Update app.py

app.py

@@ -1,274 +1,142 @@
 import gradio as gr
 import numpy as np
 import random
-import spaces
-import torch
-from diffusers import DiffusionPipeline
-from PIL import Image, ImageEnhance
 import io
 import zipfile
+from PIL import Image, ImageEnhance
+import torch
+from diffusers import DiffusionPipeline
+from transformers import CLIPTextModel, CLIPTokenizer, T5EncoderModel, T5TokenizerFast
 
-# Initialize model and settings
+# Configuration
 dtype = torch.bfloat16
 device = "cuda" if torch.cuda.is_available() else "cpu"
-pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16).to(device)
+
+# Load the model
+pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=dtype).to(device)
 
 # Constants
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 2048
 
-# Helper function to apply image filters
-def apply_filters(image, brightness, contrast, saturation):
-    enhancer = ImageEnhance.Brightness(image)
-    image = enhancer.enhance(brightness)
-    
-    enhancer = ImageEnhance.Contrast(image)
-    image = enhancer.enhance(contrast)
-    
-    enhancer = ImageEnhance.Color(image)
-    image = enhancer.enhance(saturation)
-    
-    return image
-
-# Helper function to create a ZIP file of images
-def create_zip(images):
-    zip_buffer = io.BytesIO()
-    with zipfile.ZipFile(zip_buffer, "a", zipfile.ZIP_DEFLATED) as zip_file:
-        for i, img in enumerate(images):
-            img_byte_arr = io.BytesIO()
-            img.save(img_byte_arr, format="PNG")
-            zip_file.writestr(f"image_{i+1}.png", img_byte_arr.getvalue())
-    return zip_buffer.getvalue()
-
-@spaces.GPU(duration=190)
-def infer(prompt, num_outputs=1, seed=42, randomize_seed=False, width=1024, height=1024, guidance_scale=5.0, num_inference_steps=28, output_format="png", brightness=1.0, contrast=1.0, saturation=1.0, style="None", progress=gr.Progress(track_tqdm=True)):
-    images = []
-    seeds = []
-
+# Define the inference function
+def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, guidance_scale=5.0, num_inference_steps=28, num_variations=1, brightness=1.0, contrast=1.0, saturation=1.0, style="Style1"):
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
+    generator = torch.Generator().manual_seed(seed)
     
-    for _ in range(num_outputs):
-        generator = torch.Generator(device=device).manual_seed(seed)
+    images = []
+    for _ in range(num_variations):
         image = pipe(
-            prompt=prompt, 
-            width=width, 
-            height=height, 
-            num_inference_steps=num_inference_steps, 
-            generator=generator, 
+            prompt=prompt,
+            width=width,
+            height=height,
+            num_inference_steps=num_inference_steps,
+            generator=generator,
             guidance_scale=guidance_scale
         ).images[0]
-
-        # Apply filters
-        image = apply_filters(image, brightness, contrast, saturation)
-
-        # Apply preset style if selected
-        if style != "None":
-            # Example of applying a filter based on a selected style
-            # You would expand this to include actual styles, e.g., cartoon, watercolor, etc.
-            if style == "Black & White":
-                image = image.convert("L").convert("RGB")
-            elif style == "Sepia":
-                sepia_filter = np.array(image)
-                sepia_filter = np.dot(sepia_filter[...,:3], [0.393, 0.769, 0.189])
-                sepia_filter = np.clip(sepia_filter, 0, 255)
-                image = Image.fromarray(sepia_filter.astype('uint8'))
-        
+        # Apply image adjustments
+        image = adjust_image(image, brightness, contrast, saturation)
         images.append(image)
-        seeds.append(seed)
-        seed += 1  # Increment seed for the next image
-
-    # Optionally convert image format
-    if output_format.lower() != "png":
-        images = [img.convert(output_format.upper()) for img in images]
     
-    return images, seeds, create_zip(images)
+    # Apply style (dummy implementation, replace with actual style application)
+    images = [apply_style(img, style) for img in images]
+    
+    return images, seed
 
