Chibizilla has returned - in XL and now also in DiT! I have created the training images (50 each in the standard XL-resolution as well as 2x upscaled) locally using the model "GhostbustersXL mix", so I would recommend that model with this LoRA, but it should work well with any Anime model.
Recommended tags I used for training:
1girl, ((chibi)), (solo), (cute face), (human only), godzilla-cosplay, godzilla-costume, godzilla-hat, scales, tail with scales, gloves, destroyed city background. (It's her natural habitat after all!)
I also put "Godzilla" in the negative prompts to get more consistency, which worked really well.
Additional potential prompts: atomic breath attack, roaring, charging stance, glowing_scales,. Disclaimer: atomic breath is still pretty inconsistent, I think it's because the AI is so used to showing off the face / mouth of characters, that it doesn't want to obscure it with a big energybeam coming from there. The success chance is way higher than in SD models, though!
Have fun!
