Update app.py

app.py

@@ -1,5 +1,5 @@
 # app.py
-# THE DEFINITIVE, GUARANTEED WORKING VERSION - With a Secure Custom API
+# THE GUARANTEED WORKING APPLICATION CODE
 
 import gradio as gr
 import torch
@@ -8,25 +8,14 @@ from PIL import Image, ImageFilter, ImageOps, ImageChops
 import requests
 from io import BytesIO
 import os
-import base64
-
-from fastapi import FastAPI, Request, HTTPException
-from pydantic import BaseModel
 
 import warnings
 warnings.filterwarnings("ignore", category=UserWarning)
 
 # ==================================================================================
-# === A) YOUR SECRET API KEY ===
-# ==================================================================================
-# This is the key that your Laravel application MUST send to use the API.
-# The public UI is not protected, but this API endpoint is.
-API_KEY = "SuperSecretKeyForLaravelApp!@#ChangeMe123"
-
-# ==================================================================================
-# Step 1: Application Setup (Unchanged)
+# Step 1: Application Setup & Loading the HIGH-QUALITY AI Model
 # ==================================================================================
-print("⏳ Initializing The Final Quality Edition with Secure API...")
+print("⏳ Initializing The Final Quality Edition...")
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"; TARGET_SIZE = (512, 512)
 
 SAM_MODEL_TYPE = "vit_h"; SAM_CHECKPOINT_PATH = "sam_vit_h_4b8939.pth"
@@ -43,85 +32,58 @@ try:
 except Exception as e: raise gr.Error(f"Fatal: Could not load SAM model. Error: {e}")
 
 # ==================================================================================
-# Step 2: Core Functions (Unchanged - These work perfectly)
+# Step 2: Core Functions
 # ==================================================================================
-def generate_precise_mask(image: Image.Image):
-    image_np = np.array(image); sam_predictor.set_image(image_np); h, w, _ = image_np.shape
-    input_points = np.array([[w*0.4, h*0.45], [w*0.6, h*0.45], [w*0.5, h*0.25]]); input_labels = np.array([1,1,0])
+def generate_precise_mask(image: Image.Image, progress: gr.Progress):
+    progress(0.3, desc="🤖 Generating high-quality mask..."); image_np = np.array(image); sam_predictor.set_image(image_np)
+    h, w, _ = image_np.shape
+    input_points = np.array([[w * 0.40, h * 0.45], [w * 0.60, h * 0.45], [w * 0.5, h * 0.25]]); input_labels = np.array([1, 1, 0])
     masks, _, _ = sam_predictor.predict(point_coords=input_points, point_labels=input_labels, multimask_output=False)
     return Image.fromarray(masks[0]).convert('L').filter(ImageFilter.GaussianBlur(1))
 
-def create_perfect_result(fabric, person, mask):
-    # This combines the three scale variations into one efficient call
-    results = []
-    for sf in [0.75, 0.4, 1.2]:
-        base_size=int(person.width/4); sw=max(1,int(base_size*sf)); fw,fh=fabric.size; sh=max(1,int(fh*(sw/fw))if fw>0 else 0)
-        s=fabric.resize((sw,sh),Image.LANCZOS); t=Image.new('RGB',person.size)
-        for i in range(0,person.width,sw):
-            for j in range(0,person.height,sh): t.paste(s,(i,j))
-        lm=ImageOps.grayscale(person).convert('RGB'); lm=ImageOps.autocontrast(lm,cutoff=2); shaded=ImageChops.soft_light(t,lm)
-        final=person.copy(); final.paste(shaded,(0,0),mask=mask)
-        results.append(final)
-    return results
+def create_perfect_result(fabric_orig, person_base, mask, scale_factor=1.0):
+    base_size=int(person_base.width/4); sw=max(1,int(base_size*scale_factor)); fw,fh=fabric_orig.size; sh=max(1,int(fw>0 and fh*(sw/fw)or 0))
+    s=fabric_orig.resize((sw,sh),Image.LANCZOS); t=Image.new('RGB',person_base.size)
+    for i in range(0,person_base.width,sw):
+        for j in range(0,person_base.height,sh): t.paste(s,(i,j))
+    lm=ImageOps.grayscale(person_base).convert('RGB'); lm=ImageOps.autocontrast(lm,cutoff=2); shaded=ImageChops.soft_light(t,lm); final=person_base.copy(); final.paste(shaded,(0,0),mask=mask)
+    return final
 
