Datasets:

---

hexuan21

/

VideoFeedback_bad5k

like 1

[ "2004379_00.jpg", "2004379_01.jpg", "2004379_02.jpg", "2004379_03.jpg", "2004379_04.jpg", "2004379_05.jpg", "2004379_06.jpg", "2004379_07.jpg", "2004379_08.jpg", "2004379_09.jpg", "2004379_10.jpg", "2004379_11.jpg", "2004379_12.jpg", "2004379_13.jpg", "2004379_14.jpg", "2004379_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Tour of an art gallery with many beautiful works of art in different styles.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "6000252_00.jpg", "6000252_01.jpg", "6000252_02.jpg", "6000252_03.jpg", "6000252_04.jpg", "6000252_05.jpg", "6000252_06.jpg", "6000252_07.jpg", "6000252_08.jpg", "6000252_09.jpg", "6000252_10.jpg", "6000252_11.jpg", "6000252_12.jpg", "6000252_13.jpg", "6000252_14.jpg", "6000252_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a dog jumping in a field\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2004002_00.jpg", "2004002_01.jpg", "2004002_02.jpg", "2004002_03.jpg", "2004002_04.jpg", "2004002_05.jpg", "2004002_06.jpg", "2004002_07.jpg", "2004002_08.jpg", "2004002_09.jpg", "2004002_10.jpg", "2004002_11.jpg", "2004002_12.jpg", "2004002_13.jpg", "2004002_14.jpg", "2004002_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"happy couple, birthday celebration, candles, restaurant, family, celebrate, party, 4k, realistic, hyper realism, high definition, clear faces, symmetrical faces, models, supermodel\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "7002438_00.jpg", "7002438_01.jpg", "7002438_02.jpg", "7002438_03.jpg", "7002438_04.jpg", "7002438_05.jpg", "7002438_06.jpg", "7002438_07.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a lovely robot with two big eyes\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "3006236_00.jpg", "3006236_01.jpg", "3006236_02.jpg", "3006236_03.jpg", "3006236_04.jpg", "3006236_05.jpg", "3006236_06.jpg", "3006236_07.jpg", "3006236_08.jpg", "3006236_09.jpg", "3006236_10.jpg", "3006236_11.jpg", "3006236_12.jpg", "3006236_13.jpg", "3006236_14.jpg", "3006236_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a ship in the night, ethereal\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "1002228_00.jpg", "1002228_01.jpg", "1002228_02.jpg", "1002228_03.jpg", "1002228_04.jpg", "1002228_05.jpg", "1002228_06.jpg", "1002228_07.jpg", "1002228_08.jpg", "1002228_09.jpg", "1002228_10.jpg", "1002228_11.jpg", "1002228_12.jpg", "1002228_13.jpg", "1002228_14.jpg", "1002228_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"matrix rain effect that is written in geez and amharic letters realistic\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "9000043_00.jpg", "9000043_01.jpg", "9000043_02.jpg", "9000043_03.jpg", "9000043_04.jpg", "9000043_05.jpg", "9000043_06.jpg", "9000043_07.jpg", "9000043_08.jpg", "9000043_09.jpg", "9000043_10.jpg", "9000043_11.jpg", "9000043_12.jpg", "9000043_13.jpg", "9000043_14.jpg", "9000043_15.jpg", "9000043_16.jpg", "9000043_17.jpg", "9000043_18.jpg", "9000043_19.jpg", "9000043_20.jpg", "9000043_21.jpg", "9000043_22.jpg", "9000043_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Naruto and Sasuke standing back to back, determination etched on their faces as they prepare to face Madara highly detailed\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "1002063_00.jpg", "1002063_01.jpg", "1002063_02.jpg", "1002063_03.jpg", "1002063_04.jpg", "1002063_05.jpg", "1002063_06.jpg", "1002063_07.jpg", "1002063_08.jpg", "1002063_09.jpg", "1002063_10.jpg", "1002063_11.jpg", "1002063_12.jpg", "1002063_13.jpg", "1002063_14.jpg", "1002063_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"an urbex in a old dark mansion, first person view, the guy filming fell in a hole threw the wood floor\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "4003712_00.jpg", "4003712_01.jpg", "4003712_02.jpg", "4003712_03.jpg", "4003712_04.jpg", "4003712_05.jpg", "4003712_06.jpg", "4003712_07.jpg", "4003712_08.jpg", "4003712_09.jpg", "4003712_10.jpg", "4003712_11.jpg", "4003712_12.jpg", "4003712_13.jpg", "4003712_14.jpg", "4003712_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A nervous candidate walks up to the door of a company with a resume in hand.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "4000490_00.jpg", "4000490_01.jpg", "4000490_02.jpg", "4000490_03.jpg", "4000490_04.jpg", "4000490_05.jpg", "4000490_06.jpg", "4000490_07.jpg", "4000490_08.jpg", "4000490_09.jpg", "4000490_10.jpg", "4000490_11.jpg", "4000490_12.jpg", "4000490_13.jpg", "4000490_14.jpg", "4000490_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"I want the sides of the face to move, the eyes to blink, the forearms to move. mouth can move\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0000872_00.jpg", "0000872_01.jpg", "0000872_02.jpg", "0000872_03.jpg", "0000872_04.jpg", "0000872_05.jpg", "0000872_06.jpg", "0000872_07.jpg", "0000872_08.jpg", "0000872_09.jpg", "0000872_10.jpg", "0000872_11.jpg", "0000872_12.jpg", "0000872_13.jpg", "0000872_14.jpg", "0000872_15.jpg", "0000872_16.jpg", "0000872_17.jpg", "0000872_18.jpg", "0000872_19.jpg", "0000872_20.jpg", "0000872_21.jpg", "0000872_22.jpg", "0000872_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"2d cartoon style, a cat is eating a fish.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0001049_00.jpg", "0001049_01.jpg", "0001049_02.jpg", "0001049_03.jpg", "0001049_04.jpg", "0001049_05.jpg", "0001049_06.jpg", "0001049_07.jpg", "0001049_08.jpg", "0001049_09.jpg", "0001049_10.jpg", "0001049_11.jpg", "0001049_12.jpg", "0001049_13.jpg", "0001049_14.jpg", "0001049_15.jpg", "0001049_16.jpg", "0001049_17.jpg", "0001049_18.jpg", "0001049_19.jpg", "0001049_20.jpg", "0001049_21.jpg", "0001049_22.jpg", "0001049_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Reveal the glittering gems in the treasure chest with a sparkling effect.