import gradio as gr import numpy as np import random import torch import spaces from PIL import Image from diffusers import FlowMatchEulerDiscreteScheduler from optimization import optimize_pipeline_ from qwenimage.pipeline_qwenimage_edit_plus import QwenImageEditPlusPipeline from qwenimage.transformer_qwenimage import QwenImageTransformer2DModel from qwenimage.qwen_fa3_processor import QwenDoubleStreamAttnProcessorFA3 import math from huggingface_hub import hf_hub_download from safetensors.torch import load_file import os from PIL import Image import os import gradio as gr # --- Model Loading --- dtype = torch.bfloat16 device = "cuda" if torch.cuda.is_available() else "cpu" pipe = QwenImageEditPlusPipeline.from_pretrained("Qwen/Qwen-Image-Edit-2509", transformer= QwenImageTransformer2DModel.from_pretrained("linoyts/Qwen-Image-Edit-Rapid-AIO", subfolder='transformer', torch_dtype=dtype, device_map='cuda'),torch_dtype=dtype).to(device) pipe.load_lora_weights( "dx8152/Qwen-Edit-2509-Multiple-angles", weight_name="镜头转换.safetensors", adapter_name="angles" ) # pipe.load_lora_weights( # "lovis93/next-scene-qwen-image-lora-2509", # weight_name="next-scene_lora-v2-3000.safetensors", adapter_name="next-scene" # ) pipe.set_adapters(["angles"], adapter_weights=[1.]) pipe.fuse_lora(adapter_names=["angles"], lora_scale=1.25) # pipe.fuse_lora(adapter_names=["next-scene"], lora_scale=1.) pipe.unload_lora_weights() pipe.transformer.__class__ = QwenImageTransformer2DModel pipe.transformer.set_attn_processor(QwenDoubleStreamAttnProcessorFA3()) optimize_pipeline_(pipe, image=[Image.new("RGB", (1024, 1024)), Image.new("RGB", (1024, 1024))], prompt="prompt") MAX_SEED = np.iinfo(np.int32).max def build_camera_prompt(rotate_deg, move_forward, vertical_tilt, wideangle): prompt_parts = [] # Rotation if rotate_deg != 0: direction = "left" if rotate_deg > 0 else "right" if direction == "left": prompt_parts.append(f"将镜头向左旋转{abs(rotate_deg)}度 Rotate the camera {abs(rotate_deg)} degrees to the left.") else: prompt_parts.append(f"将镜头向右旋转{abs(rotate_deg)}度 Rotate the camera {abs(rotate_deg)} degrees to the right.") # Move forward / close-up if move_forward >= 5: prompt_parts.append("将镜头转为特写镜头 Turn the camera to a close-up.") elif move_forward >= 1: prompt_parts.append("将镜头向前移动 Move the camera forward.") # Vertical tilt if vertical_tilt <= -1: prompt_parts.append("将相机转向鸟瞰视角 Turn the camera to a bird's-eye view.") elif vertical_tilt >= 1: prompt_parts.append("将相机切换到仰视视角 Turn the camera to a worm's-eye view.") # Lens option if wideangle: prompt_parts.append(" 将镜头转为广角镜头 Turn the camera to a wide-angle lens.") final_prompt = " ".join(prompt_parts).strip() return final_prompt if final_prompt else "" @spaces.GPU def infer_camera_edit( image, prev_output, rotate_deg, move_forward, vertical_tilt, wideangle, seed, randomize_seed, true_guidance_scale, num_inference_steps, height, width, ): prompt = build_camera_prompt(rotate_deg, move_forward, vertical_tilt, wideangle) print(f"Generated Prompt: {prompt}") if randomize_seed: seed = random.randint(0, MAX_SEED) generator = torch.Generator(device=device).manual_seed(seed) # Choose input image (prefer uploaded, else last output) pil_images = [] if image is not None: if isinstance(image, Image.Image): pil_images.append(image.convert("RGB")) elif hasattr(image, "name"): pil_images.append(Image.open(image.name).convert("RGB")) elif prev_output is not None: pil_images.append(prev_output.convert("RGB")) if len(pil_images) == 