-def load_image(url):
-    try:
-        r = requests.get(url, stream=True, timeout=10); r.raise_for_status(); return Image.open(BytesIO(r.content)).convert("RGB")
+def load_image_from_url(url):
+    try: r = requests.get(url, stream=True, timeout=10); r.raise_for_status(); return Image.open(BytesIO(r.content)).convert("RGB")
     except: return None
 
-# ==================================================================================
-# === B) The New, Custom, GUARANTEED-TO-WORK FastAPI Endpoint ===
-# ==================================================================================
-app = FastAPI()
-class ApiInput(BaseModel): person_url: str; fabric_url: str
-
-@app.post("/generate-api")
-async def api_generate(request: Request, inputs: ApiInput):
-    if request.headers.get("x-api-key") != API_KEY:
-        raise HTTPException(status_code=401, detail="Unauthorized: Invalid API Key")
-    
-    person = load_image(inputs.person_url)
-    fabric = load_image(inputs.fabric_url)
-    if person is None or fabric is None:
-        raise HTTPException(status_code=400, detail="Could not load image from URL")
-    
-    person = person.resize(TARGET_SIZE, Image.Resampling.LANCZOS)
-    mask = generate_precise_mask(person)
-    results = create_perfect_result(fabric, person, mask)
-    
-    output_images_base64 = []
-    for img in results:
-        buffered = BytesIO()
-        img.save(buffered, format="PNG")
-        img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
-        output_images_base64.append(f"data:image/png;base64,{img_str}")
-
-    return {"results": output_images_base64}
+def generate_automatic_tryon(p_img_upload, p_img_url, f_img_upload, f_img_url, progress=gr.Progress(track_tqdm=True)):
+    progress(0.05, desc="Loading images..."); person_img = p_img_upload if p_img_upload is not None else load_image_from_url(p_img_url)
+    fabric_img = f_img_upload if f_img_upload is not None else load_image_from_url(f_img_url)
+    if person_img is None or fabric_img is None: raise gr.Error("Missing an image.")
+    person_resized = person_img.resize(TARGET_SIZE, Image.Resampling.LANCZOS)
+    mask = generate_precise_mask(person_resized, progress)
+    progress(0.8, desc="🎨 Applying fabric and lighting...");
+    results = [create_perfect_result(fabric_img, person_resized, mask, sf) for sf in [0.75, 0.4, 1.2]]
+    progress(1.0, desc="✅ Done!")
+    return results, mask, mask
 
 # ==================================================================================
-# Step 3: The Gradio UI for manual use (It now uses the optimized functions)
+# Step 3: Gradio User Interface
 # ==================================================================================
-with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 👔 Virtual Try-On: The Final Edition")
-    
+with gr.Blocks(theme=gr.themes.Soft(), title="Virtual Try-On: Final Quality Edition") as demo:
+    gr.Markdown("# 👔 Virtual Try-On: The Final Quality Edition")
     with gr.Row():
         with gr.Column(scale=2):
-            p_url = gr.Textbox(label="Person in Suit URL", value="https://img.freepik.com/premium-photo/business-man-suit-white-transparent-background_457222-10395.jpg")
-            f_url = gr.Textbox(label="Fabric Pattern URL", value="https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcT3ajOubmAW68ZHz8aFu4WjfwBxPk1bUsd3ng&s")
+            p_upload = gr.Image(type="pil", label="Person in Suit")
+            p_url = gr.Textbox(label="Person URL")
+            f_upload = gr.Image(type="pil", label="Fabric Pattern")
+            f_url = gr.Textbox(label="Fabric URL")
             btn = gr.Button("Generate Perfect Result", variant="primary")
         with gr.Column(scale=3):
             gallery = gr.Gallery(columns=3, object_fit="cover", height=512)
+            mask_display = gr.Image(label="The Final, Precise Mask Used")
     
-    def ui_fn(person_url, fabric_url):
-        person = load_image(person_url)
-        fabric = load_image(fabric_url)
-        if person is None or fabric is None: raise gr.Error("Missing an image.")
-        person_resized = person.resize(TARGET_SIZE, Image.Resampling.LANCZOS)
-        mask = generate_precise_mask(person_resized)
-        results = create_perfect_result(fabric, person_resized, mask)
-        return results
-
-    btn.click(fn=ui_fn, inputs=[p_url, f_url], outputs=[gallery])
+    btn.click(
+        fn=generate_automatic_tryon,
+        inputs=[p_upload, p_url, f_upload, f_url],
+        outputs=[gallery, mask_display, mask_display]
+    )
 
-# Mount the FastAPI app and the Gradio app together
-app = gr.mount_gradio_app(app, demo, path="/")
+demo.launch()