MoYou AI QWen is a model based on the fine-tuned integration of Qwen image. The model adopts distributed multi-stage training, integration, and differentiation to achieve pollution-free coexistence of multi-style, multi-category, and multi-concept, and uses Qwen2.5vl for annotation, fully aligned with clip. Additionally, it has strong compatibility with the native lora of Qwen image.

GGUF version available at: MYHuman MoYou AI QWen GGUF Q4/6/8

Model Features

Superior portrait generation capability: It can freely respond to male, female, young, old, single, and multiple individuals. It also allows free adjustment of strong depth of field, weak depth of field, or no depth of field effects.

Wide applicability: Covers multiple textures (film, studio, web images, etc.), multiple styles (modern, ancient, fantasy, cosplay, etc.), various special compositions (grid, breaking out of the frame, etc.), and supports batch stable output for both Chinese and English posters.

Easy to use: Supports descriptions in Chinese and reverse generation via Qwen. Excellent responsiveness.

Local usage suggestion: Since Qwen image requires significant VRAM locally (about 24g), online usage is recommended. In ComfyUI, select "Browse Templates" from the dropdown menu and find the Qwen image template workflow for use.

Using long prompts in Chinese is recommended for better results.

Sampler: euler, res_multistep

Scheduler: simple, sgm_uniform

Steps: 20, 30, 50

CFG: 3.5, 4

Model sampling algorithm AuraFlow offset: 3.1

Recommended resolutions: 1152×1536, 992×1776, 928×1984, 1536×2048 (both landscape and portrait orientations)

Reverse prompt suggestion: Use the Ollama node with the Qwen2.5vl model for reverse generation.

If you find it helpful, please share more images for support. Thank you~

Goose Groups:

MoYou Alchemy Pavilion Taiqing Pavilion: 858495398

MoYou Alchemy Pavilion Shangqing Pavilion: 764481772

Also, please follow my Xiaohongshu and Douyin accounts: yy469370733