Z-Image Base
Hello! This time I decided to create a new version of my LoRA for Z-image.
Large uncut penises. Unfortunately, I couldn’t find a sufficiently detailed and high-quality LoRA for the Z-image base model, so I decided to train my own. It was a meticulous process. I gathered around 4,000 images for training and rented a server for it. And here is the result.
I’m planning to create a similar LoRA for Flux 2 Klein, as well as other LoRAs focused on male anatomy and gay sex.
If you have any ideas or would like me to train a LoRA for you, I can do it for a reasonable fee - message me on Telegram @lolkagyr
There’s also a donation link for those who would like to support the project. It will help cover server rental costs and the creation of new LoRAs.
https://donatello.to/pavlo_kliuchanskyi
FLUX
A LoRA for uncut penises, trained on a dataset of 500+ images with different angles and close-ups.
The goal was to teach FLUX to generate massive uncut variants with realistic texture and anatomical detail.
The results so far are stable, although there are a few common FLUX-related glitches (for example, the penis sometimes appears in the wrong place or separately in the hand). I plan to continue training and improving quality in the future.
Works best at LoRA weight 0.7–1.0.
I train on a GPU with limited power, so the process takes a lot of time.
If you like my work and want to support further improvements, I’d appreciate a subscription on Patreon or a donation. All support goes towards upgrading my GPU to create even better models.
Description
FAQ
Comments (11)
use Peter Norths
hi, I love this lora so much. would you be interested in training it in chroma or sharing your dataset by any chance? thank you <3
Wonderful! :D I love this!
How can your aesthetics be so perverted? It's all about exaggerated, perverted, anime-like, and anti-realistic psychological aesthetics. C-site is practically a pervert club.
LMFAOOO GET A FUCKING LIFE OUT OF WEIRDO DWEEB
Tell me you have micro-penis without telling me
@davedrewhull898 That’s funny coming from someone who clearly doesn’t know how to size their checkpoints properly. Your LoRA and base model shapes are completely mismatched, buddy.
Talk about projection—no wonder you’re so hyper-focused on things being 'too small to fit the tensor'. Go back to school, learn what a dimensions matrix is, and stop crying to developers because you can't read a basic stack trace
@ajeasternellis Imagine getting a standard tensor mismatch error on Node 57 and your first instinct is to project your own physical insecurities onto the dev instead of checking your LoRA compatibilities.
The console literally told you The new shape must be larger than the original tensor. Your IQ is clearly the only thing lacking dimensions here. Fix your workflow, clear your VRAM, and stop self-reporting on GitHub
@davedrewhull898 You're spending way too much time thinking about my dimensions when you should be worrying about your own Tensor shapes. A standard Flux LoRA mismatch has you completely broken. Skill issue. Fix your config or cry elsewhere.
@meishaonan What's your problem? I never wrote anything to you, dude. Were you so offended by people reacting to your comment that you came back months later just to keep talking shit? Maybe you need some help?
People who try to use overly complex words to belittle others and make themselves look smarter genuinely confuse me