SDDj supercharges Aseprite with offline AI animation

By vramkickedin | | 2 min read
Closeup of a translucent hand flipping over a stack of translucent paper.

SDDj is a local image generation and animation extension for Aseprite that combines Stable Diffusion with AnimateDiff. It runs entirely on your computer, generating images and animations directly within the popular pixel art editor without needing internet access.

The developer, known as FeelTheFonk, created this tool to bring fast, precise animation generation to Aseprite users. The project draws inspiration from older animation tools like EasyToon while adding modern AI capabilities. Everything runs offline with no cloud services or API keys required.

Core extension capabilities

  • Text-to-image, image-to-image, and inpainting generation modes.
  • ControlNet support for spatial conditioning including OpenPose, Canny, and Scribble.
  • AnimateDiff integration for smooth multi-frame animation consistency.
  • Audio reactivity that maps sound features to animation parameters.
  • Post-processing pipeline with background removal and pixel-art optimization.
  • Prompt scheduling for frame-indexed keyframe animations with transitions.

Artists and hobbyists who want AI-generated content directly in their pixel art workflow can benefit from this integration. Game developers working on retro-style projects may find the pixel-art pipeline particularly useful for rapid prototyping without switching between separate applications.

Development status and requirements

The developer notes the software is in early stages with more work planned. Currently, the tool only runs on Windows and requires PowerShell 7 along with Visual Studio 2022 C++ workloads.

'The software is in its early stages; there's still a lot of work to be done,'

the developer stated. Testing has focused on NVIDIA GPUs, with the developer achieving 512x512 generation in under a second on an RTX 4060 Mobility with 8GB VRAM.

The architecture uses a lightweight Lua frontend inside Aseprite that communicates via WebSockets to a Python backend handling all machine learning operations. Performance optimizations include Hyper-SD 8-step inference, DeepCache for feature reuse, and torch.compile for faster processing.

Get SDDj on GitHub.