This lora was specially trained on a pair 32F natural breasts. I know that there are already many other breasts lora for Flux.1 D out there, but I wanted to try and train my own lora and I like the shape of this specific cup-size.I tried to turn down appearance changes as much as I could with my actual knowledge.I’m using this lora with a 0.50 strength in my Comfyui workflows when it is the only Lora. When I use two Loras I adjust the other lora to 0.70 strength and by the images with three Loras the 32F and Extra Oomph going down to 0.30 per Lora and the other becomes 0.60 strength.It works good together with the Lora Extra Oomph which you find here <a target="_blank" rel="ugc" href="https://civarchive.com/models/1330180/a-little-extra-oomph-huge-clothed-natural-breasts-for-flux">A Little Extra "Oomph" - Huge Clothed Natural Breasts for Flux - v1.0 | Flux LoRA | Civitai</a>. I generated the example pictures with the checkpoint Crimson Witch which you find here. <a target="_blank" rel="ugc" href="https://civarchive.com/models/1604909/crimson-witch-flux">Crimson Witch FLUX - v1.0 - NF4 | Flux Checkpoint | Civitai</a> Some of my other uploaded images were generated with this lora as well. The Dataset for this training consists of 26 images, where most of them where nude images of the breasts and some of the images contained different clothing’s. Because of that it should be possible to generate NFSW images, but I haven’t tested it yet. All feedback and tips are welcomed.Flux2. D Version This version is trained on the same dataset as the version for Flux.1 D but was trained for Flux.2 D.For comparison reasons I used the same prompts as for the Flux. 1 D version so you can compare them better. Alle images were generated with the Base Flux.2 D model in ComfyUI with a strength of 0.80. I also generated an anime style version of all pictures.

32F natural breasts Lora for Flux1. D and Flux2. D

This version is trained on the same dataset as the version for Flux.1 D but was trained for Flux.2 D.For comparison reasons I used the same prompts as for the Flux. 1 D version so you can compare them better. Alle images were generated with the Base Flux.2 D model in ComfyUI with a strength of 0.80. I also generated an anime style version of all pictures.

Files with SHA256: 4da974a6f8f9c47c813fe6bcd6db288c9efef05d8905244c0d86746087d68cbe