Upload 3 files

app.py

@@ -0,0 +1,198 @@
+# import cv2
+# import gradio as gr
+# import numpy as np
+# import os
+# from PIL import Image
+# from try_on_diffusion_client import TryOnDiffusionClient   # <‑ the file you just pulled
+
+# # --- 1.  Initialise the API client -----------------------------------------
+# API_URL = "https://try-on-diffusion.p.rapidapi.com"   # RapidAPI endpoint
+# API_KEY = os.getenv("RAPIDAPI_KEY")                  # put your key in an env var
+# client  = TryOnDiffusionClient(base_url=API_URL, api_key=API_KEY)
+
+# # --- 2.  The Gradio callback ------------------------------------------------
+
+# def try_on(user_img, outfit_img, height, chest, waist, sleeve):
+#     # 1️⃣  Force 3‑channel RGB and uint8 dtype
+#     user_rgb   = user_img.convert("RGB")
+#     outfit_rgb = outfit_img.convert("RGB")
+
+#     avatar_bgr   = cv2.cvtColor(np.array(user_rgb,   dtype=np.uint8), cv2.COLOR_RGB2BGR)
+#     clothing_bgr = cv2.cvtColor(np.array(outfit_rgb, dtype=np.uint8), cv2.COLOR_RGB2BGR)
+
+#     # 2️⃣  Call the API (unchanged)
+#     resp = client.try_on_file(
+#         clothing_image = clothing_bgr,
+#         avatar_image   = avatar_bgr,
+#         seed           = -1
+#     )
+
+#     if resp.status_code != 200 or resp.image is None:
+#         return None, f"API error {resp.status_code}: {resp.error_details}"
+
+#     result_rgb = cv2.cvtColor(resp.image, cv2.COLOR_BGR2RGB)
+
+#     # 3️⃣  Simple fit message
+#     fit_msg = "✅ Fit looks OK"
+#     if chest < 80:
+#         fit_msg = "⚠️ Chest measurement suggests a tight fit."
+
+#     return Image.fromarray(result_rgb), fit_msg
+
+
+# # --- 3. Create the Gradio Interface ------------------------------------------
+# demo = gr.Interface(
+#     fn=try_on,
+#     inputs=[
+#         gr.Image(type="pil", label="Upload your image"),
+#         gr.Image(type="pil", label="Upload outfit image"),
+#         gr.Number(label="Height (cm)"),
+#         gr.Number(label="Chest (cm)"),
+#         gr.Number(label="Waist (cm)"),
+#         gr.Number(label="Sleeve length (cm)")
+#     ],
+#     outputs=[
+#         gr.Image(type="pil", label="Try-On Result"),
+#         gr.Text(label="Fit Advice")
+#     ],
+#     title="👕 Virtual Try-On",
+#     description="Upload your image and an outfit to see how it might look on you!"
+# )
+
+# # --- 4. Launch the app -------------------------------------------------------
+# if __name__ == "__main__":
+#     print("Launching Virtual Try-On App...")
+#     demo.launch()
+
+
+import cv2
+import gradio as gr
+import numpy as np
+import os
+from PIL import Image
+from try_on_diffusion_client import TryOnDiffusionClient
+
+# --- 1. Initialize the API client -----------------------------------------
+API_URL = "https://try-on-diffusion.p.rapidapi.com"
+API_KEY = os.getenv("RAPIDAPI_KEY")  # This will be set as a secret in HF Spaces
+
+if not API_KEY:
+    print("⚠️ RAPIDAPI_KEY not found. Please set it in Hugging Face Spaces secrets.")
+
+client = TryOnDiffusionClient(base_url=API_URL, api_key=API_KEY)
+
+# --- 2. The Gradio callback ------------------------------------------------
+def try_on(user_img, outfit_img, height, chest, waist, sleeve):
+    """
+    Process the virtual try-on request
+    """
+    # Check if API key is available
+    if not API_KEY:
+        return None, "❌ API key not configured. Please contact the app administrator."
+    
+    # Validate inputs
+    if user_img is None or outfit_img is None:
+        return None, "❌ Please upload both user image and outfit image."
+    
+    try:
+        # Convert images to RGB and ensure uint8 dtype
+        user_rgb = user_img.convert("RGB")
+        outfit_rgb = outfit_img.convert("RGB")
+
