more changes

app.py

@@ -236,16 +236,14 @@ DEFAULT_CONFIG = {
     'log_light_power': 3.0
 }
 
-def process_garment(input_type, text_prompt, base_text_prompt, target_image, base_image, mesh_target_image, source_mesh_type, custom_mesh, epochs, learning_rate, clip_weight, delta_clip_weight, progress=gr.Progress()):
+def process_garment(input_type, text_prompt, base_text_prompt, mesh_target_image, source_mesh_type, custom_mesh, epochs, learning_rate, clip_weight, delta_clip_weight, progress=gr.Progress()):
     """
     Main function to process garment generation
     
     Args:
-        input_type: Either "Text", "Image", or "Image to Mesh" to determine the processing mode
+        input_type: Either "Text" or "Image to Mesh" to determine the processing mode
         text_prompt: Text description of target garment (for text mode)
         base_text_prompt: Text description of base garment (for text mode)
-        target_image: Image of target garment style (for image mode)
-        base_image: Optional image of base garment (for image mode)
         mesh_target_image: Image for generating a 3D mesh (for image to mesh mode)
         source_mesh_type: Type of source mesh to use as starting point (for image to mesh mode)
         custom_mesh: Optional custom source mesh file (.obj)
@@ -278,7 +276,7 @@ def process_garment(input_type, text_prompt, base_text_prompt, target_image, bas
                     mesh_target_image.save(target_mesh_image_path)
                 else:
                     print(f"Unsupported image type: {type(mesh_target_image)}")
-                    return None
+                    return "Error: Could not process the uploaded image. Please try a different image format."
                 
                 print(f"Target mesh image saved to {target_mesh_image_path}")
                 
@@ -295,71 +293,6 @@ def process_garment(input_type, text_prompt, base_text_prompt, target_image, bas
                     'fashion_text': False,
                 })
                 
-            elif input_type == "Image" and target_image is not None:
-                # Image-based processing
-                progress(0.05, desc="Preparing image inputs...")
-                
-                # Save target image to temp directory
-                target_image_path = os.path.join(temp_dir, "target_image.jpg")
-                
-                # Handle different possible image formats from Gradio
-                if target_image is None:
-                    return None
-                
-                print(f"Target image type: {type(target_image)}")
-                
-                if isinstance(target_image, str):
-                    # Image is a file path
-                    print(f"Copying image from path: {target_image}")
-                    shutil.copy(target_image, target_image_path)
-                elif isinstance(target_image, np.ndarray):
-                    # Image is a numpy array from Gradio
-                    print(f"Converting numpy array image with shape: {target_image.shape}")
-                    # Make sure the array is in RGB format (convert if grayscale)
-                    if len(target_image.shape) == 2:
-                        target_image = np.stack([target_image] * 3, axis=2)
-                    elif target_image.shape[2] == 4:  # RGBA
-                        target_image = target_image[:,:,:3]  # Drop alpha channel
-                    
-                    img = Image.fromarray(target_image.astype(np.uint8))
-                    img.save(target_image_path)
-                    print(f"Saved numpy array as image to: {target_image_path}")
-                elif hasattr(target_image, 'save'):
-                    # Image is a PIL image
-                    print("Saving PIL image")
-                    target_image.save(target_image_path)
-                else:
-                    print(f"Unsupported image type: {type(target_image)}")
-                    return None
-                
-                print(f"Target image saved to {target_image_path}")
-                
-                # Save base image if provided, or use default
-                if base_image is not None:
-                    base_image_path = os.path.join(temp_dir, "base_image.jpg")
-                    if isinstance(base_image, str):
-                        shutil.copy(base_image, base_image_path)
-                    elif isinstance(base_image, np.ndarray):
-                        img = Image.fromarray(base_image)
-                        img.save(base_image_path)
-                    elif hasattr(base_image, 'save'):
-                        base_image.save(base_image_path)
-                    else:
-                        # If format is unsupported, use target image
-                        base_image_path = target_image_path
-                    print(f"Base image saved to {base_image_path}")
-                else:
-                    # If no base image, use a placeholder or default
-                    base_image_path = target_image_path
-                    print("No base image provided, using target image as base")
-                
-                # Configure for image-based processing
-                config.update({
-                    'image_prompt': target_image_path,
-                    'base_image_prompt': base_image_path,
-                    'fashion_image': True,
-                    'fashion_text': False
-                })
             else:
                 # Text-based processing
                 if not text_prompt or len(text_prompt.strip()) == 0:
@@ -416,37 +349,26 @@ def process_garment(input_type, text_prompt, base_text_prompt, target_image, bas
             try:
                 # Check if loop is available
                 if loop is None:
-                    error_message = "Error: The garment generation engine could not be loaded due to dependency issues."
+                    error_message = "Error: The garment generation engine could not be loaded due to dependency issues. This is likely due to compatibility issues between PyTorch and torchvision in the current environment."
                     print(error_message)
-                    # Create an error file to return
-                    error_file = os.path.join(temp_dir, "error_report.txt")
-                    with open(error_file, 'w') as f:
-                        f.write("Garment3DGen Error Report\n\n")
-                        f.write(error_message + "\n\n")
-                        f.write("This is likely due to compatibility issues between PyTorch and torchvision.")
-                    return error_file
+                    return error_message
                 
