Add boner to existing photos or create an image with one from a text prompt with one of these image LoRAs (Qwen Image Edit 2511 or Flux.2-Klein 9B). You can then animate someone playing with that bulge with my LTX-2 or WAN 2.2 LoRAs. See details below for each LoRA modelFor all models, reducing the strength is recommended when using with other LoRA, and if you want a less dramatic effect, lower the strength.Please share your creations here (click Add post under the corresponding model used), and check out my <a target="_blank" rel="ugc" href="http://coachbate.ai">coachbate.ai</a> site for content featuring all my LoRAs in action.Ideogram 4Trained with same dataset but not with the control (non-bulge) set of images, captioned with bbox around underboner in training. You can create an element and the bbox should honor the bulge size, you can use 'underboner' or more descriptive if you want to include the direction, it is great at diagonal and horizontal and vertical not as great. If you want more of a bulge coming at you, make the bbox relatively small and it squeezes it in there usually. The underlying model knows all sorts of characters, real and imaginary so have fun. Unless you describe your faces, this v1 of the Ideogram 4 lora will use faces from my dataset, which I'm in the process of cropping and captioning out better. Flux.2-Klein 9BTrained with image pairs to help add bulges to existing images or create new images from a text prompt. Version 2 added more target and control pairs, including some loose fitting clothing also. This model will shift the color sometimes if you don't prompt it to preserve exact material and exact color if you are doing an edit. You can see the color shift with the navy blue scrubs showcase example.Qwen Image Edit 2511Trained with 30 target and control images, this can add bulges to images using the Qwen Image Edit 2511 model. I'm getting very high success rate especially for one person. No mask needed, just simple prompt (see showcase samples) Trigger word: underbonerLTX-2.3This is an i2v LoRA for LTX-2.3 so need to have a starting image with the bulge you want them to rub or show off. Use my Qwen Image Edit 2511 or Flux.2 Klein 9B LoRA to give any image an underboner, then animate it with this one. Replaced with a 10k step trained LoRA 4/23/26.WAN 2.2i2v model so you need a bulge, ideally in the center of the image for best results. It is good for foreplay, or teasing scenes you can run on long frame counts since doesn't matter if it loops. It helps fill time for longer videos while teasing the viewer until they whip it out. I did notice some issues where they will start rubbing on their belly or somewhere other than their junk, so it may take several iterations to get what you want.

UnderBoner by CoachBate (Bulge)

underboner v1.1 trained for 5000 steps for Klein 9B model with 30 target and control images

Files with SHA256: 372c2370e4ca6d48c45c98373d4640e88f21ba53e44a88b7396e613a75c4c234