MOSS-Image-SFT makeup-free LoRA

Note: All model files and custom nodes below must be downloaded via the client. If you do not have the client, Download >>>

mossqiansuyan is a facial makeup-free LoRA adapter designed for MOSS‑Image‑SFT. It focuses on generating lifelike portrait photos with a natural makeup-free look. It reduces cosmetic effects and preserves authentic, subtle skin texture for realistic portrait creation. It must be used together with the MOSS‑Image‑SFT base model. MOSS‑Image is an open-source text-to-image model released by OpenMOSS. It supports both Chinese and English prompts. Its standout feature is precise spatial control over objects and text in the frame, which works great for posters and graphic design. It has two versions: Base and SFT. The Base version is the raw base model, mainly for developers to modify and continue training. It performs poorly for regular users and follows instructions weakly. The SFT version is supervised fine-tuned from Base. It is ready for direct use by ordinary users, with better instruction understanding and more stable layout and in-image text rendering. SFT is recommended for daily creation. mossqiansuyan is a community LoRA dedicated to portraits for the SFT version. While keeping the original model’s layout control ability, it delivers natural makeup-free facial effects and produces realistic portraits with genuine skin texture.

MOSS-Image-SFT-8-steps-dmd >1 files

Name	Size	Target Dir
MOSS-Image-SFT-8-steps-dmd.safetensors	1.19 GB	loras

MOSS-Image-VAE >2 files

Name	Size	Target Dir
flux2-vae_small.safetensors	249.52 MB	vae
flux2-vae.safetensors	336.21 MB	vae

MOSS-Image-SFT makeup-free LoRA >10 files

Name	Size	Target Dir
mossqiansuyan_c1-st1000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st10000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st2000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st3000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st4000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st5000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st6000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st7000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st8000.safetensors	77.72 MB	loras
mossqiansuyan_c1-st9000.safetensors	77.72 MB	loras

MOSS-Image >4 files

Name	Size	Target Dir
moss_image_base_int8_convrot.safetensors	5.41 GB	diffusion_models
moss_image_sft_int8_convrot.safetensors	5.41 GB	diffusion_models
moss_image_base_bf16.safetensors	10.16 GB	diffusion_models
moss_image_sft_bf16.safetensors	10.16 GB	diffusion_models

MOSS-Image Text Encoders >3 files

Name	Size	Target Dir
qwen3.5_9b_w4a8_gptq.safetensors	7.34 GB	text_encoders
qwen3.5_9b_int8_convrot.safetensors	9.75 GB	text_encoders
qwen3.5_9b_bf16.safetensors	19.31 GB	text_encoders