Want to support my work or help fund the training of this dataset on other models? Join the Patreon in my profile, and if you do - thank you!

Krea 2 V1.3D:

Alright, alright. I figured out a way to greatly restore texture back. It wasn't more training or different settings, and that's all I'm going to say about it. I deserve to have some trade secrets, no? I made it version 1.3D because the D stands for detail. It was actually try F, as in I was totally...frustrated...with how many lengthy tries it took.

Anyways, I had versions that took it further but it tended to have more broken anatomy. So, the same rules still apply if you want something to look like a photo:

Be sure to mention that it's a photo/photograph in the prompt. Do NOT use "photorealistic" or any other term that doesn't actually mean a photo. Krea 2 was trained on a ton of artwork so using any of those types of terms will yeet it back to bad texture.

Some rare prompts might still want to go over to the anime-side, but I still have the photo slider directly on SNOFS Krea. Simply set it to somewhere around .5-2 on the strength and you're probably golden. This was trained on SNOFS 1.2 so I'll be putting out an updated version at some point:
https://civarchive.com/models/2823820/photodetail-slider-for-snofs-krea

The same goes for some specific terms. When in doubt, use the words listed below instead of their synonyms.

My two-stage sampler is still handy for multiple reasons:
https://github.com/Auryg/Krea-2-Two-Stage-Sampler

The idea for the two-stage sampler that you can generate without the turbo lora for the first bit, which helps with variation and prompt adherence, and then bump to a second stage with the turbo lora to keep things fast. For the three-stage variation, you can then bump back to doing it without the turbo lora for the last bit if you want to use a negative prompt. You can also generate at a lower resolution to start and then bump it up.

Also, please, for the love of god try SNOFS by itself before you go adding a bunch of other general NSFW loras or models to it that screw up anatomy.

Ideogram:

Ideogram model has been updated, and is available here: https://civarchive.com/models/2781404/sex-nudes-other-fun-stuff-ideogram-snofs

General Information:

SNOFS was trained on natural language (or JSON, for Ideogram), not tags. It will work best if you use full sentences to describe what you want.

Not using ComfyUI/your inference software doesn't support lokr? I've put up a merged version here. You can also use the merged base model to train off of: https://civarchive.com/models/2416142/snofs-sex-nudes-and-other-fun-stuff-flux-2-klein-9b-base-and-distilled?modelVersionId=2985440

Here's a list of some of the terms that work well:

• anus

• blowjob

• boudoir

• condoms

• deepthroat

• braless

• cowgirl position

• cum

• cunnilingus (be specific and maybe put kissing in the negative prompt)

• deepthroat

• dildo

• doggystyle position

• fingering (anal and vaginal)

• hand in panties

• handjob

• hitachi magic wand

• implied blowjob

• ipcam / nightvision ipcam

• masturbating (might want to put penis in negative prompt, or specify what she's rubbing for women)

• massage

• missionary position

• naked, nude, etc.

• penis

• pregnant (and can specify trimester)

• prone position

• reverse cowgirl position

• sex

• sheer

• snapchat (and caption/text/etc)

• selfie (and mirror selfie)

• spooning position

• strap-on dildo

• tentacles

• licking testicles

• undressing

• vagina

• wet clothes

Depending on the version, the following might work:

• anal sex

• anilingus

But also keep in mind that it was trained on stuff like "her panties are pulled down to her thighs," not "panty pull."

These models are under the following license:

https://huggingface.co/Ashen3/SNOFS

Flux 2 Klein 9b V1.4:

Additional training. Some of the training was done using https://github.com/BuffaloBuffaloBuffaloBuffalo/ai-toolkit-perceptual , training against depth. Considering how much of SNOFS is two people intermingled with close skin colors, it seemed like a novel idea. It did seem to rapidly help with that sort of thing. On the downside, it seemed to create a bit of a texture issue on very close up images. I did some more training after to try to bring that back and was somewhat successful, but I think I'd need to increase the weight decay to really make that happen. Since everything else was in a good state I decided to release as-is. If you do have that texture issue, try adding "goosebumps" as a negative prompt.

Flux 2 Klein 9b V1.2:

More training - anal still doesn't work super reliably. Added images with terms like 'condom-wrapped penis,' 'boudoir' and 'anilingus' (again, doesn't work super great yet).

Flux 2 Klein 9b V1.1:

Additional training means far less body horror, even on the distilled version (but, you know, still some there). When using the distilled version of the model try playing around with more steps, adding a little cfg, etc.

Flux 2 Klein 9b V1:

Flux 2 Klein's awesome VAE means it picks up fine details incredibly well. While it still needs more training, I have some other stuff to train in the meantime so I thought it was worth it to push this out now as it can do some things incredibly well. Expect some body horror, especially if you use it with the distilled version of the model for text-to-image. I found that perhaps using more steps than 4 was helpful with the distilled version, but I also didn't try it much. Using this with the base model has far less anatomy issues. I expect them both to improve further with more training.

