这是通过东方女性写真摄影原图训练而成的模型，志在人物摄影方面得到更真实的质量
以1000张真实的摄影原图训练而成
想要获得BF16版的话，请移步到：https://www.runninghub.ai?inviteCode=rh-v1456上使用，如果觉得好用，再联系我，我所有的BF16版，都在上面
This model was trained on 1,000 original, unedited photos of Asian women, aiming to deliver more realistic and high-quality results in portrait photography.
If you'd like to use the BF16 version, please head over to: https://www.runninghub.ai?inviteCode=rh-v1456
Feel free to try it out first! If you find it useful, just reach out to me—I host all my BF16 versions on that platform.
Description
FAQ
Comments (7)
I tried it out today, it's perfect. I'd also like to suggest that the next version should have more realistic dissections.
解剖方面还暂不考虑，因为我没有这方面的数据集，又不想用AI生成的图做训练。当然有谁难提供这方面非AI生成的数据集的话，我可以去制作这方面的内容
就只有女性吗？男性没有做训练优化吗？会不会男生女相？
我没有做男性优化，但底模还在，所以不用担心男生女相
I really like that you released int4 today, but even though I followed all the settings exactly, I still encountered an impossible situation: insufficient video memory.
I'm not really familiar with video, I only work with images
@vkbgwyc520886330 you misunderstood the TS, he meant a VRAM.
@woaiyitiaofeng check your text encoder. It's an LLM and also consumes the VRAM. Try 8-bit or 4-bit TEs:
> 8-bit:
https://civitai.red/models/2731465/qwen3-vl-4b-abliterated-comfyui-krea-2-text-encoder-bf16-fp8
https://civitai.red/models/2728378/qwen3-vl-instruct-4b-heretic-7refusal-convrot-alternate-krea-2-text-encoder-for-comfyui
> 4-bit:
nvfp4 version:
https://huggingface.co/DreamFast/Qwen3-VL-4b-Heretic-ComfyUI/tree/main
int4 ConvRot (provides worse detailed pictures on my taste than nvfp4):
https://huggingface.co/Merserk/qwen3vl-4b-int4-convrot/tree/main
https://huggingface.co/Skydexter/Krea2-Turbo-int4-ConvRot/tree/main