Update: Krea 2 version added.
Dat gap. You asked for it, so here it is. Use "thighgap" as the trigger word. Adding "skinny" helps, of course, but not necessary. Best around 0.9 strength. Works well enough most of the time.
For a discussion of realistic AI image/music/video generation, join us at our new Discord server: https://discord.gg/B6BSCbdAJX
Description
First release. Use at .9 strength w/trigger word "thighgap"
FAQ
Comments (5)
post your dataset and I'll train it as part of a qwen lora.
I think, maybe, possibly, you could use z-image to generate a large batch of images first, and then use them to train Qwen's LoRa.
@spriac No..... Do not train loras on synthetic, it sucks.
My dataset came from hand-picked images on imagefap. Lots there to choose from.
I'm probably going to use one of the SDXL-based LoRAs to make a dataset just like I did with my waist LoRA soon to do ZIT, Qwen, Wan 2.1, Klein 9B, and LTX-2
Also the waist LoRA helps but biases it towards the front of the body, so I might make a new version of those later
I'm going to try do it in a lower rank too, either 4 or 8
Also the gaps are going to be way bigger than this. It's going to be a straight line from the top of the inner thigh to the knee.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.