A text-to-image generation model built on Stable Diffusion 1.5 and fine-tuned with multiple LoRA modules.
Generated images are released under Creative Commons Zero (CC0), free to use without attribution.
(Base model weights are derived from SD 1.5, licensed under CreativeML OpenRAIL-M; all LoRA modules used are CC0-1.0)
Model Description
Free-Image-Generation is a fine-tuned text-to-image model based on the Stable Diffusion 1.5 architecture with 1.07 billion parameters, merging multiple LoRA modules.
- Base weights: SD 1.5 (CreativeML OpenRAIL-M license)
- LoRA modules: CC0-1.0
- Generated outputs: CC0 — fully free to use without attribution
All images generated with this model are released under the Creative Commons Zero (CC0) license, making them freely usable for any purpose without attribution requirements.
Features
- 1.07B parameters
- Text-to-image generation
- Fine-tuned with multiple LoRA modules
- Compatible with Hugging Face Diffusers library
- Generates images under CC0 license
Installation
Prerequisites
- Python 3.7 or higher
- CUDA-compatible GPU (recommended)
Setup
Install the required dependencies:
pip install diffusers transformers torch
Usage Examples
Basic Usage
from diffusers import DiffusionPipeline
import torch
# Load model from Hugging Face
pipe = DiffusionPipeline.from_pretrained("aiyouthalliance/Free-Image-Generation")
# Move to GPU if available
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
pipe = pipe.to(device)
# Generate image
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]
# Save the generated image
image.save("astronaut_jungle.png")
Memory-Optimized Usage
# Load in half precision for memory efficiency
pipe = DiffusionPipeline.from_pretrained(
"aiyouthalliance/Free-Image-Generation",
torch_dtype=torch.float16
)
pipe = pipe.to(device)
Advanced Parameter Customization
# Customize generation parameters
image = pipe(
prompt="A futuristic cityscape with flying cars, neon lights, detailed, 8k",
num_inference_steps=50,
guidance_scale=7.5,
negative_prompt="blurry, low quality, distorted"
).images[0]
Model Structure
The model repository contains the following components:
- vae: The variational autoencoder component
- unet: The U-Net denoising component
- text_encoder: The text encoder for processing prompts
- tokenizer: For tokenizing the text inputs
- scheduler: Controls the denoising process
- safety_checker: For filtering content
- feature_extractor: For image processing
License
- Base model weights: OpenRAIL-M (derivative of SD 1.5)
- LoRA modules: CC0-1.0
- Generated images: CC0
Users should comply with OpenRAIL-M restrictions for derivative use or redistribution of the base model weights.
Technical Specifications
| Feature | Specification |
|---|---|
| Total Parameters | 1.07 Billion |
| Base Architecture | Stable Diffusion 1.5 |
| Fine-Tuning | Multiple LoRA modules merged |
| Output Resolution | 512×512 pixels |
Citation
@misc{aiyouthalliance2025freeimagegen,
author = {AI Youth Alliance},
title = {Free-Image-Generation},
year = {2025},
publisher = {Hugging Face},
journal = {Hugging Face repository},
howpublished = {\url{https://huggingface.co/aiyouthalliance/Free-Image-Generation}}
}
Contact & Support
- Hugging Face: aiyouthalliance