Add boner or bulge to existing photos or create an image with one from a text prompt with this Flux.2 Klein 9B LoRA. It was trained with image pairs as an edit model, so that is where it shines. Trigger word: underboner should do the trick but you can add more detail as to size, direction, etc.
Reducing the strength is recommended when using with other LoRA, and if you want a less dramatic effect.
Please share your creations here (click Add post under the corresponding model used), and check out my coachbate.ai site for content featuring all my LoRAs in action.
Flux.2-Klein 9B
Trained with image pairs to help add bulges to existing images or create new images from a text prompt. Version 3 is built on version 2, and it added even more target and control pairs, focusing on realism. Version 2 added some loose fitting clothing. This model will shift the color sometimes if you don't prompt it to preserve exact material and exact color.
Version 1 can be found here
Description
Trained for 8000 steps total, added some loose fitting clothes to the training image pairs and more angles
FAQ
Comments (4)
The world needs (I do) a flaccid penis print lora. Most loras transform bulge in a blob of cloth, specially the good ones. Please train one for Flux1.d so much good lora work on Flux1d to transition to flux2 (and flux2 is really ram hungry!).
get me a dataset and I'll build it. This was trained on target and control image pairs so works best with image models. Try the boogu image/edit or krea 2 versions. Krea 2 isn't image edit model but i used the control images as a regulation dataset when training.
@coachbate ok ok, if a dataset means a set of picture to work with, that can be arranged. But I dont know how to tag them or crop them in batch.
@kcouuock i've got tools to help automate that part. 20-60 images or 50-100 5 second videos (5-10 total minutes of video) for ltx wan loras can usually get decent results