7 ms·

loras have been fixing that for years

• dragonwriter 2mo ago
  Other than for pure t2v usage you probably don't need a LoRA for much, I have seen evidence that it is knowledgeable enough that it can handle a fair amount of anatomy looking and behaving reasonably with just relevant cues in control (for the fl2va model) or reference (for the r2va model) images. And the r2va model can also use video input for motion reference.