an1x3r/anima-artist-mixer

This is an add-on for ComfyUI, a popular visual workflow tool used to generate AI images. It targets a specific AI image model called Anima and solves a particular problem: when you ask Anima to blend the styles of two or more artists, the results tend to look muddy rather than like a real mixture. The root cause is how Anima processes text descriptions internally, causing artist names listed together in a single prompt to interfere with one another. This plugin works around that by encoding each artist's name separately through the model's text processing pipeline, then combining the resulting style signals at a deeper point inside the model called the cross-attention layer. By injecting the blended style there instead of at the text input stage, the interference is avoided. The practical outcome is that style-similar artists mix in a more predictable and controlled way, though the README is honest that very different artistic styles may still end up as a compromise blend rather than a clean mix. Using it is fairly straightforward once installed. You place your artist names, separated by commas or line breaks, in one text box, and your main image description in another. A dedicated node called AnimaArtistPack handles all the splitting and encoding automatically. You then wire its output into the rest of your ComfyUI workflow in place of the standard text connection. Installation requires only cloning the repository into ComfyUI's custom nodes folder and restarting the application. No extra libraries are needed. The plugin works only with the Anima model and will not function with other image generators. Generation speed slows as you add more artists, roughly 1.4 times longer for four artists and 1.7 times for eight compared to a single artist. The README recommends limiting which layers or generation steps the mixing applies to as a way to recover speed without much quality loss.