About MOSS Portraits

MOSS Portraits is a free library of photorealistic AI-generated portrait images of East Asian women. Every image ships with the complete English prompt that produced it, the generation settings, and tips for modifying the prompt to create your own characters.

The Model

All portraits were generated with MOSS-Image, a 5-billion-parameter diffusion-transformer (DiT) image model, run entirely locally on a consumer RTX 4060 Ti GPU via ComfyUI. Images are produced at the native resolution of 768 × 1152 (native, unupscaled) — we label resolutions honestly and never upscale-and-claim-4K.

Why We Publish Prompts

An image without its prompt is only inspiration; a prompt is reusable knowledge. Publishing both makes the library useful for designers, writers and AI enthusiasts who need portrait reference they can actually reproduce and edit.

AI Content Disclosure

All persons shown on this site are AI-generated and do not exist. No real individual is depicted. Portraits are synthetic images produced by an AI model; any resemblance to real persons is coincidental.

Contact

Feedback, takedown requests or prompt contributions: chaoshanaxing@gmail.com