Z-image turbo V1

does great at the breasts. the crotch area tho feels heavily dependent on the position/pose the subject is in to get a good result. ultimately a great start. looking forward to see what the update brings.

Is there a version compatible with DiffusionBee?

Thanks, man! I loved it.

I'm getting pretty good results on average, although it has issues (might be the model) with some positions - also at the recommended strength all of the faces start looking very similar, same flushed cheeks, same lip color. It starts looking more painterly/pony than photographic

can i use my own LoRA on top of that?

where i can get workflow for it?

With what checkpoint is this best used? Flex.1-Dev or one of the NSFW-Flux-Checkpoints?

what is the Trigger word for this LORA ??

amazing

This is actually rather good for both Asians and different levels of clothes... and that is rather hard to come by

Also prompt adherence has been okay

eh. not too bad not great also. it changes the lora face

can I use it for commercial purposes?

works for flux.1 schnell, dont know why but it does

This works in interesting ways on almost any prompt with a low weighting. I use it alone and quite often. Something about randomizing echoes. In a way, it begins to determine a personal style.

how to download this for novita ai

its a great model, but for some reason i keep generating extra limbs and the body is never perfect. what settings are best besides setting at 0.8? steps? etc? thanks in advance

it works with Schnell, apparently

how can i add this LOra to my tensor?

What prompt do I need to use?

I use <lora:NSFW_master:0.8>. But then what?

why does this Lora change my girl's face quite significantly while using the recommended 0.8 and really all weights. Do you have any suggestions?

Hi there, for some reason i'm not able to download your lora. Looks like link is broken. Could be this be fixed? thanks

Works great for me. Faces a Bit blurry but ITS ok

Can anyone use it comfortably, help me out?

good

if only I can get it to work, I didn't understand anything about flux this one I have the error: Cannot execute because a node is missing the class_type property.: Node ID '#13' It's the "string leteral" node that causes problems when I try to load a worklow from an image. What's that?

希望能有gguf版本的

i'd like to use it in wavespeed.ai, need licence?

Hi, I'm learning more about LoRA creation, and I noticed you had roughly 20k steps, which seems like a lot, even for a rank 128 LoRA. Are you able to share the thought process behind this? How large was your training set? Tags per photo?

Looking great, need to try this out!

bro, your model isn't working, every time it's generating blank gray solid color image. i think you should take a look and please update the model soon. love yo

dosnt work please help: Cannot execute because a node is missing the class_type property.: Node ID '#13'

What are the trigger words?

Hello, I am a beginner and I can't figure out how to use it. Each time I end up with models overlapping one another! And my models are dressed every time even if I put naked in the prompt. Can you give me an example of a prompt that works? Thank you.

Anyone got a workflow for this? I can't figure it out

Nice LORA, but it tends to overdo the lower ribs.

I dont get anything out of it that doesn´t look like a plastic cartoon. Using dev.fp8. with standard settings.

This is the best LORA I've been able to find for that particular goal but - it changes faces - by a lot. I've tested with different FLUX tunes. I need to run it at 0.6 to do not mess up the nipples/vaginas etc., lower values mess them up - but it changes faces. When I lower it to 0.5, I start seeing errors on nipples, when I reduce it to 0.4, they also appear on crotch - but faces get fixed.

Because of that, I need to turn it on/off for detailers - I usually run it only with a body detailer and then - run another detailer just for faces on top of that. Some other LoRAs and some mixes with it inside do not change the faces while preserving details but they respond to prompts substantially worse. For instance, I am able to generate saggy tits in both small and big sizes with this, I am not able to do it with any other NSFW LoRA with Flux - but those damn faces...

Is it fixable somehow? Cheers!

Flux LoRA Fine-Tune Causing Weird Hands/Legs — How Do I Fix Anatomy Issues?

I used the Flux Dev LoRA Trainer on Replicate as my base model to fine-tune an image generation model using my own dataset. My goal was to generate both SFW and NSFW images. These are the training parameters I applied:

30 high-resolution images (mixed SFW + NSFW)

input_images: open("C:/actress verify images minimal captions v2.zip", "rb")

steps: 2500

learning_rate: 0.0004

trigger_word: “Jahnvrix”

lora_rank: 16

caption_dropout_rate: 0.05

resolution: 1024

optimizer: adamw8bit

batch_size: 1

autocaption: False (I used my own captions)

The fine-tuning completes successfully, and normal prompts generate good results. However, when prompts involve legs, hands, or similar anatomy, the output contains distorted or incorrect anatomy (weird hands, broken legs, incorrect fingers, etc.).

I'm looking for guidance on how to fix this issue.
Is there any solution that can help improve anatomical accuracy during or after fine-tuning?

s there a step-by-step instruction on how to use this? I'd be grateful. I'm a newbie.

whats the trigger word?

Would this lora work with img to image?

I can't get this to work on Fooocus. I put the file in the LORA folder and picked it in advanced settings with a weight of 0.8 as recommended. The images aren't coming out as NSFW. I have the default in the other settings, for instance the model etc. Please help.

Unfortunately this Lora has the exact same sshs_legacy_hash as

Hyper Futanari - Big Futa Dick and Balls
https://civitai.com/models/672165/hyper-futanari-big-futa-dick-and-balls

(e3b0c442)

This can cause problems if both Loras are available

For the Z Image Turbo version... does the lora contain concept for sex acts? It doesn't seem to be able to generate the penis correctly either. Though I admit it could be an issue with Z-Image having trouble with more than 1 loras.