                 # Run the loop with error handling
                 loop(config)
             except RuntimeError as e:
                 print(f"Runtime error during processing: {e}")
                 if "operator torchvision::nms does not exist" in str(e):
-                    print("This is a known issue with PyTorch/torchvision version incompatibility.")
-                    
-                # Create error file
-                error_file = os.path.join(temp_dir, "error_report.txt")
-                with open(error_file, 'w') as f:
-                    f.write("Garment3DGen Error Report\n\n")
-                    f.write(f"Error during processing: {str(e)}\n\n")
-                    f.write("Please check that you have compatible versions of PyTorch and torchvision installed.")
-                return error_file
+                    error_message = "Error: PyTorch/torchvision version incompatibility detected. This is a known issue in some environments."
+                    print(error_message)
+                    return error_message
+                else:
+                    error_message = f"Runtime error during processing: {str(e)}"
+                    print(error_message)
+                    return error_message
             except Exception as e:
                 print(f"Error during processing: {e}")
-                error_file = os.path.join(temp_dir, "error_report.txt")
-                with open(error_file, 'w') as f:
-                    f.write("Garment3DGen Error Report\n\n")
-                    f.write(f"Error: {str(e)}")
-                return error_file
+                error_message = f"Error during processing: {str(e)}"
+                return error_message
                 
             progress(0.9, desc="Processing complete, preparing output...")
             
@@ -526,18 +448,17 @@ def create_interface():
         that can be used for virtual try-on applications.
         
         ## How to use:
-        1. Choose **Text**, **Image**, or **Image to Mesh** input mode using the radio button below
+        1. Choose **Text** or **Image to Mesh** input mode using the radio button below
         2. For **Text** mode: Enter descriptions of your target and base garment styles
-        3. For **Image** mode: Upload an image of your desired garment style
-        4. For **Image to Mesh** mode: Upload an image to generate a 3D mesh directly
-        5. Click "Generate 3D Garment" to create your 3D mesh file
+        3. For **Image to Mesh** mode: Upload an image to generate a 3D mesh directly and select a base mesh type
+        4. Click "Generate 3D Garment" to create your 3D mesh file
         """)
         
         with gr.Row():
             with gr.Column():
                 # Input type selector
                 input_type = gr.Radio(
-                    choices=["Text", "Image", "Image to Mesh"],
+                    choices=["Text", "Image to Mesh"],
                     value="Text",
                     label="Generation Method"
                 )
@@ -556,23 +477,7 @@ def create_interface():
                         value="simple t-shirt"
                     )
                 
-                # Image inputs (hidden by default)
-                with gr.Group(visible=False) as image_group:
-                    target_image = gr.Image(
-                        label="Target Garment Image",
-                        sources=["upload", "webcam"],
-                        type="numpy",
-                        interactive=True
-                    )
-                    gr.Markdown("*Upload an image of the desired garment style*")
-                    
-                    base_image = gr.Image(
-                        label="Base Garment Image (Optional)",
-                        sources=["upload", "webcam"],
-                        type="numpy",
-                        interactive=True
-                    )
-                    gr.Markdown("*Upload a base garment image (optional)*")
+
                 
                 # Image to Mesh inputs (hidden by default)
                 with gr.Group(visible=False) as image_to_mesh_group:
@@ -643,9 +548,9 @@ def create_interface():
                 gr.Markdown("""
                 ## Tips:
                 
-                - For text mode: Be specific in your descriptions
-                - For image mode: Use clear, front-facing garment images
-                - For image to mesh mode: Use clear garment images to generate a 3D mesh directly
+                - For text mode: Be specific in your descriptions (e.g., "red leather jacket with zippers")
+                - For image to mesh mode: Use clear, front-facing garment images to generate a 3D mesh directly
+                - Choose the appropriate base mesh type that matches your target garment
                 - Higher epochs = better quality but longer processing time
                 - Output files can be downloaded by clicking on them
                 
@@ -658,20 +563,16 @@ def create_interface():
         # Define a function to handle mode changes with clearer UI feedback
         def update_mode(mode):
             text_visibility = mode == "Text"
-            image_visibility = mode == "Image"
             image_to_mesh_visibility = mode == "Image to Mesh"
             status_msg = f"Mode changed to {mode}. "
             
             if text_visibility:
                 status_msg += "Enter garment descriptions and click Generate."
-            elif image_visibility:
-                status_msg += "Upload garment images and click Generate."
             else:
                 status_msg += "Upload a garment image and select mesh type, then click Generate."
                 
             return (
                 gr.Group.update(visible=text_visibility),
-                gr.Group.update(visible=image_visibility),
                 gr.Group.update(visible=image_to_mesh_visibility),
                 status_msg
             )
@@ -682,7 +583,10 @@ def create_interface():
                 result = process_garment(*args)
                 if result is None:
                     return None, "Processing failed. Please check the logs for details."
-                return result, "Processing completed successfully! Download your 3D garment file below."
+                elif isinstance(result, str) and result.startswith("Error:"):
+                    return None, result
+                else:
+                    return result, "Processing completed successfully! Download your 3D garment file below."
             except Exception as e:
                 import traceback
                 print(f"Error in interface: {str(e)}")
@@ -693,7 +597,7 @@ def create_interface():
         input_type.change(
             fn=update_mode,
             inputs=[input_type],
-            outputs=[text_group, image_group, image_to_mesh_group, status_output]
+            outputs=[text_group, image_to_mesh_group, status_output]
         )
         
         # Connect the button to the processing function with error handling
@@ -703,8 +607,6 @@ def create_interface():
                 input_type,
                 text_prompt,
                 base_text_prompt,
-                target_image,
-                base_image,
                 mesh_target_image,
                 source_mesh_type,
                 custom_mesh,