Right now, for text-to-image I recommend the base model. For editing, I recommend the distilled model. Note that SNOFS wasn't specifically trained on any image pairs for editing.

Training details (skip to the version 1.3 details below if you just want to know what this model can at least somewhat do right now):

I trained this as a factor 4 lokr using AI Toolkit this time. I used AI-Toolkit because when I started the training the other options had issues with their lycoris output and ComfyUI.

I think my starting learning rate was way too high at 1e-4 with an effective batch size of 4-6 or so. I quickly decreased it but it was perhaps still too high starting at 5e-5. I'm running a different training run at 1e-5 right now and it's still learning quite quickly. I might try to further train this at a very low LR and see what happens instead of starting fresh. Note: this is probably largely because of my large lokr size. I wanted to ensure I had "room" for all of the concepts but it can make things spicy.

I think the main issue people are coming into with training both this and Z-Image are what timesteps you train on. This was mostly trained on a high shift value of 3-5 as in inference Flux 2 Klein stays above the 800 timestep mark for most of the generation and maybe does 1 step out of 50 at below 200. I found I needed to test as I went and see where the generations went wrong and try to adjust on the fly.

Version 1.3:

Further training to further refine things. This might be the last version; I wasn't really making this for myself and I'm guessing the community wants me to make something for Z-Image. I'll at least try that out once the base model is out.

Note that the list is not exhaustive at all. It was trained on natural language (and that's how you should prompt!), so many concepts are in there.

Version 1.2:

Further training, expanded the dataset even more.

Also, I see a lot of people mixing this with other NSFW general loras. I'd recommend you try it by itself first.

Note: While you can use the lightning lora with this, keep in mind it won't lead to the best results. It's great for testing prompts, but it tends to mess with anatomy, smooth out texture, and lead to less variation on the same prompt.

Version 1:

This past weekend I was gone. I decided to let my 5090 chug along making a lokr for Qwen on ~5,000 hand fixed captions on sex, nudes, and other fun stuff of hand picked images with hand removed watermarks. I wasn't expecting it to get so good so quickly, so I did a few more night's worth of training. I'll do some additional training at some point here but it's already good enough to play around with.

It can do basic sex positions, blowjobs, cum, selfies, dildos, snapchat selfies with captions, etc. Female genitals are still a bit hit and miss, male genitals aren't bad. With it being a lokr and it being trained on so many images it's wildly flexible and can be used with perfect likeness of other loras.

Note that sometimes it'll do the wrong sex position even if you name it, and I'm unsure why as the captions have no errors. It will perhaps clear up a bit with more training.

I used Musubi Tuner and it was a heck of time getting it to train a lokr. I had to use another lycoris library for it (which is somewhere in the issues on the github page, IIRC), but it's possible the main one has Qwen support by now. Here are my training settings, though note that I reduced my LR over time and I also started with sigmoid timestep sampling. I was training at 640x640 and 1328x1328 buckets:

accelerate launch --num_cpu_threads_per_process 1 --mixed_precision bf16 src\musubi_tuner\qwen_image_train_network.py `

--dit Q:\AI\Models\DiffusionModels\qwen_image_bf16.safetensors `

--vae Q:\AI\Models\VAE\qwen_vae_for_training.safetensors `

--text_encoder Q:\AI\Models\CLIP\qwen_2.5_vl_7b.safetensors `

--dataset_config S:\AI\Musubi\datasetWoman.toml `

--sdpa --mixed_precision bf16 `

--gradient_accumulation_steps 4 `

--timestep_sampling qinglong_qwen `

--optimizer_type adamw8bit `

--learning_rate 3e-4 --lr_scheduler linear --lr_scheduler_min_lr_ratio=1e-5 --lr_warmup_steps 150 `

--blocks_to_swap 25 `

--gradient_checkpointing --gradient_checkpointing_cpu_offload --max_data_loader_n_workers 2 --persistent_data_loader_workers `

--network_module lycoris.kohya `

--network_args "algo=lokr" "factor=10" "bypass_mode=False" "use_fnmatch=True" "target_module=Linear" `

"target_name=unet.transformer_blocks.*.attn.to_q" `

"target_name=unet.transformer_blocks.*.attn.to_k" `

"target_name=unet.transformer_blocks.*.attn.to_v" `

"target_name=unet.transformer_blocks.*.attn.to_out.0" `

"target_name=unet.transformer_blocks.*.attn.add_q_proj" `

"target_name=unet.transformer_blocks.*.attn.add_k_proj" `

"target_name=unet.transformer_blocks.*.attn.add_v_proj" `

"target_name=unet.transformer_blocks.*.attn.to_add_out" `

"target_name=unet.transformer_blocks.*.img_mlp.net.0.proj" `

"target_name=unet.transformer_blocks.*.img_mlp.net.2" `

--network_dim 1000000000 `

--save_every_n_steps 250 --max_train_epochs 10--logging_dir=logs `

--output_dir Q:/AI/Models/Trained/Loras/Musubi/QwenWoman --output_name WomanGirls