Update app.py

app.py

@@ -1,43 +1,54 @@
 import os
 import random
+import time
 import gradio as gr
 import torch
-from diffusers import StableDiffusionPipeline  # Changed import
-from transformers import CLIPTextModel, CLIPTokenizer
+from diffusers import DiffusionPipeline, LCMScheduler
 
-# Configuration
-MODEL_ID = "CompVis/stable-diffusion-v1-4"  # Changed to working model
+# Configuration - Using optimized Flux model with LCM
+MODEL_ID = "ByteDance/Hyper-SD"
 MODEL_CACHE = "model_cache"
 os.makedirs(MODEL_CACHE, exist_ok=True)
 
+# Load model with CPU optimizations
 def get_pipeline():
-    pipe = StableDiffusionPipeline.from_pretrained(
+    pipe = DiffusionPipeline.from_pretrained(
         MODEL_ID,
         torch_dtype=torch.float32,
         cache_dir=MODEL_CACHE,
         safety_checker=None,
         use_safetensors=True
     )
+    
+    # Configure for fast generation
+    pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
     pipe = pipe.to("cpu")
     pipe.enable_attention_slicing()
     return pipe
 
 # Load model
+print("Loading model...")
+start_time = time.time()
 pipeline = get_pipeline()
+load_time = time.time() - start_time
+print(f"Model loaded in {load_time:.2f} seconds")
 
 def generate_image(
     prompt: str,
-    negative_prompt: str = "",
+    negative_prompt: str = "blurry, low quality, cartoon, drawing, text",
     width: int = 768,
     height: int = 768,
     seed: int = -1,
-    guidance_scale: float = 7.5,
-    num_inference_steps: int = 25
+    guidance_scale: float = 2.0,
+    num_inference_steps: int = 4
 ):
+    # Set seed if not provided
     if seed == -1:
         seed = random.randint(0, 2147483647)
     generator = torch.Generator(device="cpu").manual_seed(seed)
     
+    # Generate image with timing
+    start_gen = time.time()
     with torch.no_grad():
         image = pipeline(
             prompt=prompt,
@@ -48,34 +59,78 @@ def generate_image(
             num_inference_steps=num_inference_steps,
             generator=generator
         ).images[0]
+    gen_time = time.time() - start_gen
     
-    return image, seed
+    print(f"Generated {width}x{height} image in {gen_time:.2f} seconds")
+    return image, seed, f"Generated in {gen_time:.2f}s | Loaded in {load_time:.2f}s"
 
-with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 🌀 FLUX-Pro Image Generator")
+# Create optimized interface
+with gr.Blocks(theme=gr.themes.Soft(primary_hue="teal")) as demo:
+    gr.Markdown("""
+    # ⚡ FLUX Turbo Generator
+    **Professional Quality Images · Lightning Fast CPU Generation**
+    """)
     
     with gr.Row():
-        with gr.Column():
-            prompt = gr.Textbox(label="Prompt", lines=3)
-            negative_prompt = gr.Textbox(label="Negative Prompt", value="blurry, low quality")
-            generate_btn = gr.Button("Generate", variant="primary")
+        with gr.Column(scale=1):
+            prompt = gr.Textbox(
+                label="Prompt",
+                placeholder="Describe a professional, high-quality image...",
+                lines=3
+            )
+            negative_prompt = gr.Textbox(
+                label="Negative Prompt",
+                value="blurry, low quality, cartoon, drawing, text"
+            )
+            generate_btn = gr.Button("Generate Image", variant="primary")
             
-            with gr.Accordion("Advanced", open=False):
-                width = gr.Slider(512, 1024, value=768, step=64, label="Width")
-                height = gr.Slider(512, 1024, value=768, step=64, label="Height")
-                guidance = gr.Slider(1.0, 15.0, value=7.5, step=0.5, label="Guidance")
-                steps = gr.Slider(15, 50, value=25, step=1, label="Steps")
+            with gr.Accordion("Advanced Settings", open=False):
+                with gr.Row():
+                    width = gr.Slider(512, 1024, value=768, step=64, label="Width")
+                    height = gr.Slider(512, 1024, value=768, step=64, label="Height")
+                with gr.Row():
+                    guidance = gr.Slider(1.0, 5.0, value=2.0, step=0.1, label="Guidance")
+                    steps = gr.Slider(1, 8, value=4, step=1, label="Steps")
                 seed = gr.Number(label="Seed", value=-1)
         
-        with gr.Column():
-            output_image = gr.Image(label="Result", type="pil")
+        with gr.Column(scale=1):
+            output_image = gr.Image(label="Result", type="pil", height=500)
             used_seed = gr.Textbox(label="Used Seed")
-
+            perf_info = gr.Textbox(label="Performance Info")
+    
+    # Generation handler
     generate_btn.click(
         generate_image,
         inputs=[prompt, negative_prompt, width, height, seed, guidance, steps],
-        outputs=[output_image, used_seed]
+        outputs=[output_image, used_seed, perf_info]
+    )
+    
+    # Professional examples
+    gr.Examples(
+        examples=[
+            [
+                "Professional photograph of a futuristic city at golden hour, cinematic lighting, ultra-detailed",
+                "blurry, cartoon, drawing, text, watermark",
+                768,
+                768
+            ],
+            [
+                "Hyperrealistic portrait of a wise elderly man, detailed wrinkles, studio lighting, 8k resolution",
+                "anime, cartoon, deformed, ugly",
+                768,
+                1024
+            ],
+            [
+                "Majestic mountain landscape with crystal clear lake reflection, autumn colors, sharp focus",
+                "low quality, blurry, people, buildings",
+                1024,
+                768
+            ]
+        ],
+        inputs=[prompt, negative_prompt, width, height],
+        label="Professional Examples"
     )
 
+# Launch the app
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(server_name="0.0.0.0", server_port=int(os.getenv("PORT", 7860)))