My my you have been busy.

I really like your LORA. The quality is astonishing. If possible, please share your training guide. I would like to refer to your training settings, the resolution of the video training set, and the frame rate settings. I am currently learning the training of Dynamic LORA, but I can't find a way. Thank you.

Oh yeah keep doing that! Oh, I mean, that's incredible, you made six loras in 10 days, are you making loras for movies? I can totally make movies! If there's doggy lora, that will be great. Hope v2 version will come out soon.

Basic question: a LORA for 5B may be also used for 14B? and vice versa?

absolute hero. great lora,

For everyones information: This lora stabilizes the anatomy of characters quite a bit and can in effect be used as a general lora for T2I at quite low resolutions.
I recommend trying out 512x512 and 480x672 with a lora strength of 0.8-9. It speeds up things quite a bit to be able to lower the resolution.

For best results give a succinct prompt of what is happening; then what you want to happen. Works best (sometimes only) if the activity is heavily implied in the image alread.

NB: Details disappear more easily at lower resolutions, so add more steps (in the KSampler (Comfy-UI)): 30 steps is all I had patience for, but in general more steps seems to stabilize the generation and often help keep likeness (just like in SD i2i).

NBNB: Thank you, poster, for creating the range of loras you did already. Damn good work :D