Trained on wan2.1 i2v 720p model, using 24 clips ~3 seconds each of LSD dance video.
EXAMPLE PROMPT :
a [subject] doing L5D_move , [describe the subject clothing and background]
a [subject] doing L5D_move by pulling down her [bottom wear] revealing her bottom belly
you may lower the weight if you found too much movement (weight recommended 0.8)
i recommend using this AI Tools to run : WAN 2.1 IMAGE TO VIDEO with LoRA
Changelog :
V0.2
Continued training with more simple caption
should be more stable
V0.1
initial training