I adjusted it to make it easier to generate Asian women.
I will continue to upgrade this model in the future.
（Sales are prohibited. Please use for personal enjoyment）
Uploading to Tensor Art is prohibited.
NSFW is OK.
I hope you enjoy creating it.
If you enter "blonde hair" into the prompt, you're more likely to get a white woman.
V2.0
This model has more Asian female elements and the contrast has been lowered a bit.
It also has some NSFW elements.
V4.0
NSFW add, Contrast reduced.
Description
FAQ
Comments (12)
This model is really great and realistic. Can you make an FP8 version of it? My GPU has limited VRAM. I’d really appreciate it.
I converted one for v5 on hf
https://github.com/qskousen/ggufy
It's as simple as just drag & drop and supports fp8.
@Kerstal doesn't say support for Z-Image?
@jimzlf You're right, is not listed. However i tested it myself and i can guarantee it supports Z-image turbo & base. EDIT: my bad, it does. is labeled as Lumina2 (ZiT, ZiB)
You can use the following custom node in comfyui. for Z-image you only need 3 nodes "Load Checkpoint > ModelFP8ConverterNode > ModelSave". Conversion takes few seconds. The new file goes into /output/ folder.
https://github.com/Shiba-2-shiba/ComfyUI_DiffusionModel_fp8_converter
The most perfect realistic model ever. I've been waiting too long QAQ
Z-Image training based on mainly Asian women "I adjusted it to make it easier to generate Asian women." xD
I got a notification that you uploaded fp8 version of v6, but I don't see the download link on top. What happened?
Once the Early Access period for fp16 ends, I will upload fp8.
wow.... i tested v6 and... it's amazing. Beautiful images and a very good prompt implementation. I used er_sde/simple 8 Steps to get a bit more details and it works very good. Basically, I encountered no errors, all the images meet my requirements. I love this checkpoint. It's my favorite at the moment. Thank you for it :)
v6 is a beautiful sweet model. ty for share.