Upload folder using huggingface_hub

app.py

@@ -0,0 +1,432 @@
+"""
+GLM-Image to Image Editing App
+A Gradio 6 application for image-to-image editing using the GLM-Image model.
+
+This app allows users to upload an image and provide a prompt to transform
+the image using the GLM-Image diffusion model.
+"""
+
+import gradio as gr
+import torch
+from diffusers.pipelines.glm_image import GlmImagePipeline
+from PIL import Image
+import os
+from datetime import datetime
+
+# Initialize the model (lazy loading for better startup performance)
+pipe = None
+
+def load_model():
+    """Load the GLM-Image model with proper configuration."""
+    global pipe
+    if pipe is None:
+        pipe = GlmImagePipeline.from_pretrained(
+            "zai-org/GLM-Image",
+            torch_dtype=torch.bfloat16,
+            device_map="cuda"
+        )
+    return pipe
+
+def validate_dimensions(height: int, width: int) -> tuple:
+    """
+    Validate and adjust dimensions to be multiples of 32.
+    GLM-Image requires height and width to be multiples of 32.
+    """
+    # Adjust to nearest multiples of 32
+    adjusted_height = (height // 32 + (1 if height % 32 != 0 else 0)) * 32
+    adjusted_width = (width // 32 + (1 if width % 32 != 0 else 0)) * 32
+    return adjusted_height, adjusted_width
+
+def get_image_dimensions(image: Image.Image) -> tuple:
+    """Get the dimensions of an uploaded PIL image."""
+    return image.size[1], image.size[0]  # height, width
+
+def process_image(
+    image: Image.Image,
+    prompt: str,
+    height: int,
+    width: int,
+    num_inference_steps: int,
+    guidance_scale: float,
+    seed: int,
+    progress: gr.Progress
+) -> tuple:
+    """
+    Process the image through the GLM-Image pipeline.
+    
+    Args:
+        image: Input PIL Image
+        prompt: Text prompt describing the desired transformation
+        height: Output height (must be multiple of 32)
+        width: Output width (must be multiple of 32)
+        num_inference_steps: Number of diffusion steps
+        guidance_scale: Guidance scale for diffusion
+        seed: Random seed for reproducibility
+        progress: Gradio progress tracker
+    
+    Returns:
+        Tuple of (output_image, status_message)
+    """
+    try:
+        # Validate inputs
+        if image is None:
+            raise ValueError("Please upload an image first.")
+        
+        if not prompt or not prompt.strip():
+            raise ValueError("Please enter a prompt describing the image transformation.")
+        
+        # Adjust dimensions to be multiples of 32
+        adjusted_height, adjusted_width = validate_dimensions(height, width)
+        
+        if adjusted_height != height or adjusted_width != width:
+            height, width = adjusted_height, adjusted_width
+        
+        # Load model if not already loaded
+        progress(0.1, desc="Loading model...")
+        model = load_model()
+        
+        # Prepare image
+        progress(0.2, desc="Processing image...")
+        input_image = image.convert("RGB")
+        
+        # Create generator with seed
+        generator = torch.Generator(device="cuda").manual_seed(seed)
+        
+        # Run the pipeline
+        progress(0.3, desc="Generating image...")
+        result = model(
+            prompt=prompt,
+            image=[input_image],
+            height=height,
+            width=width,
+            num_inference_steps=num_inference_steps,
+            guidance_scale=guidance_scale,
+            generator=generator
+        )
+        
+        output_image = result.images[0]
+        
+        progress(1.0, desc="Complete!")
+        
+        status = f"✅ Successfully generated! ({height}x{width}, {num_inference_steps} steps)"
+        return output_image, status
+        
+    except Exception as e:
+        error_msg = f"❌ Error: {str(e)}"
+        return None, error_msg
+
+def update_dimensions_from_image(image: Image.Image) -> tuple:
+    """Update height and width based on uploaded image dimensions."""
+    if image is None:
+        return 1024, 1024
+    h, w = get_image_dimensions(image)
+    # Adjust to nearest multiples of 32
+    adjusted_h = (h // 32 + (1 if h % 32 != 0 else 0)) * 32
+    adjusted_w = (w // 32 + (1 if w % 32 != 0 else 0)) * 32
+    return adjusted_h, adjusted_w
+
+def generate_random_seed() -> int:
+    """Generate a random seed for the diffusion process."""
