Sign In
Civitai turns 4. Pick up a hat.
homemodelsimagesvideos3D Modelshubsarticlescomicscruciblesupdatesshop

Enzino H3 Image Workflow

Updated: Nov 12, 2026

tool

Download

1 variant available

Archive Other

Enzino-H3-Image_v1.0.0.zip

1.1 MB

Verified:

Type
Workflows
Stats

306

Reviews

(13)

Published

Nov 12, 2026

Base Model

MiniMax H3

Hash
AutoV2C9D65D7AB8

The MiniMax H3 Community License Agreement’s grant excludes the European Union, the United Kingdom, the Republic of Korea and the United States of America.

MiniMax H3

02_workflow.png

⭐ Unlock EnzinoAI Pro for just $25/month on EnzinoAI or Ko-fi ! Exclusive models, LoRAs, custom nodes & workflows. BF16, FP8, INT8 + new releases included.

📢 Join our Discord Community

Enzino H3 Image Workflow

Five blocks, left to right, nothing to wire. The H3 image always comes first; the extras are there when you want them.

What’s inside

  • Text-to-image and photo edit. Write a prompt, or drop, paste or click a photo into the Prompt block and say what should change.

  • Qwen prompt enhancer. Turns a short idea into the full description H3 Image expects: framing, face, natural skin, clothes, setting, light. Three levels: Complete, Complete + reasoning, MiniMax original. “Use as my prompt” puts the result back in the box so you can fine-tune it.

  • MOSS-Image refiner (optional). A light second pass against the smooth look: skin texture and fine detail, same composition. Detail sets how much of the new texture is kept, and Keep the H3 colours leaves the light and colours untouched. You can add a LoRA if your model doesn’t have one built in; with a speed LoRA (DMD, turbo…) the block switches to CFG 1 for you.

  • Formats up to 20 MP. The menu shows the exact output size.

  • Up to 4 LoRAs on H3, each with its own strength.

  • Preview. Live while H3 renders, then the sharp H3 image while the refiner runs. Compare H3 and MOSS with the A/B bar, and open the full-size image in a new tab.

  • Save as PNG (the workflow is kept in the file), AVIF, WebP or JPEG.

  • Built-in help panel, English or French interface, works in classic mode and Nodes 2.0.

First run

  1. Copy the ComfyUI-Enzino-H3-Image folder into ComfyUI/custom_nodes/, then restart ComfyUI.

  2. Drag enzino_h3_image.json onto the canvas.

  3. Missing a model? ComfyUI offers a Download button, and each block lists its missing files with a link and the folder they go in. Press R to refresh.

Needs ComfyUI 0.45 or newer. No other custom nodes. The full guide with screenshots, file list and settings is in the zip.

Models

  • MiniMax H3 Image: model, text encoder and VAE (about 50 GB). Links are in the guide.

  • Optional: Qwen 3.5 9B for the enhancer, MOSS-Image for the refiner.

Tips

  • Keep 6 steps / simple: H3 Image is distilled on that grid.

  • Refiner: strength 0.15–0.25, Detail 40–60%.

  • At 16–20 MP, save as AVIF or WebP.

Models keep their own licences; check them before any commercial use.

Made by Enzino · enzinoai.com · ☕ ko-fi.com/enzinoxl