devsu's picture
Refactor image generation parameters and update examples for SDXL-Turbo model. Removed guidance scale, adjusted inference steps to 1-4, and increased image dimensions to 1024x1024.
17770cb
Raw
History Blame Contribute Delete
12.8 kB
"""
AI Image Generator - Gradio Web App
A web application for generating images using Stable Diffusion
with a user-friendly Gradio interface.
"""
import json
import logging
import os
from typing import List, Optional, Tuple
import gradio as gr
import torch
from PIL import Image
from utils.generation import ImageGenerator, save_image_with_metadata
# Configure logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
logger = logging.getLogger(__name__)
# Global image generator instance
image_generator: Optional[ImageGenerator] = None
def load_examples() -> dict:
"""Load example prompts from JSON file."""
try:
with open('examples.json', 'r', encoding='utf-8') as f:
return json.load(f)
except FileNotFoundError:
logger.warning("examples.json not found, using default examples")
return {
"examples": []
}
except Exception as e:
logger.error(f"Failed to load examples: {str(e)}")
return {
"examples": [],
}
def initialize_generator() -> ImageGenerator:
"""
Initialize the image generator with error handling.
Returns:
ImageGenerator: The initialized generator instance.
Raises:
RuntimeError: If initialization fails.
"""
try:
logger.info("Initializing Stable Diffusion model...")
generator = ImageGenerator()
logger.info("Model initialized successfully")
return generator
except Exception as e:
logger.error(f"Failed to initialize model: {str(e)}")
raise RuntimeError(f"Model initialization failed: {str(e)}")
def generate_images_interface(
prompt: str,
num_inference_steps: int,
# guidance_scale: float, # Commented out - using default value 0.0
seed: Optional[int],
width: int,
height: int,
num_images: int
) -> Tuple[List[str], str]:
"""
Generate images based on user input parameters.
Args:
prompt: Text prompt for image generation.
num_inference_steps: Number of denoising steps.
# guidance_scale: How closely to follow the prompt. # Commented out
seed: Random seed for reproducibility.
width: Image width.
height: Image height.
num_images: Number of images to generate.
Returns:
Tuple[List[str], str]: (list of image paths, status message)
"""
global image_generator
try:
# Initialize generator if not already done
if image_generator is None:
image_generator = initialize_generator()
# Validate inputs
if not prompt or not prompt.strip():
return [], "❌ Please enter a prompt"
logger.info(f"Generating {num_images} image(s) with prompt: '{prompt[:50]}...'")
# Generate images
images = image_generator.generate_images(
prompt=prompt,
num_images=num_images,
num_inference_steps=num_inference_steps,
guidance_scale=0.0, # Using default value since guidance_scale is removed from UI
width=width,
height=height,
seed=seed
)
# Save images and collect paths
saved_paths = []
for i, image in enumerate(images):
metadata = {
"num_inference_steps": num_inference_steps,
"guidance_scale": 0.0, # Using default value
"seed": seed,
"width": width,
"height": height,
"image_number": i + 1
}
filepath = save_image_with_metadata(
image=image,
prompt=prompt,
metadata=metadata
)
saved_paths.append(filepath)
status_msg = f"✅ Successfully generated {len(images)} image(s)!"
logger.info(f"Generation completed: {status_msg}")
return saved_paths, status_msg
except Exception as e:
error_msg = f"❌ Generation failed: {str(e)}"
logger.error(f"Generation error: {str(e)}")
return [], error_msg
def create_interface() -> gr.Blocks:
"""
Create the Gradio interface for the AI Image Generator.
Returns:
gr.Blocks: The configured Gradio interface.
