Localize UI to Korean and add documentation

- Translate all UI elements to Korean (labels, buttons, error messages)
- Add American Gothic painting as first example
- Add CLAUDE.md with codebase architecture and development guide

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

CLAUDE.md

@@ -0,0 +1,130 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Overview
+
+This is a Hugging Face Gradio Space for camera angle control in image editing using the Qwen Image Edit 2509 model. The application allows users to control camera rotation, movement, vertical tilt, and lens settings through a web interface. It uses optimized 4-step inference with a fused LoRA model for multiple camera angles.
+
+## Architecture
+
+### Core Components
+
+1. **app.py** - Main Gradio application
+   - Loads the Qwen Image Edit pipeline with custom transformer and FlashAttention-3 processor
+   - Loads and fuses LoRA weights from `dx8152/Qwen-Edit-2509-Multiple-angles` with scale 1.25
+   - Provides camera control UI (rotation, forward movement, vertical tilt, wide-angle lens)
+   - Generates bilingual (Chinese/English) prompts from camera controls
+   - Integrates with external video generation service (`multimodalart/wan-2-2-first-last-frame`)
+   - Implements live inference with auto-reset on image upload
+
+2. **optimization.py** - Pipeline optimization module
+   - Uses `spaces.aoti_compile()` for ahead-of-time (AOT) compilation of transformer
+   - Defines dynamic shapes for image and text sequence lengths
+   - Configures TorchInductor with coordinate descent tuning and CUDA graphs
+   - Float8 quantization code is present but commented out (line 59)
+
+3. **qwenimage/** - Custom Qwen model implementations
+   - **pipeline_qwenimage_edit_plus.py** - Custom diffusion pipeline for Qwen Image Edit
+   - **transformer_qwenimage.py** - QwenImageTransformer2DModel with double-stream architecture
+   - **qwen_fa3_processor.py** - FlashAttention-3 attention processor for joint text-image attention
+   - **__init__.py** - Package initialization (minimal)
+
+### Key Technical Details
+
+- **Model**: Uses `Qwen/Qwen-Image-Edit-2509` base with `linoyts/Qwen-Image-Edit-Rapid-AIO` transformer for fast 4-step inference
+- **LoRA**: Camera angle control LoRA from `dx8152/Qwen-Edit-2509-Multiple-angles` (镜头转换.safetensors) fused at scale 1.25
+- **Attention**: FlashAttention-3 via HuggingFace `kernels` package (`kernels-community/vllm-flash-attn3`)
+- **Optimization**: AOT compilation with dynamic shapes and CUDA graphs for ~1500s GPU duration
+- **Device**: CUDA if available, falls back to CPU
+- **Dtype**: bfloat16 throughout
+
+### Camera Prompt Building
+
+The `build_camera_prompt` function (app.py:70-99) converts slider values to bilingual prompts:
+- **Rotation**: ±45° or ±90° left/right
+- **Forward movement**: 0 (none), 1-4 (move forward), 5-10 (close-up)
+- **Vertical tilt**: -1 (bird's-eye), 0 (neutral), +1 (worm's-eye)
+- **Wide-angle**: Boolean checkbox
+
+Prompts are generated in both Chinese and English (e.g., "将镜头向左旋转45度 Rotate the camera 45 degrees to the left.").
+
+## Common Commands
+
+### Running the Application
+
+```bash
+# Install dependencies
+pip install -r requirements.txt
+
+# Run the Gradio app (launches on default port 7860)
+python app.py
+```
+
+### Development
+
+The app is designed to run on Hugging Face Spaces with ZeroGPU support. The `@spaces.GPU` decorator allocates GPU resources for inference and compilation.
+
+Key environment notes:
+- Requires CUDA GPU for optimal performance
+- FlashAttention-3 requires the `kernels` package with `kernels-community/vllm-flash-attn3`
+- Pipeline warmup happens at startup with dummy 1024x1024 images (app.py:54)
+
+### Model Loading Flow
+
+1. Load base pipeline from `Qwen/Qwen-Image-Edit-2509`
+2. Swap transformer with rapid version from `linoyts/Qwen-Image-Edit-Rapid-AIO`
+3. Load LoRA weights for camera angles
+4. Fuse LoRA at scale 1.25 and unload weights
+5. Set custom transformer class and FlashAttention-3 processor
+6. Optimize pipeline with AOT compilation
+
+## Important Implementation Details
+
+### Image Dimensions
+
+- Input images are automatically resized to maintain aspect ratio with max dimension 1024
+- Dimensions are rounded to multiples of 8 (required by the VAE)
+- See `update_dimensions_on_upload()` (app.py:191-210)
+
+### Live Inference
+
+- Control sliders trigger inference on `.release()` events
+- Wide-angle checkbox triggers on `.input()` event
+- Reset flag prevents inference during control resets
+- Previous output is stored for chaining edits
+
+### Video Generation
+
+- Optional feature to create video transitions between input and output images
+- Uses external Gradio client: `multimodalart/wan-2-2-first-last-frame`
+- Requires `x-ip-token` header from incoming request
+- Saves temporary files for API communication
+
+### Attention Processor Limitations
+
+The FlashAttention-3 processor (qwen_fa3_processor.py) does NOT support:
+- Arbitrary attention masks
+- Causal masking
+- Windowed attention or sink tokens (not plumbed through)
+
+If you need these features, you must modify the processor or fall back to standard attention.
+
+## Dependencies
+
+Core dependencies from requirements.txt:
+- diffusers (git+https://github.com/huggingface/diffusers.git)
+- transformers
+- accelerate
+- safetensors
+- peft
+- torchao==0.11.0
+- kernels (for FlashAttention-3)
+
+## Gradio Space Configuration
+
+From README.md:
+- SDK: gradio 5.49.1
+- App file: app.py
+- License: Apache 2.0
+- Inference: 4 steps (configurable via slider, default=4)