0: raise gr.Error("Please upload an image first.") result = pipe( image=pil_images, prompt=prompt, height=height if height != 0 else None, width=width if width != 0 else None, num_inference_steps=num_inference_steps, generator=generator, true_cfg_scale=true_guidance_scale, num_images_per_prompt=1, ).images[0] return result, seed, prompt # --- UI --- css = "#col-container { max-width: 800px; margin: 0 auto; }" is_reset = gr.State(value=False) def reset_all(): return [0, 0, 0, 0, False, True] def end_reset(): return False with gr.Blocks(css=css) as demo: with gr.Column(elem_id="col-container"): gr.Markdown("## 🎬 Qwen Image Edit — Camera Angle Control") gr.Markdown( "" ) with gr.Row(): with gr.Column(): image = gr.Image(label="Input Image", type="pil", sources=["upload"]) prev_output = gr.State(value=None) is_reset = gr.State(value=False) with gr.Group(): rotate_deg = gr.Slider(label="Rotate Left–Right (°)", minimum=-90, maximum=90, step=45, value=0) move_forward = gr.Slider(label="Move Forward → Close-Up", minimum=0, maximum=10, step=5, value=0) vertical_tilt = gr.Slider(label="Vertical Angle (Bird ↔ Worm)", minimum=-1, maximum=1, step=1, value=0) wideangle = gr.Checkbox(label="Wide-Angle Lens", value=False) with gr.Row(): reset_btn = gr.Button("reset settings") with gr.Accordion("Advanced Settings", open=False): seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0) randomize_seed = gr.Checkbox(label="Randomize Seed", value=True) true_guidance_scale = gr.Slider(label="True Guidance Scale", minimum=1.0, maximum=10.0, step=0.1, value=1.0) num_inference_steps = gr.Slider(label="Inference Steps", minimum=1, maximum=40, step=1, value=4) height = gr.Slider(label="Height", minimum=256, maximum=2048, step=8, value=1024) width = gr.Slider(label="Width", minimum=256, maximum=2048, step=8, value=1024) run_btn = gr.Button("Generate", variant="primary", visible=False) with gr.Column(): result = gr.Image(label="Output Image") prompt_preview = gr.Textbox(label="Processed Prompt", interactive=False) #gr.Markdown("_Each change applies a fresh camera instruction to the last output image._") inputs = [ image, prev_output, rotate_deg, move_forward, vertical_tilt, wideangle, seed, randomize_seed, true_guidance_scale, num_inference_steps, height, width ] outputs = [result, seed, prompt_preview] # Reset behavior reset_btn.click( fn=reset_all, inputs=None, outputs=[rotate_deg, move_forward, vertical_tilt, wideangle, is_reset], queue=False ).then(fn=end_reset, inputs=None, outputs=[is_reset], queue=False) # Manual generation run_event = run_btn.click(fn=infer_camera_edit, inputs=inputs, outputs=outputs) # Image upload resets image.change( fn=reset_all, inputs=None, outputs=[rotate_deg, move_forward, vertical_tilt, wideangle, is_reset], queue=False ).then(fn=end_reset, inputs=None, outputs=[is_reset], queue=False) # Live updates def maybe_infer(is_reset, *args): if is_reset: return gr.update(), gr.update(), gr.update() else: return infer_camera_edit(*args) control_inputs = [ image, prev_output, rotate_deg, move_forward, vertical_tilt, wideangle, seed, randomize_seed, true_guidance_scale, num_inference_steps, height, width ] control_inputs_with_flag = [is_reset] + control_inputs for control in [rotate_deg, move_forward, vertical_tilt, wideangle]: control.change(fn=maybe_infer, inputs=control_inputs_with_flag, outputs=outputs, show_progress="minimal") run_event.then(lambda img, *_: img, inputs=[result], outputs=[prev_output]) demo.launch()