+        avatar_bgr = cv2.cvtColor(np.array(user_rgb, dtype=np.uint8), cv2.COLOR_RGB2BGR)
+        clothing_bgr = cv2.cvtColor(np.array(outfit_rgb, dtype=np.uint8), cv2.COLOR_RGB2BGR)
+
+        # Call the API
+        resp = client.try_on_file(
+            clothing_image=clothing_bgr,
+            avatar_image=avatar_bgr,
+            seed=-1
+        )
+
+        if resp.status_code != 200 or resp.image is None:
+            error_msg = resp.error_details if resp.error_details else f"API error {resp.status_code}"
+            return None, f"❌ {error_msg}"
+
+        # Convert result back to RGB
+        result_rgb = cv2.cvtColor(resp.image, cv2.COLOR_BGR2RGB)
+
+        # Generate fit advice
+        fit_msg = generate_fit_advice(height, chest, waist, sleeve)
+
+        return Image.fromarray(result_rgb), fit_msg
+
+    except Exception as e:
+        return None, f"❌ Error processing request: {str(e)}"
+
+
+def generate_fit_advice(height, chest, waist, sleeve):
+    """
+    Generate basic fit advice based on measurements
+    """
+    advice = ["✅ Fit Analysis:"]
+    
+    if chest and chest < 80:
+        advice.append("⚠️ Chest measurement suggests the item might be tight.")
+    elif chest and chest > 120:
+        advice.append("ℹ️ Chest measurement suggests a loose fit.")
+    else:
+        advice.append("✅ Chest measurement looks good.")
+    
+    if waist and chest and abs(chest - waist) < 10:
+        advice.append("ℹ️ Similar chest and waist measurements - consider fitted styles.")
+    
+    if height and height < 160:
+        advice.append("ℹ️ For shorter stature, consider checking garment length.")
+    elif height and height > 185:
+        advice.append("ℹ️ For taller stature, ensure adequate garment length.")
+    
+    return "\n".join(advice)
+
+
+# --- 3. Create the Gradio Interface ------------------------------------------
+with gr.Blocks(theme=gr.themes.Soft(), title="Virtual Try-On") as demo:
+    gr.Markdown(
+        """
+        # 👕 Virtual Try-On App
+        
+        Upload your photo and a clothing item to see how it would look on you!
+        
+        **Tips for best results:**
+        - Use clear, well-lit photos
+        - Person should be facing forward
+        - Clothing item should be clearly visible
+        """
+    )
+    
+    with gr.Row():
+        with gr.Column():
+            user_img = gr.Image(type="pil", label="Your Photo", height=300)
+            outfit_img = gr.Image(type="pil", label="Clothing Item", height=300)
+            
+        with gr.Column():
+            gr.Markdown("### 📏 Your Measurements (Optional)")
+            height = gr.Number(label="Height (cm)", value=170, minimum=140, maximum=220)
+            chest = gr.Number(label="Chest (cm)", value=90, minimum=70, maximum=140)
+            waist = gr.Number(label="Waist (cm)", value=80, minimum=60, maximum=130)
+            sleeve = gr.Number(label="Sleeve Length (cm)", value=60, minimum=50, maximum=80)
+    
+    submit_btn = gr.Button("✨ Try On Outfit", variant="primary", size="lg")
+    
+    with gr.Row():
+        result_img = gr.Image(type="pil", label="Try-On Result", height=400)
+        fit_advice = gr.Textbox(label="Fit Advice", lines=6, max_lines=10)
+    
+    # Examples section
+    gr.Markdown("### 📸 Example Images")
+    gr.Markdown("*Upload your own images or try with sample photos*")
+    
+    submit_btn.click(
+        fn=try_on,
+        inputs=[user_img, outfit_img, height, chest, waist, sleeve],
+        outputs=[result_img, fit_advice]
+    )
+
+# --- 4. Launch the app -------------------------------------------------------
+if __name__ == "__main__":
+    print("🚀 Launching Virtual Try-On App...")
+    demo.launch()

