Update app.py

app.py

@@ -10,32 +10,26 @@ import spaces
 import torch
 from diffusers import StableDiffusionXLPipeline, EulerAncestralDiscreteScheduler
 
-# Use environment variables for flexibility
-MODEL_ID = os.getenv("MODEL_ID", "sd-community/sdxl-flash")
+if not torch.cuda.is_available():
+    DESCRIPTION += "\n<p>Running on CPU 🥶 This demo may not work on CPU.</p>"
+
+MAX_SEED = np.iinfo(np.int32).max
+CACHE_EXAMPLES = torch.cuda.is_available() and os.getenv("CACHE_EXAMPLES", "1") == "1"
 MAX_IMAGE_SIZE = int(os.getenv("MAX_IMAGE_SIZE", "4096"))
 USE_TORCH_COMPILE = os.getenv("USE_TORCH_COMPILE", "0") == "1"
 ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD", "0") == "1"
-BATCH_SIZE = int(os.getenv("BATCH_SIZE", "1"))  # Allow generating multiple images at once
 
-# Determine device and load model outside of function for efficiency
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
-pipe = StableDiffusionXLPipeline.from_pretrained(
-    MODEL_ID,
-    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
-    use_safetensors=True,
-    add_watermarker=False,
-).to(device)
-pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
-
-# Torch compile for potential speedup (experimental)
-if USE_TORCH_COMPILE:
-    pipe.compile()
-
-# CPU offloading for larger RAM capacity (experimental)
-if ENABLE_CPU_OFFLOAD:
-    pipe.enable_model_cpu_offload()
 
-MAX_SEED = np.iinfo(np.int32).max
+if torch.cuda.is_available():
+    pipe = StableDiffusionXLPipeline.from_pretrained(
+        "sd-community/sdxl-flash",
+        torch_dtype=torch.float16,
+        use_safetensors=True,
+        add_watermarker=False
+    )
+    pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
+    pipe.to("cuda")
 
 def save_image(img):
     unique_name = str(uuid.uuid4()) + ".png"
@@ -47,58 +41,50 @@ def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
         seed = random.randint(0, MAX_SEED)
     return seed
 
-@spaces.GPU(duration=35, enable_queue=True)
+@spaces.GPU(duration=30, queue=False)
 def generate(
     prompt: str,
     negative_prompt: str = "",
     use_negative_prompt: bool = False,
-    seed: int = 1,
+    seed: int = 0,
     width: int = 1024,
     height: int = 1024,
     guidance_scale: float = 3,
-    num_inference_steps: int = 30,
+    num_inference_steps: int = 25,
     randomize_seed: bool = False,
-    use_resolution_binning: bool = True, 
-    num_images: int = 1,  # Number of images to generate
+    use_resolution_binning: bool = True,
     progress=gr.Progress(track_tqdm=True),
 ):
+    pipe.to(device)
     seed = int(randomize_seed_fn(seed, randomize_seed))
-    generator = torch.Generator(device=device).manual_seed(seed)
+    generator = torch.Generator().manual_seed(seed)   
 
-    # Improved options handling
     options = {
-        "prompt": [prompt] * num_images,
-        "negative_prompt": [negative_prompt] * num_images if use_negative_prompt else None,
-        "width": width,
-        "height": height,
-        "guidance_scale": guidance_scale,
-        "num_inference_steps": num_inference_steps,
-        "generator": generator,
-        "output_type": "pil",
-    }
-
-    # Use resolution binning for faster generation with less VRAM usage
-    if use_resolution_binning:
-        options["use_resolution_binning"] = True
+        "prompt":prompt,
+        "negative_prompt":negative_prompt,
+        "width":width,
+        "height":height,
+        "guidance_scale":guidance_scale,
+        "num_inference_steps":num_inference_steps,
+        "generator":generator,
+        "use_resolution_binning":use_resolution_binning,
+        "output_type":"pil",
 
-    # Generate images potentially in batches
-    images = []
-    for i in range(0, num_images, BATCH_SIZE):
-        batch_options = options.copy()
-        batch_options["prompt"] = options["prompt"][i:i+BATCH_SIZE]
-        if "negative_prompt" in batch_options:
-            batch_options["negative_prompt"] = options["negative_prompt"][i:i+BATCH_SIZE]
-        images.extend(pipe(**batch_options).images)
+    }
+    
+    images = pipe(**options).images
 
     image_paths = [save_image(img) for img in images]
     return image_paths, seed
 
+
 examples = [
     "a cat eating a piece of cheese",
-    "a ROBOT riding a BLUE horse on Mars, photorealistic, 4k",
-    "Ironman VS Hulk, ultrarealistic",
+    "a ROBOT riding a BLUE horse on Mars, photorealistic",
+    "a cartoon of a IRONMAN fighting with HULK, wall painting",
+    "a cute robot artist painting on an easel, concept art",
     "Astronaut in a jungle, cold color palette, oil pastel, detailed, 8k",
-    "An alien holding a sign board containing the word 'Flash', futuristic, neonpunk",
+    "An alien grasping a sign board contain word 'Flash', futuristic, neonpunk, detailed",
     "Kids going to school, Anime style"
 ]
 
@@ -109,9 +95,9 @@ footer {
     visibility: hidden
 }
 '''
-
 with gr.Blocks(css=css) as demo:
-    gr.Markdown("""# SDXL Flash""")
+    gr.Markdown("""# SDXL Flash
+        ### First Image processing takes time then images generate faster.""")
     with gr.Group():
         with gr.Row():
             prompt = gr.Text(
@@ -122,15 +108,8 @@ with gr.Blocks(css=css) as demo:
                 container=False,
             )
             run_button = gr.Button("Run", scale=0)
-        result = gr.Gallery(label="Result", columns=1, show_label=False) 
+        result = gr.Gallery(label="Result", columns=1)
     with gr.Accordion("Advanced options", open=False):
-        num_images = gr.Slider(
-            label="Number of Images",
-            minimum=1,
-            maximum=4,
-            step=1,
-            value=1,
-        )
         with gr.Row():
             use_negative_prompt = gr.Checkbox(label="Use negative prompt", value=True)
             negative_prompt = gr.Text(
@@ -183,7 +162,9 @@ with gr.Blocks(css=css) as demo:
     gr.Examples(
         examples=examples,
         inputs=prompt,
-        cache_examples=False
+        outputs=[result, seed],
+        fn=generate,
+        cache_examples=CACHE_EXAMPLES,
     )
 
     use_negative_prompt.change(
@@ -210,7 +191,6 @@ with gr.Blocks(css=css) as demo:
             guidance_scale,
             num_inference_steps,
             randomize_seed,
-            num_images
         ],
         outputs=[result, seed],
         api_name="run",