Trained on 195 images of Caracal (no repeats) from Kemono Friends on Anything v3.0 Pruned FP32 (however it works on every model)
This time I added the tag "floppa" to the training so that it associates "floppa" with many features of her, and fixes the association with "caracal" and "kemono (in parenthesis!)" messing up the generation
Of course, caracal (kemono friends) still works
Vtuber art is filtered from the dataset, that will be its own model
Stats:
195 512x images (no repeats) for 20 epochs
Network dimension: 32
Base model: Anything-V3.0-pruned-fp32.ckpt
Batch size of 3 on RTX 3060
Generation tags: floppa, 1girl, ((solo)), animal ears, caracal ears, blue eyes, caracal tail, orange bowtie, white sleeveless shirt, orange elbow gloves, orange hair, orange high-waist skirt, long hair, belt, bare shoulders, orange thighhighs, hair between eyes, bangs, sidelocks, orange neckwear, large breasts, long sleeves, white belt, center frills, breast pocket
Sample images are 512x512 latent upscaled to 1024x1024 and generated on Anything v3.0 Pruned FP32