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "5000714_00.jpg", "5000714_01.jpg", "5000714_02.jpg", "5000714_03.jpg", "5000714_04.jpg", "5000714_05.jpg", "5000714_06.jpg", "5000714_07.jpg", "5000714_08.jpg", "5000714_09.jpg", "5000714_10.jpg", "5000714_11.jpg", "5000714_12.jpg", "5000714_13.jpg", "5000714_14.jpg", "5000714_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Pan Dutch street scene restaurant zoom in to window, 3d style\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "0000498_00.jpg", "0000498_01.jpg", "0000498_02.jpg", "0000498_03.jpg", "0000498_04.jpg", "0000498_05.jpg", "0000498_06.jpg", "0000498_07.jpg", "0000498_08.jpg", "0000498_09.jpg", "0000498_10.jpg", "0000498_11.jpg", "0000498_12.jpg", "0000498_13.jpg", "0000498_14.jpg", "0000498_15.jpg", "0000498_16.jpg", "0000498_17.jpg", "0000498_18.jpg", "0000498_19.jpg", "0000498_20.jpg", "0000498_21.jpg", "0000498_22.jpg", "0000498_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"medium shot of tech CEO Chicano Gilbert Gottfried lecturing at a symposium\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "2005610_00.jpg", "2005610_01.jpg", "2005610_02.jpg", "2005610_03.jpg", "2005610_04.jpg", "2005610_05.jpg", "2005610_06.jpg", "2005610_07.jpg", "2005610_08.jpg", "2005610_09.jpg", "2005610_10.jpg", "2005610_11.jpg", "2005610_12.jpg", "2005610_13.jpg", "2005610_14.jpg", "2005610_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Wounded peasant fleeing through the foggy forest\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "1001532_00.jpg", "1001532_01.jpg", "1001532_02.jpg", "1001532_03.jpg", "1001532_04.jpg", "1001532_05.jpg", "1001532_06.jpg", "1001532_07.jpg", "1001532_08.jpg", "1001532_09.jpg", "1001532_10.jpg", "1001532_11.jpg", "1001532_12.jpg", "1001532_13.jpg", "1001532_14.jpg", "1001532_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"handsome pretty beauty woman medieval, frontface, in red princess dress, in the cemetery night, wearing red and black bead thread, big ruby ​​ring and laugh, holding trident, background moon and stars and mistical supernatural\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "3002057_00.jpg", "3002057_01.jpg", "3002057_02.jpg", "3002057_03.jpg", "3002057_04.jpg", "3002057_05.jpg", "3002057_06.jpg", "3002057_07.jpg", "3002057_08.jpg", "3002057_09.jpg", "3002057_10.jpg", "3002057_11.jpg", "3002057_12.jpg", "3002057_13.jpg", "3002057_14.jpg", "3002057_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"the sun god saturn holding a scythe\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0000473_00.jpg", "0000473_01.jpg", "0000473_02.jpg", "0000473_03.jpg", "0000473_04.jpg", "0000473_05.jpg", "0000473_06.jpg", "0000473_07.jpg", "0000473_08.jpg", "0000473_09.jpg", "0000473_10.jpg", "0000473_11.jpg", "0000473_12.jpg", "0000473_13.jpg", "0000473_14.jpg", "0000473_15.jpg", "0000473_16.jpg", "0000473_17.jpg", "0000473_18.jpg", "0000473_19.jpg", "0000473_20.jpg", "0000473_21.jpg", "0000473_22.jpg", "0000473_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"jedi vs sith on an ocean\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "6001863_00.jpg", "6001863_01.jpg", "6001863_02.jpg", "6001863_03.jpg", "6001863_04.jpg", "6001863_05.jpg", "6001863_06.jpg", "6001863_07.jpg", "6001863_08.jpg", "6001863_09.jpg", "6001863_10.jpg", "6001863_11.jpg", "6001863_12.jpg", "6001863_13.jpg", "6001863_14.jpg", "6001863_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a wise old man, dressed in traditional clothes and holding a scroll in his hands.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0000908_00.jpg", "0000908_01.jpg", "0000908_02.jpg", "0000908_03.jpg", "0000908_04.jpg", "0000908_05.jpg", "0000908_06.jpg", "0000908_07.jpg", "0000908_08.jpg", "0000908_09.jpg", "0000908_10.jpg", "0000908_11.jpg", "0000908_12.jpg", "0000908_13.jpg", "0000908_14.jpg", "0000908_15.jpg", "0000908_16.jpg", "0000908_17.jpg", "0000908_18.jpg", "0000908_19.jpg", "0000908_20.jpg", "0000908_21.jpg", "0000908_22.jpg", "0000908_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"I have a picture of a football player celebrating his goal\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "4002039_00.jpg", "4002039_01.jpg", "4002039_02.jpg", "4002039_03.jpg", "4002039_04.jpg", "4002039_05.jpg", "4002039_06.jpg", "4002039_07.jpg", "4002039_08.jpg", "4002039_09.jpg", "4002039_10.jpg", "4002039_11.jpg", "4002039_12.jpg", "4002039_13.jpg", "4002039_14.jpg", "4002039_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The person wearing square glasses is Kang Zewen, accompanied by a garden dog named Xiaobai, which has been raised at home for three years.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "0002674_00.jpg", "0002674_01.jpg", "0002674_02.jpg", "0002674_03.jpg", "0002674_04.jpg", "0002674_05.jpg", "0002674_06.jpg", "0002674_07.jpg", "0002674_08.jpg", "0002674_09.jpg", "0002674_10.jpg", "0002674_11.jpg", "0002674_12.jpg", "0002674_13.jpg", "0002674_14.jpg", "0002674_15.jpg", "0002674_16.jpg", "0002674_17.jpg", "0002674_18.jpg", "0002674_19.jpg", "0002674_20.jpg", "0002674_21.jpg", "0002674_22.jpg", "0002674_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a dog falling from the sky, fast motion blur effect\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "6003628_00.jpg", "6003628_01.jpg", "6003628_02.jpg", "6003628_03.jpg", "6003628_04.jpg", "6003628_05.jpg", "6003628_06.jpg", "6003628_07.jpg", "6003628_08.jpg", "6003628_09.jpg", "6003628_10.jpg", "6003628_11.jpg", "6003628_12.jpg", "6003628_13.jpg", "6003628_14.jpg", "6003628_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"one silver UFO hovering over the great pyramid,zoom out\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0005784_00.jpg", "0005784_01.jpg", "0005784_02.jpg", "0005784_03.jpg", "0005784_04.jpg", "0005784_05.jpg", "0005784_06.jpg", "0005784_07.jpg", "0005784_08.jpg", "0005784_09.jpg", "0005784_10.jpg", "0005784_11.jpg", "0005784_12.jpg", "0005784_13.jpg", "0005784_14.jpg", "0005784_15.jpg", "0005784_16.jpg", "0005784_17.jpg", "0005784_18.jpg", "0005784_19.jpg", "0005784_20.jpg", "0005784_21.jpg", "0005784_22.jpg", "0005784_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A beautiful homemade video showing the people of Lagos, Nigeria in the year 2056. Shot with a mobile phone camera.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "6001614_00.jpg", "6001614_01.jpg", "6001614_02.jpg", "6001614_03.jpg", "6001614_04.jpg", "6001614_05.jpg", "6001614_06.jpg", "6001614_07.jpg", "6001614_08.jpg", "6001614_09.jpg", "6001614_10.jpg", "6001614_11.jpg", "6001614_12.jpg", "6001614_13.jpg", "6001614_14.jpg", "6001614_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Write a heartwarming scene where a Shiba Inu dog and a Persian cat, named Mochi and Whiskers respectively, are peacefully sleeping side by side. Capture the tranquility of their friendship and the unique characteristics of each pet as they share a cozy moment together.