<h3 id="this-lora-is-an-early-proof-of-concept-for-a-full-nsfw-finetuned-checkpoint-that-will-be-done-in-the-future-for-flux.-yqxtddkm0">This LoRA is an early proof of concept for a Full NSFW Finetuned Checkpoint that will be done in the future for FLUX.</h3><h1 id="*want-the-full-version-only-need-1500-more-buzz-to-release-the-full-fine-tuning-+-final-lora-models!-b9insl9ie">*Want the Full Version?? Only Need 1,500 more buzz to release the full fine tuning + Final Lora Models!</h1>Thank you to those who have already donated buzz towards the project!*Any help is appreciated. As you know, these models take alot of time, resources and energy. I want to release the best possible model &amp; full fine tuned model as soon as buzz is raised.COMPARISIONS BELOW (WITHOUT LORA VS WITH LORA)<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/cbe2d7b5-ff6d-4b75-b874-0d8fac0648f5/width=525/cbe2d7b5-ff6d-4b75-b874-0d8fac0648f5.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/91cf1d59-41e8-451e-bd5d-b5d5b2c8eafd/width=525/91cf1d59-41e8-451e-bd5d-b5d5b2c8eafd.jpeg" /><h3 id="(very-early-version-test-model.-the-official-version-will-be-trained-on-3x-the-dataset-steps)-shfqe98kn">(very early version test model. The official version will be trained on 3x the dataset / Steps)</h3><ul><li>Fully Captioned Dataset (High Quality .PNG Images)</li><li>Trained for 9,060 Steps</li><li>Took around 2 days to fully train</li><li>Trained on Lewds / surreal-like, realistic NSFW images.</li><li>Resolution: 1024x1024</li><li>Lora Weight: 1.0 (Suggested)</li><li>Final Goal : Full Finetune on thousands of images</li></ul><h3 id="where-the-model-excels-:-ly5i6zhk5">Where the model excels :</h3><ul><li>Creating Lewds / Art (Similar to DALL-E / Cici Style)</li><li>Pair with other LoRAs</li><li>"Flash Photograph" Prompts</li><li>Follows prompts well</li><li>Multiple Aspect Ratios (Square, Landscape, Portrait)</li><li>Gets rid off "Flux Face" when generating females</li><li>"Facials", Body Shots, etc.</li></ul><ul><li>Issues:<ul><li>More images of genitalia can be added to the dataset</li><li>No images of Men in dataset</li><li>Hands are ok.</li><li>Larger dataset to give even more flexibility</li><li>Far away faces need some retuning</li></ul></li></ul><h3 id="what-are-the-next-steps:-arqidvtwy">What Are The Next Steps?:</h3><ul><li>Remove some watermarks out of the dataset, Release the Full LoRA Version (Trained with Positions / Acts + Everything)</li><li>Release the Anime / Cartoon / Animation Version</li><li>Release the Full Finetuned Version (Everything combined into one model)</li></ul><h3 id="any-buzz-will-go-toward-the-full-lora-full-finetune-training!-xaydldbly">Any buzz will go toward the Full LoRA / Full Finetune Training!</h3>If you like the lora and can donate Buzz, its much appreciated!Leave any feedback to help improve the full model.Hide

Lewd FLUX (Beta) | NSFW

VERSION INFO : This is an earlier epoch out of the initial training. Everything is still early and in beta.

l3wd-000037 (1).safetensors

v1.0 Epoch 37 (Beta)

v1.0 NEW (Beta)

v1.0 Newest (Beta)

v1.0 (Beta)

V1 (Best Current Version)

v1.0 Epoch 40 (Beta)

v1.0 Epoch 45 (Beta)

<h3 id="this-lora-took-weeks-of-dataset-prep-captioning-etc.-so-any-feedback-help-towards-the-model-is-appreciated).-9loc4alf9">This Lora took weeks of dataset prep, captioning, etc. so any feedback / help towards the model is appreciated).</h3><h3 id="(very-early-version-testing.-official-model-will-be-trained-on-an-even-larger-dataset-+-more-steps)-7cc59epfw">(very early version testing. Official Model will be trained on an even larger dataset + more steps)</h3><ul><li>Trained to 9,060 steps</li><li>Fully Captioned Dataset with trigger included (l3wddall3)</li><li>Designed to Create photorealistic Lewds / Art (Similar to DALL-E Style)</li><li>Recommended Stength / Weight : 1</li><li>Resolution Trained : 1024x1024</li></ul><h3 id="next-plans:-8r5eiwd5k">Next Plans:</h3><ul><li>Release the Official Lora Trained on 3x Dataset &amp; Steps</li><li>Release the Full Finetune Version Trained on (Will need alot of buzz to go towards training on powerful GPUs)</li></ul><ul><li>Pros : <ul><li>Follows prompts well, Fairly flexible.</li><li>Very good at "flash photograph" prompts</li><li>Very good at the "DALL-E" / "CiCi" style &amp; feel (Not stuck to the default FLUX faces)</li></ul>What Needs to be Improved?</li></ul><ul><li>May need more images of certain ethnic groups in dataset</li><li>Very good at "flash photograph" prompts</li><li>Very good at the "DALL-E" / "CiCi" style &amp; feel (Not stuck to the default FLUX faces)</li></ul><h3 id="looking-to-release-the-full-finetune-asap-(any-action-is-appreciated)!-48imr6nm0">Looking to Release the Full Finetune ASAP (Any Action is appreciated)!</h3>Thank you for using the beta, Let me know what can be improved for the final version.

Files with SHA256: 11cbba9d5ce6c70a9c7cbb819d1e24f6b58aaa5923b29f78811127545b92377c