Update app.py

app.py

@@ -13,12 +13,12 @@ API_URL = f"https://api-inference.huggingface.co/models/{MODEL_NAME}"
 headers = {"Authorization": f"Bearer {HF_API_TOKEN}"}
 WATERMARK_TEXT = "SelamGPT"
 MAX_RETRIES = 3
-TIMEOUT = 60
-EXECUTOR = ThreadPoolExecutor(max_workers=2)
+TIMEOUT = 45  # Reduced timeout for faster failover
+EXECUTOR = ThreadPoolExecutor(max_workers=3)  # Increased workers
 
-# ===== WATERMARK FUNCTION =====
+# ===== OPTIMIZED WATERMARK FUNCTION (WebP) =====
 def add_watermark(image_bytes):
-    """Add watermark with optimized PNG output"""
+    """Add watermark with optimized WebP output (85% quality)"""
     try:
         image = Image.open(io.BytesIO(image_bytes)).convert("RGB")
         draw = ImageDraw.Draw(image)
@@ -36,69 +36,66 @@ def add_watermark(image_bytes):
         draw.text((x+1, y+1), WATERMARK_TEXT, font=font, fill=(0, 0, 0, 128))
         draw.text((x, y), WATERMARK_TEXT, font=font, fill=(255, 255, 255))
         
-        # Convert to optimized PNG
-        img_byte_arr = io.BytesIO()
-        image.save(img_byte_arr, format='PNG', optimize=True, quality=85)
-        img_byte_arr.seek(0)
-        return Image.open(img_byte_arr)
+        # Convert to WebP (faster + smaller)
+        webp_buffer = io.BytesIO()
+        image.save(webp_buffer, format="WEBP", quality=85, method=6)  # 85% quality, best compression
+        webp_buffer.seek(0)
+        return Image.open(webp_buffer)
     except Exception as e:
         print(f"Watermark error: {str(e)}")
         return Image.open(io.BytesIO(image_bytes))
 
-# ===== IMAGE GENERATION =====
+# ===== FASTER IMAGE GENERATION =====
 def generate_image(prompt):
     if not prompt.strip():
         return None, "⚠️ Please enter a prompt"
     
+    # Faster generation parameters
+    params = {
+        "height": 768,  # Slightly smaller = faster
+        "width": 768,
+        "num_inference_steps": 20,  # Reduced from 30
+        "guidance_scale": 7.0,  # Slightly lower for speed
+        "options": {"wait_for_model": False}  # Don't wait if model is loading
+    }
+    
     def api_call():
         return requests.post(
             API_URL,
             headers=headers,
             json={
                 "inputs": prompt,
-                "parameters": {
-                    "height": 1024,
-                    "width": 1024,
-                    "num_inference_steps": 30
-                },
-                "options": {"wait_for_model": True}
+                "parameters": params
             },
             timeout=TIMEOUT
         )
     
     for attempt in range(MAX_RETRIES):
         try:
-            future = EXECUTOR.submit(api_call)
-            response = future.result()
+            start_time = time.time()
+            response = EXECUTOR.submit(api_call).result()
             
             if response.status_code == 200:
-                return add_watermark(response.content), "✔️ Generation successful"
+                gen_time = time.time() - start_time
+                return add_watermark(response.content), f"✔️ Generated in {gen_time:.1f}s"
             elif response.status_code == 503:
-                wait_time = (attempt + 1) * 15
-                print(f"Model loading, waiting {wait_time}s...")
-                time.sleep(wait_time)
+                if attempt < MAX_RETRIES - 1:
+                    time.sleep(5 * (attempt + 1))  # Progressive backoff
                 continue
             else:
                 return None, f"⚠️ API Error: {response.text[:200]}"
         except requests.Timeout:
-            return None, f"⚠️ Timeout: Model took >{TIMEOUT}s to respond"
+            return None, f"⚠️ Timeout: Model took >{TIMEOUT}s"
         except Exception as e:
-            return None, f"⚠️ Unexpected error: {str(e)[:200]}"
+            return None, f"⚠️ Error: {str(e)[:200]}"
     
-    return None, "⚠️ Failed after multiple attempts. Please try later."
-
-# ===== GRADIO THEME =====
-theme = gr.themes.Default(
-    primary_hue="emerald",
-    secondary_hue="amber",
-    font=[gr.themes.GoogleFont("Poppins"), "Arial", "sans-serif"]
-)
+    return None, "⚠️ Failed after retries. Try again."
 
 # ===== GRADIO INTERFACE =====
-with gr.Blocks(theme=theme, title="SelamGPT Image Generator") as demo:
+with gr.Blocks(title="SelamGPT Image Generator") as demo:
     gr.Markdown("""
     # 🎨 SelamGPT Image Generator
-    *Powered by Stable Diffusion XL (1024x1024 PNG output)*
+    *Optimized for speed (WebP output @ 768px)*
     """)
     
     with gr.Row():
@@ -106,8 +103,7 @@ with gr.Blocks(theme=theme, title="SelamGPT Image Generator") as demo:
             prompt_input = gr.Textbox(
                 label="Describe your image",
                 placeholder="A futuristic Ethiopian city with flying cars...",
-                lines=3,
-                max_lines=5
+                lines=3
             )
             with gr.Row():
                 generate_btn = gr.Button("Generate Image", variant="primary")
@@ -115,18 +111,18 @@ with gr.Blocks(theme=theme, title="SelamGPT Image Generator") as demo:
             
             gr.Examples(
                 examples=[
-                    ["An ancient Aksumite warrior in cyberpunk armor, 4k detailed"],
-                    ["Traditional Ethiopian coffee ceremony in zero gravity"],
-                    ["Portrait of a Habesha queen with golden jewelry"]
+                    ["An ancient Aksumite warrior in cyberpunk armor"],
+                    ["Traditional Ethiopian coffee ceremony"],
+                    ["Habesha queen with jewelry"]
                 ],
                 inputs=prompt_input
             )
             
         with gr.Column(scale=2):
             output_image = gr.Image(
-                label="Generated Image",
+                label="Generated Image (WebP)",
                 type="pil",
-                format="png",
+                format="webp",  # WebP output
                 height=512
             )
             status_output = gr.Textbox(
@@ -147,5 +143,5 @@ with gr.Blocks(theme=theme, title="SelamGPT Image Generator") as demo:
     )
 
 if __name__ == "__main__":
-    demo.queue(max_size=2)
+    demo.queue(concurrency_count=3)  # Increased concurrency
     demo.launch(server_name="0.0.0.0", server_port=7860)