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2001663_00.jpg", "2001663_01.jpg", "2001663_02.jpg", "2001663_03.jpg", "2001663_04.jpg", "2001663_05.jpg", "2001663_06.jpg", "2001663_07.jpg", "2001663_08.jpg", "2001663_09.jpg", "2001663_10.jpg", "2001663_11.jpg", "2001663_12.jpg", "2001663_13.jpg", "2001663_14.jpg", "2001663_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"two kids running on grassland on sunny day with the style of oil painting\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0006382_00.jpg", "0006382_01.jpg", "0006382_02.jpg", "0006382_03.jpg", "0006382_04.jpg", "0006382_05.jpg", "0006382_06.jpg", "0006382_07.jpg", "0006382_08.jpg", "0006382_09.jpg", "0006382_10.jpg", "0006382_11.jpg", "0006382_12.jpg", "0006382_13.jpg", "0006382_14.jpg", "0006382_15.jpg", "0006382_16.jpg", "0006382_17.jpg", "0006382_18.jpg", "0006382_19.jpg", "0006382_20.jpg", "0006382_21.jpg", "0006382_22.jpg", "0006382_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a cow on the right of an elephant, front view\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "3001202_00.jpg", "3001202_01.jpg", "3001202_02.jpg", "3001202_03.jpg", "3001202_04.jpg", "3001202_05.jpg", "3001202_06.jpg", "3001202_07.jpg", "3001202_08.jpg", "3001202_09.jpg", "3001202_10.jpg", "3001202_11.jpg", "3001202_12.jpg", "3001202_13.jpg", "3001202_14.jpg", "3001202_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Generate a Cosmic Ascension Message: Ascension (Font: MODERN)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "8000858_00.jpg", "8000858_01.jpg", "8000858_02.jpg", "8000858_03.jpg", "8000858_04.jpg", "8000858_05.jpg", "8000858_06.jpg", "8000858_07.jpg", "8000858_08.jpg", "8000858_09.jpg", "8000858_10.jpg", "8000858_11.jpg", "8000858_12.jpg", "8000858_13.jpg", "8000858_14.jpg", "8000858_15.jpg", "8000858_16.jpg", "8000858_17.jpg", "8000858_18.jpg", "8000858_19.jpg", "8000858_20.jpg", "8000858_21.jpg", "8000858_22.jpg", "8000858_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"film still of bruce wayne waking up at 5 am in 2024 movie scene\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2006202_00.jpg", "2006202_01.jpg", "2006202_02.jpg", "2006202_03.jpg", "2006202_04.jpg", "2006202_05.jpg", "2006202_06.jpg", "2006202_07.jpg", "2006202_08.jpg", "2006202_09.jpg", "2006202_10.jpg", "2006202_11.jpg", "2006202_12.jpg", "2006202_13.jpg", "2006202_14.jpg", "2006202_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a dog enjoying a peaceful walk\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0004055_00.jpg", "0004055_01.jpg", "0004055_02.jpg", "0004055_03.jpg", "0004055_04.jpg", "0004055_05.jpg", "0004055_06.jpg", "0004055_07.jpg", "0004055_08.jpg", "0004055_09.jpg", "0004055_10.jpg", "0004055_11.jpg", "0004055_12.jpg", "0004055_13.jpg", "0004055_14.jpg", "0004055_15.jpg", "0004055_16.jpg", "0004055_17.jpg", "0004055_18.jpg", "0004055_19.jpg", "0004055_20.jpg", "0004055_21.jpg", "0004055_22.jpg", "0004055_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a fly landed on a jam jar\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0000260_00.jpg", "0000260_01.jpg", "0000260_02.jpg", "0000260_03.jpg", "0000260_04.jpg", "0000260_05.jpg", "0000260_06.jpg", "0000260_07.jpg", "0000260_08.jpg", "0000260_09.jpg", "0000260_10.jpg", "0000260_11.jpg", "0000260_12.jpg", "0000260_13.jpg", "0000260_14.jpg", "0000260_15.jpg", "0000260_16.jpg", "0000260_17.jpg", "0000260_18.jpg", "0000260_19.jpg", "0000260_20.jpg", "0000260_21.jpg", "0000260_22.jpg", "0000260_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A guy playing Magic The Gathering cards surrounded by creatures from the playing cards\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 4\n temporal consistency: 4\n dynamic degree: 4\n text-to-video alignment: 0\n factual consistency: 4\n" } ]

[ "8002188_00.jpg", "8002188_01.jpg", "8002188_02.jpg", "8002188_03.jpg", "8002188_04.jpg", "8002188_05.jpg", "8002188_06.jpg", "8002188_07.jpg", "8002188_08.jpg", "8002188_09.jpg", "8002188_10.jpg", "8002188_11.jpg", "8002188_12.jpg", "8002188_13.jpg", "8002188_14.jpg", "8002188_15.jpg", "8002188_16.jpg", "8002188_17.jpg", "8002188_18.jpg", "8002188_19.jpg", "8002188_20.jpg", "8002188_21.jpg", "8002188_22.jpg", "8002188_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"mad max car in europa country\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2002128_00.jpg", "2002128_01.jpg", "2002128_02.jpg", "2002128_03.jpg", "2002128_04.jpg", "2002128_05.jpg", "2002128_06.jpg", "2002128_07.jpg", "2002128_08.jpg", "2002128_09.jpg", "2002128_10.jpg", "2002128_11.jpg", "2002128_12.jpg", "2002128_13.jpg", "2002128_14.jpg", "2002128_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Tallinn old city aerial shot cinematic photorealistic\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2005692_00.jpg", "2005692_01.jpg", "2005692_02.jpg", "2005692_03.jpg", "2005692_04.jpg", "2005692_05.jpg", "2005692_06.jpg", "2005692_07.jpg", "2005692_08.jpg", "2005692_09.jpg", "2005692_10.jpg", "2005692_11.jpg", "2005692_12.jpg", "2005692_13.jpg", "2005692_14.jpg", "2005692_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"sexy lady with sexy body and gold hair and beautiful eyes\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0000624_00.jpg", "0000624_01.jpg", "0000624_02.jpg", "0000624_03.jpg", "0000624_04.jpg", "0000624_05.jpg", "0000624_06.jpg", "0000624_07.jpg", "0000624_08.jpg", "0000624_09.jpg", "0000624_10.jpg", "0000624_11.jpg", "0000624_12.jpg", "0000624_13.jpg", "0000624_14.jpg", "0000624_15.jpg", "0000624_16.jpg", "0000624_17.jpg", "0000624_18.jpg", "0000624_19.jpg", "0000624_20.jpg", "0000624_21.jpg", "0000624_22.jpg", "0000624_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a tall tree that covers the whole earth, inspired by the book of Daniel.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "6000328_00.jpg", "6000328_01.jpg", "6000328_02.jpg", "6000328_03.jpg", "6000328_04.jpg", "6000328_05.jpg", "6000328_06.jpg", "6000328_07.jpg", "6000328_08.jpg", "6000328_09.jpg", "6000328_10.jpg", "6000328_11.jpg", "6000328_12.jpg", "6000328_13.jpg", "6000328_14.jpg", "6000328_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"slow motion 3d blue vortex water flow with splashes isolated on a white background\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "8001081_00.jpg", "8001081_01.jpg", "8001081_02.jpg", "8001081_03.jpg", "8001081_04.jpg", "8001081_05.jpg", "8001081_06.jpg", "8001081_07.jpg", "8001081_08.jpg", "8001081_09.jpg", "8001081_10.jpg", "8001081_11.jpg", "8001081_12.jpg", "8001081_13.jpg", "8001081_14.jpg", "8001081_15.jpg", "8001081_16.jpg", "8001081_17.jpg", "8001081_18.jpg", "8001081_19.jpg", "8001081_20.jpg", "8001081_21.jpg", "8001081_22.jpg", "8001081_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"generate an image of a youtube profile pic about this prompt : Wild Wonders: This name implies that your channel showcases the amazing diversity and beauty of wildlife.