Trained on 198 (provided) no testicle having futanari. I haven't played around with it too much yet so it might still be crapola. Figured i'd upload it since the only other alternative sans balls had balls in the first image /facepalm
Alrighty now that I've sufficiently generated an ample amount of images leveraging it. A few pointers
Like my other models this too has a bit of a heavy handed squatting position, so negative prompt that if u don't want it (sorry it's just the clearest position to catch everything /shrug)
Flaccid works amazingly well with it
Gape/Gaping will make it look like Elon brought a Boring Company drill thru her enchanted forest.
Weighting depends on ur base model as it's fighting against countless trained images of cocks and balls together.
Penis, pussy and futa /futanari are all u need in the positive, i usually drop a testicles in the negative prompt for good measure.
Cheers.
Description
FAQ
Comments (11)
Is this really that necessary? Pretty much all idea of Pony is that it can do this kind of concepts natively. For me it takes only a minor prompt engineering to achieve the same result... I'm not saying you've done the work for nothing or this LoRA is useless, I just really don't understand the intent behind making one (despite the fact that this concept is definitely from my book).
lol i made it primarily because i stumbled over this https://civitai.com/models/417977
Also getting any model to generate futas without balls takes quite a bit of prompt engineering, 9 times out of 10 you will end up with some abomination of flesh, if you find a secret string please do share.
@voidxedis Honestly, "futanari, vagina, vulva, penis" and "testicles" in the negative do the job pretty much on any Pony based checkpoint. Depending on the checkpoint, this exact prompt might not be enough, but the rest is just playing with weights.
@latex_waifu uh huh. Pics or it didn't happen.
@latex_waifu also, why my Lora out of the THOUSANDS on this site?
@voidxedis Would've happily shared the pictures if only there was a way to upload them privately without publishing to everybody. I don't publicly post such stuff on principle.
Your LoRA because it's very nice and would've been great if it was trained on SDXL and/or Animagine, where such concept actually near impossible to get. In this case it would be massively more useful because existing LoRAs aimed at producing the same result are really subpar in terms of the image quality.
@latex_waifu oh, well in that case, you could have just asked. I can run thru training on 1.5 and sdxl just as easy, i've been support since the first month they offered it so i have plenty of buzz for training......hold my beer...
@latex_waifu first batch of sdxl was a bust, going to refine the dataset further with some additional content. Please hold. No drinking my beer tho, i'm still coming back for that.
@voidxedis No worries bro, I'll keep it cold for ya! ;D
long story short yeah sometimes it helps a lot. Adding more keywords results in a less coherent image, it's sometimes better to use a lora to change the weights outright and keep the query small
Very useful for drawing futa with real models!Thanks!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.