Qwen-Image-2.1

Quantised media packages of Qwen/Qwen-Image-2.1.

MECHA Forge

MECHA — Model Engineering & Contextual Hearth Assembly

These media packages have been forged and optimized in the MECHA Forge. Each one holds the Denoiser, Text Encoder, and VAE components packaged together for easier deployment.

Description

Qwen-Image 2.1: text-to-image model with strong prompt adherence, multi-language (EN/ZH), and text rendering. 50 steps, CFG 4.0.

Available quantizations

Quant	Denoiser	Text encoder	VAE
FP4	FP4	FP4	FP8
FP8	FP8	FP8	FP8
INT4	INT4	INT4	INT8
INT8	INT8	INT8	INT8

The VAE is shared between formats: FP4 reuses the FP8 file, INT4 reuses the INT8 file.

Recommended settings

From the source model card (adjust to taste):

Parameter	Value
Steps	50
CFG / Guidance scale	4.0 (true_cfg_scale)
Resolution	any aspect ratio
Notes	Flow-matching; seed 42 for reproducibility

Example prompts

Caught in the moment (candid, slice of life):

masterpiece, best quality, 1man, solo, candid moment, natural lighting.
A street cafe on a sunny afternoon, a bistro table under a parasol, potted
plants, warm golden light. A young man with short dark hair, lean build,
wearing a linen shirt and chinos, tan loafers, sitting with iced coffee,
relaxed smile, candid and at ease.

Fashion photography (magazine cover style):

Fashion photography, magazine cover style: a 32-year-old woman with high
cheekbones, almond eyes, and full lips; wearing an ivory silk blouse and wide-
leg trousers, gold hoops, pointed heels; soft window light, elegant typography,
8K, sharp focus.

Scenery:

Scenery: a serene mountain lake at sunrise, mist over still water, snow-capped
peaks, golden light reflecting, pine trees, wildflowers, photorealistic, 8K.