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "4001145_00.jpg", "4001145_01.jpg", "4001145_02.jpg", "4001145_03.jpg", "4001145_04.jpg", "4001145_05.jpg", "4001145_06.jpg", "4001145_07.jpg", "4001145_08.jpg", "4001145_09.jpg", "4001145_10.jpg", "4001145_11.jpg", "4001145_12.jpg", "4001145_13.jpg", "4001145_14.jpg", "4001145_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"very detailed, money showrering over man\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2002372_00.jpg", "2002372_01.jpg", "2002372_02.jpg", "2002372_03.jpg", "2002372_04.jpg", "2002372_05.jpg", "2002372_06.jpg", "2002372_07.jpg", "2002372_08.jpg", "2002372_09.jpg", "2002372_10.jpg", "2002372_11.jpg", "2002372_12.jpg", "2002372_13.jpg", "2002372_14.jpg", "2002372_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"No prompt\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "s000519_00.jpg", "s000519_01.jpg", "s000519_02.jpg", "s000519_03.jpg", "s000519_04.jpg", "s000519_05.jpg", "s000519_06.jpg", "s000519_07.jpg", "s000519_08.jpg", "s000519_09.jpg", "s000519_10.jpg", "s000519_11.jpg", "s000519_12.jpg", "s000519_13.jpg", "s000519_14.jpg", "s000519_15.jpg", "s000519_16.jpg", "s000519_17.jpg", "s000519_18.jpg", "s000519_19.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"secret mafia members hiding in sewers\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 4\n temporal consistency: 4\n dynamic degree: 4\n text-to-video alignment: 0\n factual consistency: 4\n" } ]

[ "4001338_00.jpg", "4001338_01.jpg", "4001338_02.jpg", "4001338_03.jpg", "4001338_04.jpg", "4001338_05.jpg", "4001338_06.jpg", "4001338_07.jpg", "4001338_08.jpg", "4001338_09.jpg", "4001338_10.jpg", "4001338_11.jpg", "4001338_12.jpg", "4001338_13.jpg", "4001338_14.jpg", "4001338_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"night sky with milkyway and shooting stars Message: marfa (Font: MODERN)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "6003136_00.jpg", "6003136_01.jpg", "6003136_02.jpg", "6003136_03.jpg", "6003136_04.jpg", "6003136_05.jpg", "6003136_06.jpg", "6003136_07.jpg", "6003136_08.jpg", "6003136_09.jpg", "6003136_10.jpg", "6003136_11.jpg", "6003136_12.jpg", "6003136_13.jpg", "6003136_14.jpg", "6003136_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Imaginary portrayal of Einstein surrounded by equations and thought bubbles with sketches of his ideas.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2003942_00.jpg", "2003942_01.jpg", "2003942_02.jpg", "2003942_03.jpg", "2003942_04.jpg", "2003942_05.jpg", "2003942_06.jpg", "2003942_07.jpg", "2003942_08.jpg", "2003942_09.jpg", "2003942_10.jpg", "2003942_11.jpg", "2003942_12.jpg", "2003942_13.jpg", "2003942_14.jpg", "2003942_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"One sunny morning, as Peter was walking through the village, he noticed an injured bird on the roadside. Its wing was broken, and it looked helpless. Without a second thought, Peter gently picked up the injured bird and cradled it in his hands. He decided to nurse it back to health.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "2002617_00.jpg", "2002617_01.jpg", "2002617_02.jpg", "2002617_03.jpg", "2002617_04.jpg", "2002617_05.jpg", "2002617_06.jpg", "2002617_07.jpg", "2002617_08.jpg", "2002617_09.jpg", "2002617_10.jpg", "2002617_11.jpg", "2002617_12.jpg", "2002617_13.jpg", "2002617_14.jpg", "2002617_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A woman comes to buy donuts at a cafe and sits at a table to eat donuts. At softsoftdonut shop, Scandi style, the entire shop is rice milled.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "1005924_00.jpg", "1005924_01.jpg", "1005924_02.jpg", "1005924_03.jpg", "1005924_04.jpg", "1005924_05.jpg", "1005924_06.jpg", "1005924_07.jpg", "1005924_08.jpg", "1005924_09.jpg", "1005924_10.jpg", "1005924_11.jpg", "1005924_12.jpg", "1005924_13.jpg", "1005924_14.jpg", "1005924_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"s rooster and a cat fight\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "3000367_00.jpg", "3000367_01.jpg", "3000367_02.jpg", "3000367_03.jpg", "3000367_04.jpg", "3000367_05.jpg", "3000367_06.jpg", "3000367_07.jpg", "3000367_08.jpg", "3000367_09.jpg", "3000367_10.jpg", "3000367_11.jpg", "3000367_12.jpg", "3000367_13.jpg", "3000367_14.jpg", "3000367_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"university professor teaching class walking among students\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0005424_00.jpg", "0005424_01.jpg", "0005424_02.jpg", "0005424_03.jpg", "0005424_04.jpg", "0005424_05.jpg", "0005424_06.jpg", "0005424_07.jpg", "0005424_08.jpg", "0005424_09.jpg", "0005424_10.jpg", "0005424_11.jpg", "0005424_12.jpg", "0005424_13.jpg", "0005424_14.jpg", "0005424_15.jpg", "0005424_16.jpg", "0005424_17.jpg", "0005424_18.jpg", "0005424_19.jpg", "0005424_20.jpg", "0005424_21.jpg", "0005424_22.jpg", "0005424_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"3 friend open a horror door animation\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "s000019_00.jpg", "s000019_01.jpg", "s000019_02.jpg", "s000019_03.jpg", "s000019_04.jpg", "s000019_05.jpg", "s000019_06.jpg", "s000019_07.jpg", "s000019_08.jpg", "s000019_09.jpg", "s000019_10.jpg", "s000019_11.jpg", "s000019_12.jpg", "s000019_13.jpg", "s000019_14.jpg", "s000019_15.jpg", "s000019_16.jpg", "s000019_17.jpg", "s000019_18.jpg", "s000019_19.jpg", "s000019_20.jpg", "s000019_21.jpg", "s000019_22.jpg", "s000019_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a 21 year old black guy driving a Lamborghini, Highly detailed, gs5\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 4\n temporal consistency: 2\n dynamic degree: 4\n text-to-video alignment: 0\n factual consistency: 4\n" } ]