-# Example prompts for users to try
-examples = [
-    ["a tiny astronaut hatching from an egg on the moon", 1],
-    ["a cat holding a sign that says hello world", 2],
-    ["an anime illustration of a wiener schnitzel", 3],
-]
+def adjust_image(image, brightness, contrast, saturation):
+    enhancer = ImageEnhance.Brightness(image)
+    image = enhancer.enhance(brightness)
+    enhancer = ImageEnhance.Contrast(image)
+    image = enhancer.enhance(contrast)
+    enhancer = ImageEnhance.Color(image)
+    image = enhancer.enhance(saturation)
+    return image
 
-# CSS styling for modern look
+def apply_style(image, style):
+    # Dummy style application
+    return image
+
+def download_all(images):
+    with io.BytesIO() as buffer:
+        with zipfile.ZipFile(buffer, 'w') as zipf:
+            for i, img in enumerate(images):
+                img_byte_arr = io.BytesIO()
+                img.save(img_byte_arr, format="PNG")
+                zipf.writestr(f'image_{i}.png', img_byte_arr.getvalue())
+        buffer.seek(0)
+        return buffer.getvalue()
+
+# Gradio interface
 css = """
 #col-container {
     margin: 0 auto;
-    max-width: 1000px;
-    padding: 20px;
-    background-color: #f9f9f9;
-    border-radius: 10px;
-    box-shadow: 0px 0px 20px rgba(0,0,0,0.1);
-}
-
-#title {
-    font-family: 'Arial', sans-serif;
-    color: #333;
-    text-align: center;
-    margin-bottom: 20px;
-}
-
-#advanced-settings {
-    background-color: #f1f1f1;
-    border-radius: 8px;
-    padding: 10px;
-}
-
-#output-container {
-    text-align: center;
-    margin-top: 20px;
+    max-width: 720px;
 }
 """
 
+examples = [
+    "a tiny astronaut hatching from an egg on the moon",
+    "a cat holding a sign that says hello world",
+    "an anime illustration of a wiener schnitzel",
+]
+
 with gr.Blocks(css=css) as demo:
     with gr.Column(elem_id="col-container"):
-        
-        # Title and Description
-        gr.Markdown(f"""
-        <h1 id="title">FLUX.1 [dev] - Feature-Rich Text-to-Image Generator</h1>
-        <p style="text-align:center; color:#555;">
-        Unleash your creativity with this advanced text-to-image generator. Customize prompts, generate multiple images, apply filters, and more!
-        </p>
+        gr.Markdown(f"""# FLUX.1 [dev]
+12B param rectified flow transformer guidance-distilled from [FLUX.1 [pro]](https://blackforestlabs.ai/)  
+[[non-commercial license](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)] [[blog](https://blackforestlabs.ai/announcing-black-forest-labs/)] [[model](https://huggingface.co/black-forest-labs/FLUX.1-dev)]
         """)
         
