original content based on model definition by @Defozo. this model will hopefully allow users to experience the fp16 model content on lower hardware spec'd computers.This is a merged model of a base model flux1-dev fp16 and NSFW MASTER FLUX LoRA : <a target="_blank" rel="ugc" href="https://civarchive.com/models/667086?modelVersionId=746602">https://civarchive.com/models/667086?modelVersionId=746602</a>Version 1.2 has added PussyDiffusion - Flux Lora merged in to improve vagina detail:<a target="_blank" rel="ugc" href="https://civarchive.com/models/983498/pussydiffusion-flux">https://civarchive.com/models/983498/pussydiffusion-flux</a>Version 1.2 hyper 8step models have the content of version 1.2 plus ByteDance Hyper-FLUX Acceleration LoRA has been merged in allowing images to be generated in 8 (sometimes 9 or 10) steps:<a target="_blank" rel="ugc" href="https://civarchive.com/models/691446?modelVersionId=774008">https://civarchive.com/models/691446?modelVersionId=774008</a>NOTE: When ByteDance Hyper-FLUX Acceleration LoRA is merged into a model and the model is converted into fp8 precision, the conversion process screws-up the hyper 8step acceleration blocks from the lora. This causes the model to behave like a standard (non-hyper) fp8 model. The good news is it can still be included in the image prompt and still get the benefit of hyper 8step with fp8 model.If you like this content, try a version of my other model series:<a target="_blank" rel="ugc" href="https://civarchive.com/models/967270?modelVersionId=1171422">https://civarchive.com/models/967270?modelVersionId=1171422</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/667086?modelVersionId=746602￼￼To"> To</a> convert and merge, I used scripts from: <a target="_blank" rel="ugc" href="https://github.com/kohya-ss/sd-scripts/">https://github.com/kohya-ss/sd-scripts/</a>To convert to gguf images, I used scripts from:<a target="_blank" rel="ugc" href="https://github.com/Zuntan03/EasyForge/">https://github.com/Zuntan03/EasyForge/</a><a target="_blank" rel="ugc" href="https://github.com/kohya-ss/sd-scripts/￼￼Using">Using</a> <code>--ratios 0.8</code> in the merging command for version 1.0 and <code>--ratios 0.65</code> for version 1.1requires ae, clip_l and (t5xxl_fp8_e4m3fn or t5xxl_fp16) vae/text encoderscreates fairly good nsfw content. no key words. tested with base flux image generation settings.NOTE: if you're not interested in clothing, use the version 1.0 model. it does a much better job with female anatomy. version 1.1 sacrifices quality to work better with clothed women.

NSFW MASTER FLUX fp16 (lora merged with flux1-dev fp16 saved in multiple precisions and formats)

use t5xxl_fp8_e4m3fn text encoder to reduce vram usage. content of version 1.1 merged with pussydiffusion to add more realistic vagina.notes from lora owner:Trigger words (that sometimes Flux does not follow):<ul><li>bush pussy - Shaved pussy/vagina with hair on top / landing strip</li><li>butterfly pussy - Pussy/vagina with big lips</li><li>hairy pussy - Hairy pussy/vagina, simple as that</li><li>shaved pussy - Shaved pussy/vagina, in the style of cameltoe (innie)</li><li>asian pussy - Dark lips pussy/vagina on a white skin woman</li><li>pink pussy - Pink delicate pussy/vagina</li></ul>Adding "showing her ass from the back" or something similar to show a rear view may work, as I trained rear view too for every trigger above.You can combine more than one. Inpainting them to improve the result is a good option.

Files with SHA256: 1a849dff048bfca7f6c9705999ec63047145f831da23f3f9e33c52663ba73092