+    import random
+    return random.randint(0, 2**32 - 1)
+
+# Custom theme with modern design
+custom_theme = gr.themes.Soft(
+    primary_hue="indigo",
+    secondary_hue="blue",
+    neutral_hue="slate",
+    font=gr.themes.GoogleFont("Inter"),
+    text_size="lg",
+    spacing_size="md",
+    radius_size="md"
+).set(
+    button_primary_background_fill="*primary_600",
+    button_primary_background_fill_hover="*primary_700",
+    button_secondary_background_fill="*secondary_200",
+    button_secondary_background_fill_hover="*secondary_300",
+    block_title_text_weight="600",
+    block_title_text_color="*primary_600",
+    input_background_fill="*neutral_100",
+    input_border_color="*neutral_300",
+    input_focus_border_color="*primary_400",
+)
+
+# Build the Gradio 6 application
+with gr.Blocks(theme=custom_theme, fill_height=True) as demo:
+    
+    # Header with branding
+    gr.Markdown(
+        """
+        # 🎨 GLM-Image Editor
+        Transform your images with AI-powered editing. Upload an image and describe how you want to modify it.
+        
+        <div align="center">
+        <a href="https://huggingface.co/spaces/akhaliq/anycoder">Built with anycoder</a>
+        </div>
+        """,
+        elem_classes=["header-markdown"]
+    )
+    
+    # Main content in a row
+    with gr.Row(equal_height=True):
+        # Left column - Input controls
+        with gr.Column(scale=1, min_width=350):
+            gr.Markdown("### 📤 Input")
+            
+            # Image upload
+            input_image = gr.Image(
+                label="Upload Image",
+                type="pil",
+                sources=["upload", "clipboard"],
+                elem_id="input-image",
+                height=300
+            )
+            
+            # Prompt input
+            prompt = gr.Textbox(
+                label="Prompt",
+                placeholder="Describe how you want to transform the image...\n\nExample: Replace the background with an underground station featuring an automatic escalator.",
+                lines=4,
+                max_lines=6,
+                info="Be specific about what you want to change"
+            )
+            
+            # Advanced settings accordion
+            with gr.Accordion("⚙️ Advanced Settings", open=False):
+                with gr.Row():
+                    height = gr.Number(
+                        label="Height",
+                        value=1024,
+                        minimum=64,
+                        maximum=2048,
+                        step=32,
+                        info="Will be adjusted to multiple of 32"
+                    )
+                    width = gr.Number(
+                        label="Width",
+                        value=1024,
+                        minimum=64,
+                        maximum=2048,
+                        step=32,
+                        info="Will be adjusted to multiple of 32"
+                    )
+                
+                with gr.Row():
+                    num_inference_steps = gr.Slider(
+                        label="Inference Steps",
+                        minimum=10,
+                        maximum=100,
+                        value=50,
+                        step=5,
+                        info="More steps = higher quality but slower"
+                    )
+                    guidance_scale = gr.Slider(
+                        label="Guidance Scale",
+                        minimum=0.5,
+                        maximum=3.0,
+                        value=1.5,
+                        step=0.1,
+                        info="How closely to follow the prompt"
+                    )
+                
+                with gr.Row():
+                    seed = gr.Number(
+                        label="Seed",
+                        value=42,
+                        minimum=0,
+                        maximum=2**32 - 1,
+                        step=1,
+                        info="Random seed for reproducibility"
+                    )
+                    random_seed_btn = gr.Button(
+                        "🎲 Random",
+                        size="sm",
+                        variant="secondary"
+                    )
+            
+            # Action buttons
+            with gr.Row():
+                generate_btn = gr.Button(
+                    "✨ Generate Image",
+                    variant="primary",
+                    size="lg",
+                    full_width=True
+                )
+            
+            # Clear button
+            clear_btn = gr.Button(
+                "🗑️ Clear All",
+                variant="stop",
+                size="sm"
+            )
+        
+        # Right column - Output
+        with gr.Column(scale=1, min_width=350):
+            gr.Markdown("### 📥 Output")
+            
+            # Output image display
+            output_image = gr.Image(
+                label="Generated Image",
+                type="pil",
+                elem_id="output-image",
+                height=400,
+                interactive=False
+            )
+            
+            # Status message
+            status = gr.Textbox(
+                label="Status",
+                value="Ready to generate!",
+                interactive=False,
+                show_label=True
+            )
+            
+            # Download button
+            download_btn = gr.DownloadButton(
+                "💾 Download Image",
+                value=None,
+                variant="secondary",
+                interactive=False
+            )
+            
+            # Tips section
+            with gr.Accordion("💡 Tips for Better Results", open=False):
+                gr.Markdown(
+                    """
+                    - **Be specific**: Include details about colors, style, and composition
+                    - **Background changes**: Start with "Replace the background with..." or "Change the background to..."
