This model is intended to showcase the transformation process of a character into another.
Currently the intended usage is [main subject] is transforming into [target character]
The transformation tag is very useful and both are good when weighed to about 1.2-1.45 like this in ComfyUI: (transformation:1.3).
For best results use the LoRA a weight: 0.4 - 0.6, anything above or below that may be worse.
It was trained on a carefully curated and high quality dataset.
It's currently pretty good at understanding the concept although may get confused when using some LoRAs. I'm also working on a version for Anima if this one is well received.
Currently it was trained on an Illustrious fine-tune called NovaFurry 18.A but should work with other SDXL models.
I'm also working on a better version with way better semantic and contextual understanding and stages like stage 1 - 3 of the process.
Description
First version on Anima, quite good understanding of the process and anatomy
FAQ
Comments (2)
I don't understand. Some of your samples are exactly the same for Anima base and for Anima 2.9B. What is 2.9B, how that works, what it's for if it's the same result ? I don't get it x)
I can also regenerate the images with Anima 2.9B, not sure if this will be any different but it's supposed to be better