Upload 2 files

app.py

@@ -0,0 +1,278 @@
+import base64
+import os
+import gradio as gr
+from google import genai
+from google.genai import types
+from google.genai.types import HarmBlockThreshold
+from PIL import Image
+from io import BytesIO
+import tempfile
+import warnings
+import io
+
+# Load environment variables from .env file
+
+def swap_clothing(person_image, clothing_image):
+    """
+    Generate an image where the person from the first image is wearing clothing from the second image.
+
+    Args:
+        person_image: The image containing the person
+        clothing_image: The image containing the clothing to swap
+
+    Returns:
+        The generated image with the clothing swapped and any relevant messages
+    """
+    # Capture warnings in a string buffer
+    warning_buffer = io.StringIO()
+    warnings.filterwarnings('always')  # Ensure all warnings are shown
+
+    # Initialize variables outside the try block
+    temp_files = []
+    uploaded_files = []
+    client = None
+    output_image = None
+    output_text = ""
+
+    with warnings.catch_warnings(record=True) as warning_list:
+        try:
+            # Check if both images are provided
+            if person_image is None or clothing_image is None:
+                return None, "Please upload both images."
+
+            # Get API key from environment variables
+            api_key = "AIzaSyCIrNCyWF0AO0EYVd7bJ37lBjPozQOrUjE"
+            if not api_key:
+                return None, "GEMINI_API_KEY not found in environment variables."
+
+            # Create a fresh client instance for each request
+            client = genai.Client(api_key=api_key)
+
+            # Save both uploaded images to temporary files
+            for img, prefix in [(person_image, "person"), (clothing_image, "clothing")]:
+                with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as temp_file:
+                    img.save(temp_file.name)
+                    temp_files.append(temp_file.name)
+
+            # Upload both files to Gemini with fresh file uploads
+            uploaded_files = [
+                client.files.upload(file=temp_files[0]),  # person image
+                client.files.upload(file=temp_files[1]),  # clothing image
+            ]
+
+            # Create the prompt
+            prompt = """
+Perform a precise clothing swap between two images while maintaining strict fidelity to original elements:
+
+1. PERSON IMAGE (SOURCE TO KEEP):
+- Retain EXACT pose, body proportions, and background
+- Preserve 100% of facial features, hair, and skin details
+- Maintain original lighting conditions and environment shadows
+
+2. CLOTHING IMAGE (TARGET TO TRANSFER):
+- Extract clothing item with complete pattern/texture details
+- Match garment cut/style exactly while maintaining fabric texture
+- Preserve original color palette and material appearance
+
+3. OUTPUT REQUIREMENTS:
+- Seamlessly overlay clothing from [Clothing Image] onto [Person Image]
+- Perfectly align garment with body posture/muscle contours
+- Maintain original image resolution and sharpness
+- No alterations to facial features/body shape/background elements
+- Ensure natural fold/wrinkle physics matching original pose
+- Keep accessories/body jewelry unchanged unless covered by new clothing
+
+4. SPECIAL INSTRUCTIONS:
+- Priority 1: Clothing accuracy from target image
+- Priority 2: Person/image integrity from source photo
+- Priority 3: Seamless visual integration
+- Reject any style transfer that requires pose/background modification
+"""
+
+            contents = [
+                types.Content(
+                    role="user",
+                    parts=[
+                        types.Part.from_text(text="This is the person image. Do not change the face or features of the person. Pay attention and retain the face, environment, background, pose, facial features."),
+                        types.Part.from_uri(
+                            file_uri=uploaded_files[0].uri,
+                            mime_type=uploaded_files[0].mime_type,
+                        ),
+                        types.Part.from_text(text="This is the clothing image. Swap the clothing onto the person image."),
+                        types.Part.from_uri(
+                            file_uri=uploaded_files[1].uri,
+                            mime_type=uploaded_files[1].mime_type,
+                        ),
+                        types.Part.from_text(text=prompt),
+                        types.Part.from_uri(
+                            file_uri=uploaded_files[0].uri,
+                            mime_type=uploaded_files[0].mime_type,
+                        ),
+                    ],
+                ),
+            ]
+
+            generate_content_config = types.GenerateContentConfig(
+                temperature=0.099,
+                top_p=0.95,
+                top_k=40,
+                max_output_tokens=8192,
+                response_modalities=[
+                    "image",
+                    "text",
+                ],
+                safety_settings=[
+                    types.SafetySetting(
+                        category="HARM_CATEGORY_HARASSMENT",
+                        threshold=HarmBlockThreshold.BLOCK_NONE,
+                    ),
+                    types.SafetySetting(
+                        category="HARM_CATEGORY_HATE_SPEECH",
+                        threshold=HarmBlockThreshold.BLOCK_NONE,
+                    ),
+                    types.SafetySetting(
+                        category="HARM_CATEGORY_SEXUALLY_EXPLICIT",
+                        threshold=HarmBlockThreshold.BLOCK_NONE,
+                    ),
+                    types.SafetySetting(
+                        category="HARM_CATEGORY_DANGEROUS_CONTENT",
+                        threshold=HarmBlockThreshold.BLOCK_NONE,
+                    ),
+                ],
+                response_mime_type="text/plain",
+            )
+
+            response = client.models.generate_content(
+                model="models/gemini-2.0-flash-exp",