+                    - **Style transfer**: Mention artistic styles like "in the style of" or specific artists
+                    - **Lighting**: Include lighting conditions like "soft natural lighting" or "dramatic shadows"
+                    
+                    **Common prompt patterns:**
+                    - `"Replace the background with [description]"`
+                    - `"Transform this into [style]"`
+                    - `"Add [elements] to the scene"`
+                    - `"Change the [specific part] to [description]"`
+                    """
+                )
+    
+    # Example prompts section
+    with gr.Accordion("📝 Example Prompts", open=False):
+        gr.Markdown(
+            """
+            Try these prompts to get started:
+            
+            | Transformation | Prompt |
+            |---------------|--------|
+            | Background Change | "Replace the background with a futuristic city skyline at sunset" |
+            | Style Transfer | "Transform this into an oil painting in the style of Van Gogh" |
+            | Scene Change | "Change the environment to an underwater coral reef with colorful fish" |
+            | Object Addition | "Add a red sports car parked in the foreground" |
+            | Weather Effect | "Make it look like it's raining with wet reflections on the ground" |
+            """
+        )
+    
+    # Event handlers
+    
+    # Update dimensions when image is uploaded
+    input_image.change(
+        fn=update_dimensions_from_image,
+        inputs=input_image,
+        outputs=[height, width],
+        api_visibility="private"
+    )
+    
+    # Random seed generation
+    random_seed_btn.click(
+        fn=generate_random_seed,
+        outputs=seed,
+        api_visibility="private"
+    )
+    
+    # Generate button handler
+    generate_btn.click(
+        fn=process_image,
+        inputs=[
+            input_image,
+            prompt,
+            height,
+            width,
+            num_inference_steps,
+            guidance_scale,
+            seed
+        ],
+        outputs=[output_image, status],
+        progress=gr.Progress()
+    )
+    
+    # Update download button when output is generated
+    def enable_download(img):
+        if img is not None:
+            return gr.DownloadButton(value=img, interactive=True)
+        return gr.DownloadButton(interactive=False)
+    
+    output_image.change(
+        fn=enable_download,
+        inputs=output_image,
+        outputs=download_btn,
+        api_visibility="private"
+    )
+    
+    # Clear button handler
+    def clear_all():
+        return {
+            input_image: None,
+            prompt: "",
+            output_image: None,
+            status: "Ready to generate!",
+            download_btn: gr.DownloadButton(interactive=False)
+        }
+    
+    clear_btn.click(
+        fn=clear_all,
+        outputs={
+            input_image: input_image,
+            prompt: prompt,
+            output_image: output_image,
+            status: status,
+            download_btn: download_btn
+        },
+        api_visibility="private"
+    )
+
+# Gradio 6 - ALL app parameters go in launch()!
+demo.launch(
+    theme=custom_theme,
+    css="""
+    .header-markdown {
+        text-align: center;
+        padding: 1rem;
+        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+        border-radius: 0.5rem;
+        margin-bottom: 1rem;
+    }
+    .header-markdown h1 {
+        color: white !important;
+        margin: 0;
+    }
+    .header-markdown p {
+        color: rgba(255,255,255,0.9) !important;
+        margin: 0.5rem 0 0 0;
+    }
+    .header-markdown a {
+        color: #ffd700 !important;
+        text-decoration: underline;
+    }
+    #input-image, #output-image {
+        border: 2px dashed var(--neutral-300);
+        border-radius: var(--radius-lg);
+    }
+    #input-image:hover, #output-image:hover {
+        border-color: var(--primary-400);
+    }
+    """,
+    footer_links=[
+        {"label": "Built with anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"},
+        {"label": "GLM-Image Model", "url": "https://huggingface.co/zai-org/GLM-Image"},
+        {"label": "Diffusers Library", "url": "https://github.com/huggingface/diffusers"}
+    ],
+    server_name="0.0.0.0",
+    server_port=7860,
+    debug=False
+)

requirements.txt

@@ -0,0 +1,21 @@
+Pillow
+torch
+torchvision
+torchaudio
+gradio>=6.0
+git+https://github.com/huggingface/diffusers
+git+https://github.com/huggingface/transformers
+sentencepiece
+accelerate
+tokenizers
+datasets
+numpy
+requests
+safetensors
+huggingface-hub
+fsspec
+packaging
+filelock
+regex
+tqdm
+pyyaml