README.md

metadata

title: Coloring Books
emoji: 🐠
colorFrom: pink
colorTo: green
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false

🌟 AI Coloring Book & Style Generator

A Hugging Face Space powered by Stable Diffusion XL, allowing users to generate AI-generated artwork in different styles. The app also supports LoRA fine-tuning, enabling specialized coloring book effects.

🧠 Concept

This application is built on Stable Diffusion XL (SDXL), a text-to-image generative AI model. Users can:

• Enter a text prompt to describe the image they want.
• Select a predefined artistic style (e.g., Neonpunk, Retro Cyberpunk, Dark Fantasy).
• Enable LoRA (Low-Rank Adaptation) for Coloring Book Style, which enhances black-and-white outlines.

This makes the tool useful for: ✅ Artists looking for AI-generated inspiration.
✅ Educators wanting customized coloring book images.
✅ AI enthusiasts experimenting with fine-tuned models.

---

⚙️ How It Works

1. User enters a text prompt describing the desired image.
2. User selects an artistic style from the dropdown menu.
3. User enables or disables LoRA (if they want a coloring book effect).
4. The app processes the request using Stable Diffusion XL on a GPU.
5. The final image is generated and displayed.

🚀 Runs on Hugging Face Spaces with ZeroGPU allocation!
⏳ GPU is only used during image generation, saving resources.

---

🛠️ Code Overview

📌 1. Loading the Stable Diffusion Model

The app loads the Stable Diffusion XL model and keeps it cached in memory.

@lru_cache(maxsize=1)
def load_pipeline(use_lora: bool):
    """Load Stable Diffusion XL pipeline and apply LoRA weights (if enabled)."""

    pipe = StableDiffusionXLPipeline.from_pretrained(
        "stabilityai/stable-diffusion-xl-base-1.0",
        torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
        use_safetensors=True
    )

    pipe.to("cpu")  # Keep on CPU until needed

    if use_lora:
        pipe.load_lora_weights(color_book_lora_path)

    return pipe

🔹 Uses LoRA for coloring book style when enabled.
🔹 Model stays on CPU until a request is made.

---

📌 2. Handling Image Generation

The generate_image() function processes user input, applies styles, and runs the Stable Diffusion pipeline.

@spaces.GPU  # Ensures GPU is only used during image generation
def generate_image(prompt: str, style_name: str, use_lora: bool):
    """Generate an image based on user input and selected style."""
    
    pipeline = load_pipeline(use_lora)
    pipeline.to("cuda")  # Move to GPU

    # Apply artistic style
    style_prompt = styles.get(style_name, {}).get("prompt", "")
    negative_prompt = styles.get(style_name, {}).get("negative_prompt", "")

    if use_lora:
        prompt += color_book_trigger  # Activates LoRA

    image = pipeline(
        prompt=prompt + " " + style_prompt,
        negative_prompt="blurred, ugly, watermark, low resolution, " + negative_prompt,
        num_inference_steps=20,
        guidance_scale=9.0
    ).images[0]

    pipeline.to("cpu")  # Free GPU memory

    return image

🔹 Uses predefined artistic styles.
🔹 LoRA modifies the image style for coloring book effects.
🔹 GPU is only used for generation, then freed for efficiency.

---

📌 3. Gradio UI for Hugging Face Spaces

The Gradio Interface provides a simple UI for users.

interface = gr.Interface(
    fn=generate_image,
    inputs=[
        gr.Textbox(label="Enter Your Prompt", placeholder="A cute lion"),
        gr.Dropdown(label="Select a Style", choices=list(styles.keys()), value="None"),
        gr.Checkbox(label="Use Coloring Book LoRA", value=False)
    ],
    outputs=gr.Image(label="Generated Image"),
    title="🎨 AI Coloring Book & Style Generator",
    description="Generate AI-powered art using Stable Diffusion XL. Choose a style or enable a LoRA fine-tuned coloring book effect."
)

if __name__ == "__main__":
    interface.launch()

🔹 User-friendly UI with textbox, dropdown, and checkbox controls.
🔹 Outputs a generated image based on user selections.
🔹 Runs efficiently on Hugging Face Spaces with ZeroGPU allocation.

---

📦 Installation & Running Locally

If you want to run this locally, install dependencies:

pip install torch diffusers transformers accelerate gradio peft safetensors spaces

Then, run:

python app.py

---

📝 Requirements (requirements.txt)

This app requires the following dependencies:

torch
diffusers
transformers
accelerate
gradiop
peft
safetensors
spaces

---

🌟 Future Improvements

✅ Add more LoRA styles (e.g., Anime, Watercolor, Sketch).
✅ Optimize GPU usage with mixed-precision inference.
✅ Improve UI with real-time previews.

---

👩‍💻 Author

👨‍💻 Developed by Walid Ahmed

📌 Powered by Hugging Face 🤗 and Stable Diffusion XL.

🌟 Try it out and start generating amazing AI art! 🎨