
thanks, i had the actual useful post on the other thread bookmarked but it was still a pain to go through, hopefully this one will fix that
here is my contribution, nothing new but it gives me consistent results for blowjobs, this specific prompt was for already naked/topless base pictures but even without any change in the prompt it should also work on full dressed pictures, but just in case remember to always change the prompt depending on your starting pic
Template:
https://tensor.art/template/891989052872481976
Lora: high - 900241638184486434 / low - 900241602751013371
Lora weight: 1 on both
Prompt:
the girl in the frame remains stationary in place. A cinematic video of a beautiful naked woman, she puts her hands down on her sides exposing her naked breasts. A naked man approaches the woman from the right side. his penis is large and erect. Woman looks up at him with shocked expression. He grabs her by the head with one hand and pulls it closer, then he thrusts his penis into her mouth in one clean move. his penis goes inside her mouth smoothly. He repeatedly thrusts back and forth with his hips, he is making wide movements, penetrating her mouth down to the base and smashing his hips against her nose. When her face smashes the man's groin the extended tongue of the woman wraps under the man's testicles. Woman's breasts jiggle with every movement. The video ends with the man pulling the woman's head against his hips, pushing his penis all the way inside the woman's mouth. Fast and intense motion. realistic jiggle physics on her body. The girl keep looking at the viewer as she keeps her eye's open. eye contact. Realistic animation. [Keep the womans face accurate to the origional image,distinctive facial features remain intact] [prevent womans face from distorting] [keep lighting consistent to the original image]
the lora also work for POV blowjobs but i didnt try generating it without a dick already in the picture so you would have to try it on your side
Results:
you can remove the part where the man enters the scene and just leave the rest and it should work if you already have it on the picture
If the results dont really match the prompt you can try changing the seed and try again, at least for me that fixed it when due to the picture used the actions in the video didnt match the prompt description
