This model is made for messy cumshot photos with full facial and body coverage. The LORA aims for cum coverage over all else.This is v3 of Cumziilla. This release doesn't seem to perform any better than v2, so you might want to try that out as well. It was trained on the full flux1-dev model. The model needs a large amount of prompting help to render cum properly. See example images for help with prompting - e.g. if you don't tell it cum is white, you're going to get lots of random colors. See example images for prompting advice.Changes<ul><li>Using Joy captions (highly descriptive) for training - don't notice much of a difference</li><li>Increased number of images 50%</li><li>Decreased steps to just 50 epochs vs 150 in v2. I have compared versions at different training durations and I don't notice much of a difference. The training loss doesn't change much, so I assume lower steps is the move here.</li><li>Trained on a new LORA config.</li></ul>NOTE: this LORA performs poorly against the flux1-dev-fp8 quantization of Flux. Don't be surprised if results are off. It was trained on and performs the best on flux1-dev. Not sure how to help this in terms of training. One user suggested using Heun with 8 steps for the fp8 quantized Flux.

Cumzilla Flux Edition

Much improved imho with 1024x1024 source images. These were upscaled from 512x512, I need to resize and crop the originals again to 1024x1024 for the next version. I also did this one at network rank 4, so the LORA itself is much smaller. I didn't set any trigger words, going to do that for the next version, for now see example images for how to prompt. Trained for many more (750) epochs than beta1.

Files with SHA256: 625836666d1b528ce5f6b5cda736cfbcb5a9c23e91f1422435f7dcc06e01c495