Extreme Closeup (Blowjob) MinimaxH3
Trained with close up images at the POV mentioned in the LORA
Not intended for other purposes.
Using first frame only will add flexibility to the lora
Description
FAQ
Comments (14)
Nobody has ever tried the famous Nikita D. tongue flick. Would love to see that happen.
If training could be done without likeness possibly
I've been looking for a good tongue flick but I was kinda able to do it with this one. By the way @Felldude this thing is amazing. this whole line... where can I tip? buzz?
@A_fuzzy_larkin So rather then buzz I would say post some videos to the model - Currently even if civitai detects the use of a video lora in a workflow it does not seem to post to the creator page unless made onsite - unless you go to the lora page which is admittedly a pain
@Felldude bro! I gotchu!!!
The POV version is GREATTT too! Keep up the great work!
Thanks
While I'm at it... Do you think it's possible to create a LoRA: 'General Tongue Enhancer'? Something that captures the details of various licking movements and angles, licking lips, licking penis/pussy, etc.? Ideally, it would cover all these different types of movement in a single LoRA. It would be a huge help, since tongue anatomy and movement in Minimax H3 unfortunately don't work very well... Thanks a lot for your great work.
Maybe movement, the detail seems to suffer from the granulation
destroys sound and audio following, but it looks cool
yeah, sound on this is no good but can be totally fixed with a pass of audio only in mmh3 or better, ltx2.3/2.5 for speed. 2.3's voice model is particularly good, fwiw.
The visual quality on these loras are outstanding. They make BJ scene generation exciting again. Really well done. Hope for more positions/concepts (tongue was mentioned, yes, PLEASE fix mmh3 tongue action!)
Kjnodes has an LTX lora loader advanced that let you reduce the audio latents. I don't know if that's the case with minimax, but might be worth looking into
I'm probably asking for a lot here but how about one for cunnilingus? I know a lot of us would really appreciate it.
Either way, really great work here.
It should be possible, maybe not with text to video but with first frame or ref for sure