I was thinking the other day, what would be a LoRA effect that I have never seen before, and something that wouldn't just be ripped off from PIKA or something like that. So I came up with the AMGERY lora for the new LTXV 13B Video model. The concept here is to visualize someone getting really angry, in the old "looney tunes" or "Tom &amp; Jerry" -kind of way. I built a fully synthetic dataset of assets created by myself, and trained the model on these.The trigger word is AMGERY, and a simple prompt like "AMGERY the man in this image gets really angry" works well.

Files with SHA256: c750c626fc49dc17819a1b906f2dc437b9303191f02217144f20f56f91b9ac1d