[ "8003893_00.jpg", "8003893_01.jpg", "8003893_02.jpg", "8003893_03.jpg", "8003893_04.jpg", "8003893_05.jpg", "8003893_06.jpg", "8003893_07.jpg", "8003893_08.jpg", "8003893_09.jpg", "8003893_10.jpg", "8003893_11.jpg", "8003893_12.jpg", "8003893_13.jpg", "8003893_14.jpg", "8003893_15.jpg", "8003893_16.jpg", "8003893_17.jpg", "8003893_18.jpg", "8003893_19.jpg", "8003893_20.jpg", "8003893_21.jpg", "8003893_22.jpg", "8003893_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"“The air grows tense as a hand emerges from the darkness, fingers wrapping around the cold metal of the gun’s trigger.”\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0006808_00.jpg", "0006808_01.jpg", "0006808_02.jpg", "0006808_03.jpg", "0006808_04.jpg", "0006808_05.jpg", "0006808_06.jpg", "0006808_07.jpg", "0006808_08.jpg", "0006808_09.jpg", "0006808_10.jpg", "0006808_11.jpg", "0006808_12.jpg", "0006808_13.jpg", "0006808_14.jpg", "0006808_15.jpg", "0006808_16.jpg", "0006808_17.jpg", "0006808_18.jpg", "0006808_19.jpg", "0006808_20.jpg", "0006808_21.jpg", "0006808_22.jpg", "0006808_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Fashion show model revels, bioluminescent underwater theme, coral background, high end fashion, cinematic, highly detailed\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "1000265_00.jpg", "1000265_01.jpg", "1000265_02.jpg", "1000265_03.jpg", "1000265_04.jpg", "1000265_05.jpg", "1000265_06.jpg", "1000265_07.jpg", "1000265_08.jpg", "1000265_09.jpg", "1000265_10.jpg", "1000265_11.jpg", "1000265_12.jpg", "1000265_13.jpg", "1000265_14.jpg", "1000265_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"an Angel falling into the pupil of an eye\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "8003127_00.jpg", "8003127_01.jpg", "8003127_02.jpg", "8003127_03.jpg", "8003127_04.jpg", "8003127_05.jpg", "8003127_06.jpg", "8003127_07.jpg", "8003127_08.jpg", "8003127_09.jpg", "8003127_10.jpg", "8003127_11.jpg", "8003127_12.jpg", "8003127_13.jpg", "8003127_14.jpg", "8003127_15.jpg", "8003127_16.jpg", "8003127_17.jpg", "8003127_18.jpg", "8003127_19.jpg", "8003127_20.jpg", "8003127_21.jpg", "8003127_22.jpg", "8003127_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"tracking shot of an pink astronaut in space\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "7000304_00.jpg", "7000304_01.jpg", "7000304_02.jpg", "7000304_03.jpg", "7000304_04.jpg", "7000304_05.jpg", "7000304_06.jpg", "7000304_07.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A man has a light bulb moment.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "8001373_00.jpg", "8001373_01.jpg", "8001373_02.jpg", "8001373_03.jpg", "8001373_04.jpg", "8001373_05.jpg", "8001373_06.jpg", "8001373_07.jpg", "8001373_08.jpg", "8001373_09.jpg", "8001373_10.jpg", "8001373_11.jpg", "8001373_12.jpg", "8001373_13.jpg", "8001373_14.jpg", "8001373_15.jpg", "8001373_16.jpg", "8001373_17.jpg", "8001373_18.jpg", "8001373_19.jpg", "8001373_20.jpg", "8001373_21.jpg", "8001373_22.jpg", "8001373_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A tranquil tableau of a country estate's library featured elegant wooden shelves\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "8001929_00.jpg", "8001929_01.jpg", "8001929_02.jpg", "8001929_03.jpg", "8001929_04.jpg", "8001929_05.jpg", "8001929_06.jpg", "8001929_07.jpg", "8001929_08.jpg", "8001929_09.jpg", "8001929_10.jpg", "8001929_11.jpg", "8001929_12.jpg", "8001929_13.jpg", "8001929_14.jpg", "8001929_15.jpg", "8001929_16.jpg", "8001929_17.jpg", "8001929_18.jpg", "8001929_19.jpg", "8001929_20.jpg", "8001929_21.jpg", "8001929_22.jpg", "8001929_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"an elephant\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "3000387_00.jpg", "3000387_01.jpg", "3000387_02.jpg", "3000387_03.jpg", "3000387_04.jpg", "3000387_05.jpg", "3000387_06.jpg", "3000387_07.jpg", "3000387_08.jpg", "3000387_09.jpg", "3000387_10.jpg", "3000387_11.jpg", "3000387_12.jpg", "3000387_13.jpg", "3000387_14.jpg", "3000387_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Alexander Reben is an artist who has spent the last decade creating work that explores the humor and absurdity of human nature in artificial intelligence. Alex has been creating sculptures that originate from AI-generated imagery, manually transforming those AI creations into 3D models materialized in the physical world. “My experience of using Sora was as a starting point to develop 3D sculpture. My thoughts drifted towards exploring the realm of photogrammetry and its potential applications to sculpture. The prospect of transforming video into 3D models intrigued me, as it hinted at propelling the AI system beyond its initial scope.”\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "4003686_00.jpg", "4003686_01.jpg", "4003686_02.jpg", "4003686_03.jpg", "4003686_04.jpg", "4003686_05.jpg", "4003686_06.jpg", "4003686_07.jpg", "4003686_08.jpg", "4003686_09.jpg", "4003686_10.jpg", "4003686_11.jpg", "4003686_12.jpg", "4003686_13.jpg", "4003686_14.jpg", "4003686_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a cartoon van dancing on the beach with flowers in the hair in front of karate fights\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0005847_00.jpg", "0005847_01.jpg", "0005847_02.jpg", "0005847_03.jpg", "0005847_04.jpg", "0005847_05.jpg", "0005847_06.jpg", "0005847_07.jpg", "0005847_08.jpg", "0005847_09.jpg", "0005847_10.jpg", "0005847_11.jpg", "0005847_12.jpg", "0005847_13.jpg", "0005847_14.jpg", "0005847_15.jpg", "0005847_16.jpg", "0005847_17.jpg", "0005847_18.jpg", "0005847_19.jpg", "0005847_20.jpg", "0005847_21.jpg", "0005847_22.jpg", "0005847_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"an ant wearing shield in the middle of war and fight\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "3002326_00.jpg", "3002326_01.jpg", "3002326_02.jpg", "3002326_03.jpg", "3002326_04.jpg", "3002326_05.jpg", "3002326_06.jpg", "3002326_07.jpg", "3002326_08.jpg", "3002326_09.jpg", "3002326_10.jpg", "3002326_11.jpg", "3002326_12.jpg", "3002326_13.jpg", "3002326_14.jpg", "3002326_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A space shuttle launching into orbit, with flames and smoke billowing out from the engines\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0003372_00.jpg", "0003372_01.jpg", "0003372_02.jpg", "0003372_03.jpg", "0003372_04.jpg", "0003372_05.jpg", "0003372_06.jpg", "0003372_07.jpg", "0003372_08.jpg", "0003372_09.jpg", "0003372_10.jpg", "0003372_11.jpg", "0003372_12.jpg", "0003372_13.jpg", "0003372_14.jpg", "0003372_15.jpg", "0003372_16.jpg", "0003372_17.jpg", "0003372_18.jpg", "0003372_19.jpg", "0003372_20.jpg", "0003372_21.jpg", "0003372_22.jpg", "0003372_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"tumor innervation, with cancer species include, than tumors achive to do breast, 1 minute video\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "1000552_00.jpg", "1000552_01.jpg", "1000552_02.jpg", "1000552_03.jpg", "1000552_04.jpg", "1000552_05.jpg", "1000552_06.jpg", "1000552_07.jpg", "1000552_08.jpg", "1000552_09.jpg", "1000552_10.jpg", "1000552_11.jpg", "1000552_12.jpg", "1000552_13.jpg", "1000552_14.jpg", "1000552_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"two corgis standing at a mountains top looking like they conquered the mountain\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "7001065_00.jpg", "7001065_01.jpg", "7001065_02.jpg", "7001065_03.jpg", "7001065_04.jpg", "7001065_05.jpg", "7001065_06.jpg", "7001065_07.