app.py

@@ -133,7 +133,7 @@ def infer_camera_edit(
         pil_images.append(prev_output.convert("RGB"))
 
     if len(pil_images) == 0:
-        raise gr.Error("Please upload an image first.")
+        raise gr.Error("먼저 이미지를 업로드해주세요.")
 
     if prompt == "no camera movement":
         return image, seed, prompt
@@ -153,28 +153,28 @@ def infer_camera_edit(
 def create_video_between_images(input_image, output_image, prompt: str, request: gr.Request) -> str:
     """Create a video between the input and output images."""
     if input_image is None or output_image is None:
-        raise gr.Error("Both input and output images are required to create a video.")
-    
+        raise gr.Error("비디오 생성을 위해 입력 및 출력 이미지가 모두 필요합니다.")
+
     try:
-        
+
         with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as tmp:
             input_image.save(tmp.name)
             input_image_path = tmp.name
-        
+
         output_pil = Image.fromarray(output_image.astype('uint8'))
         with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as tmp:
             output_pil.save(tmp.name)
             output_image_path = tmp.name
-            
+
         video_path = _generate_video_segment(
-            input_image_path, 
-            output_image_path, 
-            prompt if prompt else "Camera movement transformation",
+            input_image_path,
+            output_image_path,
+            prompt if prompt else "카메라 움직임 변환",
             request
         )
         return video_path
     except Exception as e:
-        raise gr.Error(f"Video generation failed: {e}")
+        raise gr.Error(f"비디오 생성 실패: {e}")
 
 
 # --- UI ---
@@ -212,42 +212,42 @@ def update_dimensions_on_upload(image):
 
 with gr.Blocks(theme=gr.themes.Citrus(), css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        gr.Markdown("## 🎬 Qwen Image Edit — Camera Angle Control")
+        gr.Markdown("## 🎬 Qwen Image Edit — 카메라 앵글 컨트롤")
         gr.Markdown("""
-            Qwen Image Edit 2509 for Camera Control ✨ 
-            Using [dx8152's Qwen-Edit-2509-Multiple-angles LoRA](https://huggingface.co/dx8152/Qwen-Edit-2509-Multiple-angles) and [Phr00t/Qwen-Image-Edit-Rapid-AIO](https://huggingface.co/Phr00t/Qwen-Image-Edit-Rapid-AIO/tree/main) for 4-step inference 💨
+            카메라 컨트롤을 위한 Qwen Image Edit 2509 ✨
+            4단계 추론을 위한 [dx8152's Qwen-Edit-2509-Multiple-angles LoRA](https://huggingface.co/dx8152/Qwen-Edit-2509-Multiple-angles)와 [Phr00t/Qwen-Image-Edit-Rapid-AIO](https://huggingface.co/Phr00t/Qwen-Image-Edit-Rapid-AIO/tree/main) 사용 💨
             """
         )
 
