Riaku Laboratory is a space for experimental builds. I will be uploading various tests and new model variants here that show interesting results but aren't quite ready for a main release. Your feedback and generations will really help me navigate and find the perfect balance!

Versions

v2.9b - An experimental attempt to improve the Anima-2.9B model by creator gazingstars321. For instructions on how to run this model in ComfyUI, please check their original page, or simply use Forge Neo.

v1 - A heavy build with a deep bake (4500 training steps in total). The main focus of this version is pushing for painterly atmosphere and complex lighting. The model can be temperamental and unstable with short prompts.