+                contents=contents,
+                config=generate_content_config,
+            )
+
+            # Add any warnings to the output text
+            if warning_list:
+                output_text += "\nWarnings:\n"
+                for warning in warning_list:
+                    output_text += f"- {warning.message}\n"
+
+            # Process the response
+            if response and hasattr(response, 'candidates') and response.candidates:
+                candidate = response.candidates[0]
+                if hasattr(candidate, 'content') and candidate.content:
+                    for part in candidate.content.parts:
+                        if part.text is not None:
+                            output_text += part.text + "\n"
+                        elif part.inline_data is not None:
+                            try:
+                                if isinstance(part.inline_data.data, bytes):
+                                    image_data = part.inline_data.data
+                                else:
+                                    image_data = base64.b64decode(part.inline_data.data)
+
+                                with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as temp_file:
+                                    temp_file.write(image_data)
+                                    temp_file_path = temp_file.name
+
+                                output_image = Image.open(temp_file_path)
+                                os.unlink(temp_file_path)
+
+                            except Exception as img_error:
+                                output_text += f"Error processing image: {str(img_error)}\n"
+            else:
+                output_text = "The model did not generate a valid response. Please try again with different images."
+
+        except Exception as e:
+            error_details = f"Error: {str(e)}\n\nType: {type(e).__name__}"
+            if warning_list:
+                error_details += "\n\nWarnings:\n"
+                for warning in warning_list:
+                    error_details += f"- {warning.message}\n"
+            print(f"Exception occurred: {error_details}")
+            return None
+
+        finally:
+            # Clean up all temporary files
+            for temp_file in temp_files:
+                if os.path.exists(temp_file):
+                    os.unlink(temp_file)
+
+            # Clean up any uploaded files if possible
+            for uploaded_file in uploaded_files:
+                try:
+                    if hasattr(client.files, 'delete') and uploaded_file:
+                        client.files.delete(uploaded_file.uri)
+                except:
+                    pass  # Best effort cleanup
+
+            # Clear the client
+            client = None
+
+        return output_image
+
+# Create the Gradio interface
+def create_interface():
+
+    custom_css = """
+    @font-face {
+        font-family: "XKCD";
+        src: url("https://raw.githubusercontent.com/ipython/xkcd-font/refs/heads/master/xkcd-script/font/xkcd-script.ttf") format("opentype");
+    }
+
+    .gradio-container {
+        font-family: "XKCD", "Courier New", monospace;
+    }
+    """
+
+    with gr.Blocks(css = custom_css,title="Virtual Clothing Try-On", theme ='gstaff/xkcd') as app:
+        gr.Markdown("# Virtual Clothing Try-On 👕 🩳 🧢")
+        gr.Markdown("Upload a photo of yourself and a photo of clothing you'd like to try on!")
+
+        with gr.Row():
+            with gr.Column():
+                person_image = gr.Image(label="Your Photo", type="pil", image_mode="RGB", height=300)
+                person_examples = [
+                 "https://i.postimg.cc/MGxQDGQP/440050401-46f62432-634e-49e6-b841-25d43945e4ed.png",
+                 "https://i.postimg.cc/fyHC49xR/440050411-c3bc6a7c-496f-493f-aad7-2e434e32fb86.png",
+                 "https://i.postimg.cc/cJNQbmVc/440050418-46dbca08-003b-4a14-8fcf-5a61b4c91e12.png",
+                 "https://i.postimg.cc/SskfJrdK/440050443-2abac8ca-f5b9-4ba9-9437-36e5ccf027d5.png",
+                 "https://i.postimg.cc/LXDBf1Q3/440050446-fc8c2be6-f33a-4ebb-a6a2-0a1d514a219c.png",
+                 "https://i.postimg.cc/kXBF6DQW/440050457-318b1e52-46af-4b29-9b21-6281a6687302.png",
+                 "https://i.postimg.cc/3xGX16bR/440050469-a5113e2c-a644-4484-a4de-2b6914796deb.png",
+                 "https://i.postimg.cc/m25j87Ld/440050479-2bfaa190-8211-4e2f-a9f0-8b1263d22e3a.jpg"
+                ]
+ 
+
+
+                gr.Examples(examples=person_examples, inputs=person_image, label="Person Image Examples")
+
+            with gr.Column():
+                clothing_image = gr.Image(label="Clothing Photo", type="pil", image_mode="RGB", height=300)
+                clothing_examples = [
+                    "https://i.postimg.cc/2SM7QWd4/0012.jpg",
+                    "https://i.postimg.cc/TPbqCPQB/07-upper.png",
+                    "https://i.postimg.cc/Znb8bxZh/440052371-e34d38d2-88cd-4ee9-a07d-57f544fcbab1.png",
+                    "https://i.postimg.cc/c41h0048/440052384-aa986119-61ee-406e-a246-29ac83f323e6.png",
+                    "https://i.postimg.cc/qq18z6M6/440052392-d63adf74-abe2-4f57-a3b7-5ce464824e54.jpg",
+                    "https://i.postimg.cc/1tZ0TZjX/garment-1.png",
+                    "https://i.postimg.cc/hjb8R3kr/garment-2.jpg",
+                    "https://i.postimg.cc/SxJGYd9S/garment-3.jpg"
+                    ]
+
+
+                gr.Examples(examples=clothing_examples, inputs=clothing_image, label="Garment Image Examples")
+
+
+            with gr.Column():
+                output_image = gr.Image(label="Result", type="pil")
+                submit_btn = gr.Button("Generate")
+
+        submit_btn.click(
+            fn=swap_clothing,
+            inputs=[person_image, clothing_image],
+            outputs=output_image
+        )
+
+    return app
+
+if __name__ == "__main__":
+    app = create_interface()
+    app.launch()

requirements.txt

@@ -0,0 +1,6 @@
+google
+google-genai
+google-generativeai
+gradio
+python-dotenv
+pillow