MOSS Image SFT | Quality & Turbo | Anikin Labs
Custom ComfyUI workflow for MOSS Image SFT, created by Anikin Labs.
This workflow includes two generation modes:
Quality Mode — for higher-quality results with more steps
Turbo Mode — for faster generation using Turbo LoRA
Features
Clean and easy-to-use layout
Separate Quality and Turbo generation paths
Shared resolution setup
Ready for prompt-based generation
Beginner-friendly structure
Includes guide notes for required files and setup
Best for
portraits
stylized images
anime and semi-realistic tests
prompt experimentation
fast comparisons between quality and speed
Notes
This is an author-assembled workflow, adapted and organized by Anikin Labs for удобного локального использования в ComfyUI.
Please make sure you install the required:
model
text encoder / clip components
VAE
Turbo LoRA (for Turbo mode)
custom nodes used in the workflow
If something is missing, check the guide note inside the workflow.
Description
Anikin Labs MOSS Image SFT Workflow — Initial Release v1.0
The first public release of my custom ComfyUI workflow for MOSS Image SFT.
Features:
Quality Mode — full-quality image generation with customizable sampling settings.
Turbo Mode — accelerated image generation using a Turbo LoRA.
Easy switching between Quality and Turbo modes.
Shared positive and negative prompts.
Support for optional additional LoRAs.
Organized and color-coded node layout.
Built-in setup guide with model download links and installation paths.
Separate image outputs for both generation modes.
Recommended settings:
Quality: 50 steps, CFG 3.0.
Turbo (4-step LoRA): 4 steps, CFG 1.0.
Requirements:
Updated ComfyUI installation with MOSS Image support.
MOSS Image SFT model, compatible text encoder and VAE.
Turbo LoRA for accelerated generation.
rgthree custom nodes for mode switching.
Created and tested by Anikin Labs.
This workflow uses third-party models and ComfyUI nodes. Their respective creators retain ownership of their work.