         with gr.Row():
-            prompt = gr.Textbox(
-                label="Prompt",
-                show_label=False,
-                max_lines=1,
-                placeholder="Enter your creative prompt...",
-                container=False,
-                interactive=True
-            )
-
-        num_outputs = gr.Slider(
-            label="Number of Variations",
-            minimum=1,
-            maximum=10,
-            step=1,
-            value=1,
-            interactive=True
-        )
-
-        run_button = gr.Button("Generate", elem_id="generate-button")
-
-        # Output image gallery and settings
-        with gr.Row(elem_id="output-container"):
-            gallery = gr.Gallery(label="Generated Images", show_label=False).style(grid=[4], height="auto")
-            output_format = gr.Radio(
-                label="Output Format",
-                choices=["png", "jpeg", "bmp"],
-                value="png",
-                interactive=True
-            )
-
-        # Advanced settings
-        with gr.Accordion("Advanced Settings", open=False, elem_id="advanced-settings"):
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
-                interactive=True
-            )
-            randomize_seed = gr.Checkbox(label="Randomize Seed", value=True, interactive=True)
-
+            prompt = gr.Textbox(label="Prompts (comma-separated)", placeholder="Enter multiple prompts separated by commas", lines=2)
+            run_button = gr.Button("Run", scale=0)
+        
+        result = gr.Gallery(label="Image Gallery").style(height=400)
+        
+        with gr.Accordion("Advanced Settings", open=False):
+            seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0)
+            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
+            
             with gr.Row():
-                width = gr.Slider(
-                    label="Width",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,
-                    interactive=True
-                )
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,
-                    interactive=True
-                )
-
+                width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
+                height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
+            
             with gr.Row():
-                guidance_scale = gr.Slider(
-                    label="Guidance Scale",
-                    minimum=1,
-                    maximum=15,
-                    step=0.1,
-                    value=5.0,
-                    interactive=True
-                )
-                num_inference_steps = gr.Slider(
-                    label="Number of Inference Steps",
-                    minimum=1,
-                    maximum=50,
-                    step=1,
-                    value=28,
-                    interactive=True
-                )
-
-            # Image filter sliders
-            brightness = gr.Slider(
-                label="Brightness",
-                minimum=0.5,
-                maximum=2.0,
-                step=0.1,
-                value=1.0,
-                interactive=True
-            )
-            contrast = gr.Slider(
-                label="Contrast",
-                minimum=0.5,
-                maximum=2.0,
-                step=0.1,
-                value=1.0,
-                interactive=True
-            )
-            saturation = gr.Slider(
-                label="Saturation",
-                minimum=0.5,
-                maximum=2.0,
-                step=0.1,
-                value=1.0,
-                interactive=True
-            )
-
-            # Preset styles
-            style = gr.Dropdown(
-                label="Preset Styles",
-                choices=["None", "Black & White", "Sepia", "Vivid Colors"],
-                value="None",
-                interactive=True
-            )
-
-        # Interactive Examples
+                guidance_scale = gr.Slider(label="Guidance Scale", minimum=1, maximum=15, step=0.1, value=3.5)
+                num_inference_steps = gr.Slider(label="Number of inference steps", minimum=1, maximum=50, step=1, value=28)
+            
+            with gr.Row():
+                num_variations = gr.Slider(label="Number of Variations", minimum=1, maximum=10, step=1, value=1)
+                brightness = gr.Slider(label="Brightness", minimum=0.0, maximum=2.0, step=0.1, value=1.0)
+                contrast = gr.Slider(label="Contrast", minimum=0.0, maximum=2.0, step=0.1, value=1.0)
+                saturation = gr.Slider(label="Saturation", minimum=0.0, maximum=2.0, step=0.1, value=1.0)
+            
+            style = gr.Dropdown(label="Select Style", choices=["Style1", "Style2", "Style3"], value="Style1")
+            download_all_button = gr.Button("Download All")
+        
         gr.Examples(
             examples=examples,
             fn=infer,
-            inputs=[prompt, num_outputs],
-            outputs=[gallery, seed],
-            cache_examples="lazy",
-            label="Try these examples:"
+            inputs=[prompt],
+            outputs=[result, seed],
+            cache_examples="lazy"
         )
 
-        # Download all images as a ZIP file
-        download_button = gr.Button("Download All Images")
-        zip_file = gr.File(label="Download", show_label=False)
-        download_button.click(
-            fn=create_zip,
-            inputs=[gallery],
-            outputs=[zip_file]
-        )
+    def run_inference(prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, num_variations, brightness, contrast, saturation, style):
+        images, seed = infer(prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, num_variations, brightness, contrast, saturation, style)
+        return images, seed
 
-        # Link button to trigger inference
-        run_button.click(
-            fn=infer,
-            inputs=[prompt, num_outputs, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, output_format, brightness, contrast, saturation, style],
-            outputs=[gallery, seed, zip_file]
-        )
+    gr.on(
+        triggers=[run_button.click, prompt.submit],
+        fn=run_inference,
+        inputs=[prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, num_variations, brightness, contrast, saturation, style],
+        outputs=[result, seed]
+    )
+
+    def download_all_callback(images):
+        return download_all(images)
+
+    download_all_button.click(fn=download_all_callback, inputs=[result], outputs=[gr.File(label="Download All Images")])
 
 demo.launch()