MOSS-Image-Base

Quantised media packages of OpenMOSS-Team/MOSS-Image-Base.

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

MOSS-Image-Base: text-to-image model supporting natural-language prompts, tag sequences, and JSON layouts for controlling text placement and object arrangement. Uses a Qwen3.5 text encoder and the FLUX.2 small VAE decoder.

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	3.0
Resolution	1024 (total pixels)
Notes	Flow-matching Euler; shift 5; negative-branch layer skip at 22

Example prompts

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

masterpiece, best quality, 1woman, solo, candid moment, natural lighting.
A garden bench under a blooming cherry tree, petals falling, soft spring
light. A 33-year-old woman with an oval face and brown eyes, graceful
build, gentle smile. Floral dress and sun hat, holding a book, mid-read,
candid and relaxed.

Fashion photography (magazine cover style):

Fashion photography, magazine cover style: a 28-year-old woman with sharp
cheekbones, piercing green eyes, and a defined jawline; wearing a crimson
silk slip dress with a plunging neckline and a gold chain belt, layered
pearl necklace, strappy stilettos; studio rim lighting, bold serif
typography, high contrast, cinematic, 8K, sharp focus, Vogue editorial.

Scenery:

Scenery: a windswept coastal cliff at golden hour, crashing waves below,
dramatic cumulus clouds, golden light on wet rocks, seagulls in flight,
wildflowers in the foreground, photorealistic, wide angle, 8K.