S D_ Pixel Art_ Sprite Sheet_ Generator

Introduction

The SD PixelArt SpriteSheet Generator is a Stable Diffusion model designed to create pixel art sprite sheets from various angles. It merges with other models to produce consistent character views and requires some adjustments in image editing software to finalize the results.

Architecture

The model is based on the Stable Diffusion framework, specifically utilizing the StableDiffusionPipeline from Hugging Face's Diffusers library. It can be exported to formats such as ONNX, MPS, and FLAX/JAX for optimization.

Training

The model achieves its results by merging with other models trained on specific subjects, allowing for consistent character portrayal across multiple views. The process includes fine-tuning settings in the img2img pipeline and employing image editing tools for refinement.

Guide: Running Locally

To run the model locally, follow these steps:

1. Install Required Libraries:

   pip install diffusers transformers scipy torch
   

2. Load the Model:

   from diffusers import StableDiffusionPipeline
   import torch
   
   model_id = "Onodofthenorth/SD_PixelArt_SpriteSheet_Generator"
   pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
   pipe = pipe.to("cuda")
   

3. Generate an Image:

   prompt = "PixelartLSS"
   image = pipe(prompt).images[0]
   image.save("./pixel.png")
   

4. GPU Recommendation: Utilize cloud GPUs such as those available on AWS, Google Cloud, or Azure to enhance performance and reduce processing time.

License

The SD PixelArt SpriteSheet Generator is licensed under the Apache-2.0 License.