This is a blowjob template involving a character giving a deepthroat while groping the testicles with a close-up perspective focusing only on the face trained by the artist Cyberunique. Lora weight around 0.7 - 1.0 works best. Around 0.7 - 0,8 you'll have some flexibility with the lora, but around weight 0.9 - 1.0 you'll most likely have close-up shots. This mainly works best with human/humanoid characters, but furry/anthro characters work as well. You might need to lower this lora to 0.7 - 0.8 if you want to use furry characters though since having it at normal strength might bring some weird results.
Positive Prompts: CybBJ, 1boy, 1girl, sex, 69, 69 position, blowjob, oral, fellatio, deepthroat, all the way to the base, balls deep, ball fondling, grabbing balls, close-up, face focus, female focus, penis, balls, blushing, looking at viewer
Optional: cum on face, cum in mouth
Description
FAQ
Comments (9)
I like the general Lora, but one problem is it hard forces a style
I'd say try and change the checkpoint or make the weight .8 or .9. Not sure what else I can recommend otherwise.
Mirror POV version of this would be possible? There are so few POV's centered on balls.
You'd need to send a link for reference so I understand what you mean. And yeah I wish there were more loras focused on balls, It's really hard to generate anything relating to balls for whatever reason.
@FurkamiDood Yes, that's exactly what I was referring to, I'm not an expert but I've noticed that the POVs that ask for something related to the balls just doesn't work, it automatically becomes blowjob or something weird. that's why I was referring to the idea of the mirror POV, maybe a POV where the viewer sees the girl's head and her back while in a mirror in front of him the action on the balls is reflected would be a factible solution. Something similar to "POV mirror refection fellatio" but if the girl is the other way around, I don't know if it would work.
https://civitai.com/models/1779299/pov-mirror-refection-fellatio-illustrious
@servonat That would be tough as hell because I don't think there's much artists out there who've done that before aside from BJs. Your best bet is to probably combine loras together to get what you're trying to do.
@FurkamiDood yeah, I've already tried that (it didn't work out XD) and I don't know why I always forget that models need to be trained with previous images, and what they do is permute them with the instructions you give them, changing the character, the posture, and things like that. But if you ask for too much, it starts to get messy. Mixing models doesn't allow them to generate something so new, the results are always permutations, allowing you to add “add-ons” but not radically change a result. Out of curiosity, how many images are needed to train a model?
@servonat 8 is probably the absolute minimum you'll need if you want decent results. But the higher it is the better it'll be. Usually 40 - 60 images will net you great results.
@FurkamiDood Umm, that's a considerable number, I'll keep thinking about it, but in any case thanks for the info.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.