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a recruitment post of english learning class for kids\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "0003085_00.jpg", "0003085_01.jpg", "0003085_02.jpg", "0003085_03.jpg", "0003085_04.jpg", "0003085_05.jpg", "0003085_06.jpg", "0003085_07.jpg", "0003085_08.jpg", "0003085_09.jpg", "0003085_10.jpg", "0003085_11.jpg", "0003085_12.jpg", "0003085_13.jpg", "0003085_14.jpg", "0003085_15.jpg", "0003085_16.jpg", "0003085_17.jpg", "0003085_18.jpg", "0003085_19.jpg", "0003085_20.jpg", "0003085_21.jpg", "0003085_22.jpg", "0003085_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A planet that can be seen from the garden approaches.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "1003817_00.jpg", "1003817_01.jpg", "1003817_02.jpg", "1003817_03.jpg", "1003817_04.jpg", "1003817_05.jpg", "1003817_06.jpg", "1003817_07.jpg", "1003817_08.jpg", "1003817_09.jpg", "1003817_10.jpg", "1003817_11.jpg", "1003817_12.jpg", "1003817_13.jpg", "1003817_14.jpg", "1003817_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Harvard Lab in 1967: Stanley Milgram in a Harvard laboratory setting with assistants and research equipment around.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2001622_00.jpg", "2001622_01.jpg", "2001622_02.jpg", "2001622_03.jpg", "2001622_04.jpg", "2001622_05.jpg", "2001622_06.jpg", "2001622_07.jpg", "2001622_08.jpg", "2001622_09.jpg", "2001622_10.jpg", "2001622_11.jpg", "2001622_12.jpg", "2001622_13.jpg", "2001622_14.jpg", "2001622_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"two funny space ships flying in the sky like a friends playing\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0006252_00.jpg", "0006252_01.jpg", "0006252_02.jpg", "0006252_03.jpg", "0006252_04.jpg", "0006252_05.jpg", "0006252_06.jpg", "0006252_07.jpg", "0006252_08.jpg", "0006252_09.jpg", "0006252_10.jpg", "0006252_11.jpg", "0006252_12.jpg", "0006252_13.jpg", "0006252_14.jpg", "0006252_15.jpg", "0006252_16.jpg", "0006252_17.jpg", "0006252_18.jpg", "0006252_19.jpg", "0006252_20.jpg", "0006252_21.jpg", "0006252_22.jpg", "0006252_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Rapper Gone Fludd makes cotton candy and then eats it\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "2001924_00.jpg", "2001924_01.jpg", "2001924_02.jpg", "2001924_03.jpg", "2001924_04.jpg", "2001924_05.jpg", "2001924_06.jpg", "2001924_07.jpg", "2001924_08.jpg", "2001924_09.jpg", "2001924_10.jpg", "2001924_11.jpg", "2001924_12.jpg", "2001924_13.jpg", "2001924_14.jpg", "2001924_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Three cute girls are walking in the park.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "4001026_00.jpg", "4001026_01.jpg", "4001026_02.jpg", "4001026_03.jpg", "4001026_04.jpg", "4001026_05.jpg", "4001026_06.jpg", "4001026_07.jpg", "4001026_08.jpg", "4001026_09.jpg", "4001026_10.jpg", "4001026_11.jpg", "4001026_12.jpg", "4001026_13.jpg", "4001026_14.jpg", "4001026_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"naruto standing besides luffy on the top of a mountain\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "1000277_00.jpg", "1000277_01.jpg", "1000277_02.jpg", "1000277_03.jpg", "1000277_04.jpg", "1000277_05.jpg", "1000277_06.jpg", "1000277_07.jpg", "1000277_08.jpg", "1000277_09.jpg", "1000277_10.jpg", "1000277_11.jpg", "1000277_12.jpg", "1000277_13.jpg", "1000277_14.jpg", "1000277_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a man looking at rain in a window with a lightning storm\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "4003161_00.jpg", "4003161_01.jpg", "4003161_02.jpg", "4003161_03.jpg", "4003161_04.jpg", "4003161_05.jpg", "4003161_06.jpg", "4003161_07.jpg", "4003161_08.jpg", "4003161_09.jpg", "4003161_10.jpg", "4003161_11.jpg", "4003161_12.jpg", "4003161_13.jpg", "4003161_14.jpg", "4003161_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a girl in the sky trying to catch a car in the road\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "0006654_00.jpg", "0006654_01.jpg", "0006654_02.jpg", "0006654_03.jpg", "0006654_04.jpg", "0006654_05.jpg", "0006654_06.jpg", "0006654_07.jpg", "0006654_08.jpg", "0006654_09.jpg", "0006654_10.jpg", "0006654_11.jpg", "0006654_12.jpg", "0006654_13.jpg", "0006654_14.jpg", "0006654_15.jpg", "0006654_16.jpg", "0006654_17.jpg", "0006654_18.jpg", "0006654_19.jpg", "0006654_20.jpg", "0006654_21.jpg", "0006654_22.jpg", "0006654_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Japan night flying candles on the background Fujiyama and lots of blooming sakuras\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "1005000_00.jpg", "1005000_01.jpg", "1005000_02.jpg", "1005000_03.jpg", "1005000_04.jpg", "1005000_05.jpg", "1005000_06.jpg", "1005000_07.jpg", "1005000_08.jpg", "1005000_09.jpg", "1005000_10.jpg", "1005000_11.jpg", "1005000_12.jpg", "1005000_13.jpg", "1005000_14.jpg", "1005000_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"what do you mean by best quality?\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "7001639_00.jpg", "7001639_01.jpg", "7001639_02.jpg", "7001639_03.jpg", "7001639_04.jpg", "7001639_05.jpg", "7001639_06.jpg", "7001639_07.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Batman in the Batcave makes music on his computer\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "4001157_00.jpg", "4001157_01.jpg", "4001157_02.jpg", "4001157_03.jpg", "4001157_04.jpg", "4001157_05.jpg", "4001157_06.jpg", "4001157_07.jpg", "4001157_08.jpg", "4001157_09.jpg", "4001157_10.jpg", "4001157_11.jpg", "4001157_12.jpg", "4001157_13.jpg", "4001157_14.jpg", "4001157_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"girl changes her skin like a snake, front view, looking in camera, pieces of skin falling down, camera zoom in, high definition, intricate details, high quality\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "5001750_00.jpg", "5001750_01.jpg", "5001750_02.jpg", "5001750_03.jpg", "5001750_04.jpg", "5001750_05.jpg", "5001750_06.jpg", "5001750_07.jpg", "5001750_08.jpg", "5001750_09.jpg", "5001750_10.jpg", "5001750_11.jpg", "5001750_12.jpg", "5001750_13.jpg", "5001750_14.jpg", "5001750_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a bear fighting with a