         with gr.Row():
             with gr.Column():
-                image = gr.Image(label="Input Image", type="pil")
+                image = gr.Image(label="입력 이미지", type="pil")
                 prev_output = gr.Image(value=None, visible=False)
                 is_reset = gr.Checkbox(value=False, visible=False)
 
-                with gr.Tab("Camera Controls"):
-                    rotate_deg = gr.Slider(label="Rotate Right-Left (degrees °)", 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.Tab("카메라 컨트롤"):
+                    rotate_deg = gr.Slider(label="좌우 회전 (각도 °)", minimum=-90, maximum=90, step=45, value=0)
+                    move_forward = gr.Slider(label="전진 → 클로즈업", minimum=0, maximum=10, step=5, value=0)
+                    vertical_tilt = gr.Slider(label="수직 앵글 (조감 ↔ 앙각)", minimum=-1, maximum=1, step=1, value=0)
+                    wideangle = gr.Checkbox(label="광각 렌즈", value=False)
                 with gr.Row():
-                        reset_btn = gr.Button("Reset")
-                        run_btn = gr.Button("Generate", variant="primary")
+                        reset_btn = gr.Button("초기화")
+                        run_btn = gr.Button("생성", variant="primary")
 
-                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)
+                with gr.Accordion("고급 설정", open=False):
+                    seed = gr.Slider(label="시드", minimum=0, maximum=MAX_SEED, step=1, value=0)
+                    randomize_seed = gr.Checkbox(label="랜덤 시드", value=True)
+                    true_guidance_scale = gr.Slider(label="가이던스 스케일", minimum=1.0, maximum=10.0, step=0.1, value=1.0)
+                    num_inference_steps = gr.Slider(label="추론 단계", minimum=1, maximum=40, step=1, value=4)
+                    height = gr.Slider(label="높이", minimum=256, maximum=2048, step=8, value=1024)
+                    width = gr.Slider(label="너비", minimum=256, maximum=2048, step=8, value=1024)
 
             with gr.Column():
-                result = gr.Image(label="Output Image", interactive=False)
-                prompt_preview = gr.Textbox(label="Processed Prompt", interactive=False)
-                create_video_button = gr.Button("🎥 Create Video Between Images", variant="secondary", visible=False)
+                result = gr.Image(label="출력 이미지", interactive=False)
+                prompt_preview = gr.Textbox(label="처리된 프롬프트", interactive=False)
+                create_video_button = gr.Button("🎥 이미지 간 비디오 생성", variant="secondary", visible=False)
                 with gr.Group(visible=False) as video_group:
-                    video_output = gr.Video(label="Generated Video", show_download_button=True, autoplay=True)
+                    video_output = gr.Video(label="생성된 비디오", show_download_button=True, autoplay=True)
                     
     inputs = [
         image,rotate_deg, move_forward,
@@ -292,6 +292,7 @@ with gr.Blocks(theme=gr.themes.Citrus(), css=css) as demo:
     # Examples
     gr.Examples(
         examples=[
+            ["https://upload.wikimedia.org/wikipedia/commons/thumb/c/cc/Grant_Wood_-_American_Gothic_-_Google_Art_Project.jpg/1697px-Grant_Wood_-_American_Gothic_-_Google_Art_Project.jpg", 0, 0, 0, False, 0, True, 1.0, 4, 1024, 768],
             ["tool_of_the_sea.png", 90, 0, 0, False, 0, True, 1.0, 4, 568, 1024],
             ["monkey.jpg", -90, 0, 0, False, 0, True, 1.0, 4, 704, 1024],
             ["metropolis.jpg", 0, 0, -1, False, 0, True, 1.0, 4, 816, 1024],