Everyone's favourite movie short king... Now in Chroma!
Massive shout-out to PepitoPalotes for their feedback on the last version. The new version should be much easier to prompt. It's still by no means perfect, but hopefully now he can hold a beer bottle without looking like a dog lol.
This is trained on the same dataset as the MCU Rocket Raccoon illustrious model.
Also my bad for the uninspiring sample pictures. Caption-based prompting is still strange to me.
Other than that I'm counting the days until a good Chroma finetune comes out so we can see its full potential.
Description
Reworked training captions for better prompt adherence. It might not be immediately apparent in the example pics, but using the lora is much easier now.