leon\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "1006553_00.jpg", "1006553_01.jpg", "1006553_02.jpg", "1006553_03.jpg", "1006553_04.jpg", "1006553_05.jpg", "1006553_06.jpg", "1006553_07.jpg", "1006553_08.jpg", "1006553_09.jpg", "1006553_10.jpg", "1006553_11.jpg", "1006553_12.jpg", "1006553_13.jpg", "1006553_14.jpg", "1006553_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"silent hill 2 man from monsters go pro footage fast movement\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "6000009_00.jpg", "6000009_01.jpg", "6000009_02.jpg", "6000009_03.jpg", "6000009_04.jpg", "6000009_05.jpg", "6000009_06.jpg", "6000009_07.jpg", "6000009_08.jpg", "6000009_09.jpg", "6000009_10.jpg", "6000009_11.jpg", "6000009_12.jpg", "6000009_13.jpg", "6000009_14.jpg", "6000009_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"rotate the record as if playing on a record player for 60 seconds\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "1004513_00.jpg", "1004513_01.jpg", "1004513_02.jpg", "1004513_03.jpg", "1004513_04.jpg", "1004513_05.jpg", "1004513_06.jpg", "1004513_07.jpg", "1004513_08.jpg", "1004513_09.jpg", "1004513_10.jpg", "1004513_11.jpg", "1004513_12.jpg", "1004513_13.jpg", "1004513_14.jpg", "1004513_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"SS troops marching in Washington DC\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "3004698_00.jpg", "3004698_01.jpg", "3004698_02.jpg", "3004698_03.jpg", "3004698_04.jpg", "3004698_05.jpg", "3004698_06.jpg", "3004698_07.jpg", "3004698_08.jpg", "3004698_09.jpg", "3004698_10.jpg", "3004698_11.jpg", "3004698_12.jpg", "3004698_13.jpg", "3004698_14.jpg", "3004698_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"No prompt\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0006598_00.jpg", "0006598_01.jpg", "0006598_02.jpg", "0006598_03.jpg", "0006598_04.jpg", "0006598_05.jpg", "0006598_06.jpg", "0006598_07.jpg", "0006598_08.jpg", "0006598_09.jpg", "0006598_10.jpg", "0006598_11.jpg", "0006598_12.jpg", "0006598_13.jpg", "0006598_14.jpg", "0006598_15.jpg", "0006598_16.jpg", "0006598_17.jpg", "0006598_18.jpg", "0006598_19.jpg", "0006598_20.jpg", "0006598_21.jpg", "0006598_22.jpg", "0006598_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Fireworks Technician setting up the explossives to the mortars\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "5000774_00.jpg", "5000774_01.jpg", "5000774_02.jpg", "5000774_03.jpg", "5000774_04.jpg", "5000774_05.jpg", "5000774_06.jpg", "5000774_07.jpg", "5000774_08.jpg", "5000774_09.jpg", "5000774_10.jpg", "5000774_11.jpg", "5000774_12.jpg", "5000774_13.jpg", "5000774_14.jpg", "5000774_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Ironman home alone in a modern luxury house\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 2\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "6001835_00.jpg", "6001835_01.jpg", "6001835_02.jpg", "6001835_03.jpg", "6001835_04.jpg", "6001835_05.jpg", "6001835_06.jpg", "6001835_07.jpg", "6001835_08.jpg", "6001835_09.jpg", "6001835_10.jpg", "6001835_11.jpg", "6001835_12.jpg", "6001835_13.jpg", "6001835_14.jpg", "6001835_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"the birth of Venus in the rays of the sun on the sea horizon\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0005916_00.jpg", "0005916_01.jpg", "0005916_02.jpg", "0005916_03.jpg", "0005916_04.jpg", "0005916_05.jpg", "0005916_06.jpg", "0005916_07.jpg", "0005916_08.jpg", "0005916_09.jpg", "0005916_10.jpg", "0005916_11.jpg", "0005916_12.jpg", "0005916_13.jpg", "0005916_14.jpg", "0005916_15.jpg", "0005916_16.jpg", "0005916_17.jpg", "0005916_18.jpg", "0005916_19.jpg", "0005916_20.jpg", "0005916_21.jpg", "0005916_22.jpg", "0005916_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Ghosts of the past lost in time\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2004933_00.jpg", "2004933_01.jpg", "2004933_02.jpg", "2004933_03.jpg", "2004933_04.jpg", "2004933_05.jpg", "2004933_06.jpg", "2004933_07.jpg", "2004933_08.jpg", "2004933_09.jpg", "2004933_10.jpg", "2004933_11.jpg", "2004933_12.jpg", "2004933_13.jpg", "2004933_14.jpg", "2004933_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"The universe will make way for you.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "2000560_00.jpg", "2000560_01.jpg", "2000560_02.jpg", "2000560_03.jpg", "2000560_04.jpg", "2000560_05.jpg", "2000560_06.jpg", "2000560_07.jpg", "2000560_08.jpg", "2000560_09.jpg", "2000560_10.jpg", "2000560_11.jpg", "2000560_12.jpg", "2000560_13.jpg", "2000560_14.jpg", "2000560_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"medical staff testing plants in lab\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "4002503_00.jpg", "4002503_01.jpg", "4002503_02.jpg", "4002503_03.jpg", "4002503_04.jpg", "4002503_05.jpg", "4002503_06.jpg", "4002503_07.jpg", "4002503_08.jpg", "4002503_09.jpg", "4002503_10.jpg", "4002503_11.jpg", "4002503_12.jpg", "4002503_13.jpg", "4002503_14.jpg", "4002503_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"fungus growing 20 inches high in Brazil jungle\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0003893_00.jpg", "0003893_01.jpg", "0003893_02.jpg", "0003893_03.jpg", "0003893_04.jpg", "0003893_05.jpg", "0003893_06.jpg", "0003893_07.jpg", "0003893_08.jpg", "0003893_09.jpg", "0003893_10.jpg", "0003893_11.jpg", "0003893_12.jpg", "0003893_13.jpg", "0003893_14.jpg", "0003893_15.jpg", "0003893_16.jpg", "0003893_17.jpg", "0003893_18.jpg", "0003893_19.jpg", "0003893_20.jpg", "0003893_21.jpg", "0003893_22.jpg", "0003893_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"yuna and tidus on the calm lands fighting fire fans final fantasy X HD\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "1001898_00.jpg", "1001898_01.jpg", "1001898_02.jpg", "1001898_03.jpg", "1001898_04.jpg", "1001898_05.jpg", "1001898_06.jpg", "1001898_07.jpg", "1001898_08.jpg", "1001898_09.jpg", "1001898_10.jpg", "1001898_11.jpg", "1001898_12.jpg", "1001898_13.jpg", "1001898_14.jpg", "1001898_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"create a 2d character for me form different angle of face\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0002696_00.jpg", "0002696_01.jpg", "0002696_02.jpg", "0002696_03.jpg", "0002696_04.jpg", "0002696_05.jpg", "0002696_06.jpg", "0002696_07.jpg", "0002696_08.jpg", "0002696_09.jpg", "0002696_10.jpg", "0002696_11.jpg", "0002696_12.jpg", "0002696_13.jpg", "0002696_14.jpg", "0002696_15.jpg", "0002696_16.jpg", "0002696_17.jpg", "0002696_18.jpg", "0002696_19.jpg", "0002696_20.jpg", "0002696_21.jpg", "0002696_22.jpg", "0002696_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"the two teenage womenboth look scared. women with the red hair looks more scare than the woman with blue hair. both teen girls are sitting on the bed. inside a house. this scene needs to look cinematic.