"""
# Custom CSS for better styling
css = """
.gradio-container {
max-width: 100% !important;
margin: 0 !important;
padding: 0 !important;
}
.main-header {
text-align: center;
margin-bottom: 2rem;
}
.param-section {
background: #f8f9fa;
padding: 1rem;
border-radius: 8px;
margin-bottom: 1rem;
}
.examples-section {
background: #f0f8ff;
padding: 1.5rem;
border-radius: 12px;
margin-bottom: 2rem;
border: 2px solid #e1f5fe;
}
.gradio-row {
gap: 2rem !important;
max-width: 100% !important;
}
.gradio-column {
min-width: 0 !important;
flex: 1 !important;
}
.gradio-blocks {
max-width: 100% !important;
}
"""
with gr.Blocks(css=css, title="AI Image Generator") as interface:
# Header
gr.HTML("""
<div class="main-header">
<h1>🎨 AI Image Generator — powered by Stable Diffusion</h1>
<p style="font-size: 1.1em; color: #666;">Enter a prompt to generate an image in seconds</p>
</div>
""")
# Load examples data first
examples_data = load_examples()
# Create examples for different categories
all_examples = []
# Add examples from categories
if "examples" in examples_data:
for category in examples_data["examples"]:
if "prompts" in category:
for prompt_data in category["prompts"][:2]: # Take first 2 from each category
if isinstance(prompt_data, dict):
all_examples.append([
prompt_data["text"],
prompt_data["num_inference_steps"],
# prompt_data["guidance_scale"], # Commented out - using default value
prompt_data["seed"],
prompt_data["width"],
prompt_data["height"],
1 # num_images default
])
else:
all_examples.append([prompt_data, 2, None, 1024, 1024, 1]) # Updated to match new parameters
# Limit to 8 examples for better UI
all_examples = all_examples[:8]
with gr.Row():
# Left column - Input parameters
with gr.Column(scale=2):
gr.Markdown("### 📝 Input Parameters")
# Main prompt input
prompt_input = gr.Textbox(
label="Prompt",
placeholder="A beautiful sunset over mountains, digital art",
lines=3,
max_lines=5
)
# Generation parameters
with gr.Group():
gr.Markdown("#### ⚙️ Generation Settings")
num_inference_steps = gr.Slider(
minimum=1,
maximum=4,
value=2,
step=1,
label="Inference Steps",
info="More steps = better quality, slower generation"
)
# guidance_scale = gr.Slider(
# minimum=1.0,
# maximum=20.0,
# value=7.5,
# step=0.1,
# label="Guidance Scale",
# info="How closely to follow the prompt"
# )
seed = gr.Number(
label="Seed (optional)",
value=None,
precision=0,
info="Leave empty for random generation"
)
# Image parameters
with gr.Group():
gr.Markdown("#### 🖼️ Image Settings")
with gr.Row():
width = gr.Number(
label="Width",
value=1024,
minimum=512,
maximum=1024,
step=64,
precision=0
)
height = gr.Number(
label="Height",
value=1024,
minimum=512,
maximum=1024,
step=64,
precision=0
)
num_images = gr.Slider(
minimum=1,
maximum=4,
value=1,
step=1,
label="Number of Images",
info="Generate multiple variations"
)
# Generate button
generate_btn = gr.Button(
"🎨 Generate Images",
variant="primary",
size="lg"
)
# Right column - Output
with gr.Column(scale=2):
gr.Markdown("### 🖼️ Generated Images")
# Status message
status_output = gr.Textbox(
label="Status",
interactive=False,
value="Ready to generate images!"
)
# Image gallery
gallery = gr.Gallery(
label="Generated Images",
show_label=True,
elem_id="gallery",
columns=2,
rows=2,
height="auto",
object_fit="contain"
)
# Footer
gr.HTML("""
<div style="text-align: center; margin-top: 2rem; padding: 1rem; border-top: 1px solid #eee;">
<p>
Powered by <a href="https://huggingface.co/docs/diffusers/using-diffusers/sdxl_turbo" target="_blank">Stable Diffusion XL Turbo</a> |
Built with <a href="https://gradio.app" target="_blank">Gradio</a>
</p>
</div>
""")
# Examples section - now that all components are defined
with gr.Group():
gr.Markdown("### 💡 Example Prompts - Click to Auto-Fill Form")
gr.Examples(
examples=all_examples,
inputs=[
prompt_input,
num_inference_steps,
# guidance_scale, # Commented out - removed from UI
seed,
width,
height,
num_images
],
label="",
elem_id="examples-section"
)
# Event handlers
generate_btn.click(
fn=generate_images_interface,
inputs=[
prompt_input,
num_inference_steps,
# guidance_scale, # Commented out - removed from UI
seed,
width,
height,
num_images
],
outputs=[gallery, status_output]
)
return interface
def main():
"""Main function to launch the Gradio interface."""
try:
logger.info("Starting AI Image Generator...")
# Create and launch interface
interface = create_interface()
# Launch with appropriate settings
interface.launch(
server_name="0.0.0.0",
server_port=7860,
share=False,
show_error=True,
quiet=False
)
except Exception as e:
logger.error(f"Failed to start application: {str(e)}")
raise
if __name__ == "__main__":
main()