Concept
An experimental small/micro penis LoRA for MiniMax H3 i trained a month or so ago and hadn't gotten around to releasing. Minimax isn't great at cocks to start with but pretty much every mix or cock lora generates giant erect circumcised cocks. Even the flaccid penis lora generates pretty large cocks. The original goal was just to create a better cock lora but trimmed the scope down to small cocks only as that was the major missing feature. Foreskin is supported but its super hit and miss, i've tried captioning it a whole bunch of ways and none seem to give the result i want but it can work sometimes.
This couples well with the softsuck lora hence the timing. The lora also couples pretty well with @diogod's excellent male anatomy lora (https://civarchive.com/models/2839513/male-anatomy-ass-penis-testicles-and-glans-h3). The second example video uses both loras together to get a better overall result.
Usage
Should work in both T2V and I2V as an anatomy anchor. It works much like a slider lora where the stronger you put the weight the smaller the cock. I'd recommend no lower than 0.6 and no higher than 1.2.
Prompt
So first thing to note is i somehow managed to misspell flaccid and the captions all contain flacid instead. so make sure you use flacid when describing the state of the penis. Some sample prompts to get you started, though Minimax really does need longer prompts than just these. Check the samples for good full prompts.
Size (small/micro)
a small/micro penis.State (flacid/erect)
a small flacid/erect penis.Foreskin(intact/circumcised)
a small intact/circumcised penis.Foreskin position (retracted / extended-covering / partially retracted)
a small intact penis with the foreskin retracted / extended / partially retracted.I do tend to find that it tends to do darker shafts by default. I think this is largely due to the lack of visible shaft in most of the input images. You can try and prompt the shaft color as being pale or dark but it has mixed results.
Training
The model was trained on 100 images at 2mp for about 10000 steps. Captioning was automated using Qwen3 VL 32B heretic. Training was run using diffusion pipe on 3x3090s. I used the diffusion pipe image only training adapter to avoid images slowing motion and it seems to have worked well.
Description
FAQ
Comments (28)
lets be honest that's not that small right ?
depends on which one but the first pic no its not really. Sadly porn has warped what people think is normal / regular sized. I will say there are some actual micro penises in the dataset. But you have to push the lora weight up to get that size.
@imb101 it was a joke if i could post a gif a man sweating i would of done lol
lol @crafted101 fair enough bro.. ! i clearly need to go outside and touch some grass ><
@imb101 Why though? You answered his question respectfully, which is a lot more than I can say for a lot of uploaders on here. Sides, winter is rolling in, won't be much grass to touch, soon..
@Lazman its almost summer here where i am
@Lazman Well unfortunately as someone who mostly makes non-mainstream content I get comments on forums from folks who somehow think my kinks/content are tied to some political agenda. When reality my kinks are just my kinks and have nothing to do with some imagined political viewpoint.
This in turn results in me being pretty guarded when responding to comments on the whole. Probably more so than i need to be. Hence my suggesting that i needed to maybe stop gooning and go outside :)
I will say i hate this weird alignment that's happened in society over the past decade or so that if you like kink X you must have political view Y. Anyway.. back to gooning outside <3
@crafted101 Philippines?
@imb101 Kind-based politics? That's new to me. Cept maga being a buncha self-casterated dickless loonatics. Lol.. I mean, I'm not the partisan type myself, but those people are a whole new brand of stupid, it's hard not to have fun insulting them. I am gettin sick to death of hearing 'trump' everywhere I go though.. People act like the one man is somehow responsible for half a century of near global economic decline..
@imb101 'Kink-based' not 'Kind-based'.
Civit went full retard and decided to block me from commenting cuz I commented too much(answered several responses in a row..). So, I wasn't even able to make a single letter correction.. SMH..
@imb101 Correction, this is my other account, and civit didn't properly log me into my other account when I told it to, so that just made me look like a moron, especially since, immediately after realizing that, civit continued refusing to let me post anymore, despite letting me post that one message..
This site has gone to absolute trash since they split it to red and blue.. Actually, some time before then, but it's buggered AF now..
Either way, apologies for the drama over a 1 letter error.. I just needed to vent.. Such a simple thing, corps gotta make into a half hour event..
@Lazman no think more middle earth
@crafted101 Hawaii? Or Mordor?
@shanezwicker697 the shire
Doesn't work for me at all. Tried 3 different models and all the results look like ground beef.
Edit: Also your samples don't contain workflows.
And yes, i have that "Save Image" thing installed that you need to import most workflows, because that's what most people save with apparently.
I don't tend to post my workflows as it means cleaning things up (i use a bunch of my own unpublished nodes) and I'm too lazy to do that.
The default base model with the 1.2 lightxv2 turbo lora even with a really short prompt should get you something half decent. if you're using mixes i cannot tell what they've added which might impact how this lora works. If you're actually looking for help feel free to message me with your workflow/prompts and i'll see if i can help you.
"Default base model"
You mean the one that's like.. what was it? 320 GB in size?
I am using the 2 "Pruned Int8" ones, fl2va and ref2va, and an Eros one, "ErosMax Beta5"
You could at least add your prompts somewhere. You just wrote some examples and then say "This but longer!". Well... i have no idea what you want me to add... And if my models are not messing things up, it should be the prompt, which is practically just what you wrote as examples.
@SDGFDGFHFd46346 open one of my samples on the civ website.. scroll down find the section that says ‘Prompt’ .. select the copy button and paste it into comfy…. Base model means the unmixed version of which runs on your hardware.. so based on what you wrote pruned int8 fl2va.
@imb101 If, hypothetically, this were trained on a model that was 10-20X the size, wold it even be guaranteed to work on the one that's so much smaller? I mean, that would imply working on a model that may have had the specific tweaked parameters removed a few times over..
Also, I would like to ask, does it work for uncut. I only see mutilated wankers in the samples, but then, I've got women turned off, cuz I don't want to see that (at least, not in the nude/sexualized context), so if there's any uncut in the samples containing women, I can't see those.
Thanks for the mention! I'll try it with my LoRA! looks great
Both at 0.7 seems to work well for me.
Thanks for the hard work, especially since this has been an underrepresented area in video gen up to this point. Would a REF2VA version be easily feasible by any chance? No worries at all if that's out of scope or too much of a headache.
i've not trained a ref2vid lora yet. i'll investigate and see how hard it is.
@imb101 I've only done a few tests with this so far as ref2va, but it appears to improve the "giant slab of ooze" issue that minimax has with penises, so this looks promising!
would also be interested in this
@everwake888728 @datlurkaa So ref2vid does work but its far from perfect. For best results i'd suggest you either add a ref of the penis if it isn't in the inputs already, or add diogods male anatomy lora at about 0.7 strength. the example i posted has the penis as a ref input.
@everwake888728 Ick.. Tbh, it's disgusting how underrepresented males are when it comes to image/video generation in contrast to females. I mean, to the extent that the models understand 'human' as 'female'.. Not literally, but that's pretty much how the model sees it..
It also don't help that so many people seem to be into this relatively new form of 'bara' which is basically just muscular men with boobs.. So, that further biases the models towards female anatomy.
PS: I got nothin against women, in general. Just against this societal bias and stereotyping of men to the extent of nearly dismissing their right to exist. It's a twisted thing to see once ya notice it..
Thank u!