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "2006440_00.jpg", "2006440_01.jpg", "2006440_02.jpg", "2006440_03.jpg", "2006440_04.jpg", "2006440_05.jpg", "2006440_06.jpg", "2006440_07.jpg", "2006440_08.jpg", "2006440_09.jpg", "2006440_10.jpg", "2006440_11.jpg", "2006440_12.jpg", "2006440_13.jpg", "2006440_14.jpg", "2006440_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"Hermione granger lookalike Looking directly at the camera while moving around\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "3000799_00.jpg", "3000799_01.jpg", "3000799_02.jpg", "3000799_03.jpg", "3000799_04.jpg", "3000799_05.jpg", "3000799_06.jpg", "3000799_07.jpg", "3000799_08.jpg", "3000799_09.jpg", "3000799_10.jpg", "3000799_11.jpg", "3000799_12.jpg", "3000799_13.jpg", "3000799_14.jpg", "3000799_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"margot robbie and scarlet johanson dance in underwear, crowd of people looking surprised in the background\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 3\n dynamic degree: 1\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "1003588_00.jpg", "1003588_01.jpg", "1003588_02.jpg", "1003588_03.jpg", "1003588_04.jpg", "1003588_05.jpg", "1003588_06.jpg", "1003588_07.jpg", "1003588_08.jpg", "1003588_09.jpg", "1003588_10.jpg", "1003588_11.jpg", "1003588_12.jpg", "1003588_13.jpg", "1003588_14.jpg", "1003588_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"In a world on the brink of the abyss, in the year 2023, a brave young woman rises to face an uncertain destiny.\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "5000516_00.jpg", "5000516_01.jpg", "5000516_02.jpg", "5000516_03.jpg", "5000516_04.jpg", "5000516_05.jpg", "5000516_06.jpg", "5000516_07.jpg", "5000516_08.jpg", "5000516_09.jpg", "5000516_10.jpg", "5000516_11.jpg", "5000516_12.jpg", "5000516_13.jpg", "5000516_14.jpg", "5000516_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"video of city busses stranded in flooded streets of New York city\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 2\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "2002301_00.jpg", "2002301_01.jpg", "2002301_02.jpg", "2002301_03.jpg", "2002301_04.jpg", "2002301_05.jpg", "2002301_06.jpg", "2002301_07.jpg", "2002301_08.jpg", "2002301_09.jpg", "2002301_10.jpg", "2002301_11.jpg", "2002301_12.jpg", "2002301_13.jpg", "2002301_14.jpg", "2002301_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"soccer game blue and yellow shirts Message: boca (Font: MODERN)\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "2004328_00.jpg", "2004328_01.jpg", "2004328_02.jpg", "2004328_03.jpg", "2004328_04.jpg", "2004328_05.jpg", "2004328_06.jpg", "2004328_07.jpg", "2004328_08.jpg", "2004328_09.jpg", "2004328_10.jpg", "2004328_11.jpg", "2004328_12.jpg", "2004328_13.jpg", "2004328_14.jpg", "2004328_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"a zoom in birds flying across the sky scene\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 2\n text-to-video alignment: 0\n factual consistency: 3\n" } ]

[ "0004617_00.jpg", "0004617_01.jpg", "0004617_02.jpg", "0004617_03.jpg", "0004617_04.jpg", "0004617_05.jpg", "0004617_06.jpg", "0004617_07.jpg", "0004617_08.jpg", "0004617_09.jpg", "0004617_10.jpg", "0004617_11.jpg", "0004617_12.jpg", "0004617_13.jpg", "0004617_14.jpg", "0004617_15.jpg", "0004617_16.jpg", "0004617_17.jpg", "0004617_18.jpg", "0004617_19.jpg", "0004617_20.jpg", "0004617_21.jpg", "0004617_22.jpg", "0004617_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"vertical format video. Dark volcanic wasteland. Tall dark solitary tower. Tower is style of Lord of the Rings, tower of Sauron. Flaming eye on top of tower. Tower falls over and crumbles to dust. 5 second duration for video\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "1006567_00.jpg", "1006567_01.jpg", "1006567_02.jpg", "1006567_03.jpg", "1006567_04.jpg", "1006567_05.jpg", "1006567_06.jpg", "1006567_07.jpg", "1006567_08.jpg", "1006567_09.jpg", "1006567_10.jpg", "1006567_11.jpg", "1006567_12.jpg", "1006567_13.jpg", "1006567_14.jpg", "1006567_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"tens of thousands of starlings swirling in the sky nightfall hyperrealistic very wide angle\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 1\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]

[ "0004319_00.jpg", "0004319_01.jpg", "0004319_02.jpg", "0004319_03.jpg", "0004319_04.jpg", "0004319_05.jpg", "0004319_06.jpg", "0004319_07.jpg", "0004319_08.jpg", "0004319_09.jpg", "0004319_10.jpg", "0004319_11.jpg", "0004319_12.jpg", "0004319_13.jpg", "0004319_14.jpg", "0004319_15.jpg", "0004319_16.jpg", "0004319_17.jpg", "0004319_18.jpg", "0004319_19.jpg", "0004319_20.jpg", "0004319_21.jpg", "0004319_22.jpg", "0004319_23.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"A green gorge among the snowy mountains in which hordes of warriors fight in front of the portal that calls for an army for one of the parties\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 3\n temporal consistency: 3\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 2\n" } ]

[ "6004793_00.jpg", "6004793_01.jpg", "6004793_02.jpg", "6004793_03.jpg", "6004793_04.jpg", "6004793_05.jpg", "6004793_06.jpg", "6004793_07.jpg", "6004793_08.jpg", "6004793_09.jpg", "6004793_10.jpg", "6004793_11.jpg", "6004793_12.jpg", "6004793_13.jpg", "6004793_14.jpg", "6004793_15.jpg" ]

[ { "from": "human", "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos,\nplease watch the following frames of a given video and see the text prompt for generating the video,\nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, both the consistency of objects or humans and the smoothness of motion or movements\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nfor each dimension, output a float number from 1.0 to 4.0,\nthe higher the number is, the better the video performs in that sub-score, \nthe lowest 1.0 means Bad, the highest 4.0 means Perfect/Real (the video is like a real video)\nHere is an output example:\nvisual quality: 3.2\ntemporal consistency: 2.7\ndynamic degree: 4.0\ntext-to-video alignment: 2.3\nfactual consistency: 1.8\n\nFor this video, the text prompt is \"cultists from a surrealist 1960s art film\",\nall the frames of video are as follows:\n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> " }, { "from": "gpt", "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 0\n factual consistency: 1\n" } ]