requirements.txt

@@ -0,0 +1,6 @@
+gradio
+requests
+pillow
+numpy
+opencv-python-headless
+try-on-diffusion-client

try_on_diffusion_client.py

@@ -0,0 +1,130 @@
+import cv2
+import numpy as np
+import requests
+from requests_toolbelt.multipart.encoder import MultipartEncoder
+from urllib.parse import urlparse
+import logging
+import json
+from io import BytesIO
+from dataclasses import dataclass
+
+
+@dataclass
+class TryOnDiffusionAPIResponse:
+    status_code: int
+    image: np.ndarray = None
+    response_data: bytes = None
+    error_details: str = None
+    seed: int = None
+
+
+class TryOnDiffusionClient:
+    def __init__(self, base_url: str = "http://localhost:8000/", api_key: str = ""):
+        self._logger = logging.getLogger("try_on_diffusion_client")
+        self._base_url = base_url
+        self._api_key = api_key
+
+        if self._base_url[-1] == "/":
+            self._base_url = self._base_url[:-1]
+
+        parsed_url = urlparse(self._base_url)
+
+        self._rapidapi_host = parsed_url.netloc if parsed_url.netloc.endswith(".rapidapi.com") else None
+
+        if self._rapidapi_host is not None:
+            self._logger.info(f"Using RapidAPI proxy: {self._rapidapi_host}")
+
+    @staticmethod
+    def _image_to_upload_file(image: np.ndarray) -> tuple:
+        _, jpeg_data = cv2.imencode(".jpg", image, [int(cv2.IMWRITE_JPEG_QUALITY), 99])
+        jpeg_data = jpeg_data.tobytes()
+
+        fp = BytesIO(jpeg_data)
+
+        return "image.jpg", fp, "image/jpeg"
+
+    def try_on_file(
+        self,
+        clothing_image: np.ndarray = None,
+        clothing_prompt: str = None,
+        avatar_image: np.ndarray = None,
+        avatar_prompt: str = None,
+        avatar_sex: str = None,
+        background_image: np.ndarray = None,
+        background_prompt: str = None,
+        seed: int = -1,
+        raw_response: bool = False,
+    ) -> TryOnDiffusionAPIResponse:
+        url = self._base_url + "/try-on-file"
+
+        request_data = {"seed": str(seed)}
+
+        if clothing_image is not None:
+            request_data["clothing_image"] = self._image_to_upload_file(clothing_image)
+
+        if clothing_prompt is not None:
+            request_data["clothing_prompt"] = clothing_prompt
+
+        if avatar_image is not None:
+            request_data["avatar_image"] = self._image_to_upload_file(avatar_image)
+
+        if avatar_prompt is not None:
+            request_data["avatar_prompt"] = avatar_prompt
+
+        if avatar_sex is not None:
+            request_data["avatar_sex"] = avatar_sex
+
+        if background_image is not None:
+            request_data["background_image"] = self._image_to_upload_file(background_image)
+
+        if background_prompt is not None:
+            request_data["background_prompt"] = background_prompt
+
+        multipart_data = MultipartEncoder(fields=request_data)
+
+        headers = {"Content-Type": multipart_data.content_type}
+
+        if self._rapidapi_host is not None:
+            headers["X-RapidAPI-Key"] = self._api_key
+            headers["X-RapidAPI-Host"] = self._rapidapi_host
+        else:
+            headers["X-API-Key"] = self._api_key
+
+        try:
+            response = requests.post(
+                url,
+                data=multipart_data,
+                headers=headers,
+            )
+        except Exception as e:
+            self._logger.error(e, exc_info=True)
+            return TryOnDiffusionAPIResponse(status_code=0)
+
+        if response.status_code != 200:
+            self._logger.warning(f"Request failed, status code: {response.status_code}, response: {response.content}")
+
+        result = TryOnDiffusionAPIResponse(status_code=response.status_code)
+
+        if not raw_response and response.status_code == 200:
+            try:
+                result.image = cv2.imdecode(np.frombuffer(response.content, np.uint8), cv2.IMREAD_COLOR)
+            except:
+                result.image = None
+        else:
+            result.response_data = response.content
+
+        if result.status_code == 200:
+            if "X-Seed" in response.headers:
+                result.seed = int(response.headers["X-Seed"])
+        else:
+            try:
+                response_json = (
+                    json.loads(result.response_data.decode("utf-8")) if result.response_data is not None else None
+                )
+
+                if response_json is not None and "detail" in response_json:
+                    result.error_details = response_json["detail"]
+            except:
+                result.error_details = None
+
+        return result