wowsuch ai
Live

speach1sdef178Model

MiniMax-H3-X2-Detail-v1

A modified MiniMax H3 VAE checkpoint with two uses: 2× video decoding and optional detail enhancement of an RGB reference image. The release includes a ComfyUI custom node and workflow, and requires a separate fast decode node for the 2× workflow.

What it does

Use it in ComfyUI to decode MiniMax H3 video latents at 2× resolution or to enhance spatial detail in a reference image before reference-to-video generation.

Availability
Working model; tested custom node, workflow, and research report released
Pricing model
Not established
Deployment
ComfyUI
Privacy
Not established
Licence
MiniMax H3 Community License Agreement
Capability
video generationimage generation
Modality
videoimage
Access
open weights
Deployment
on-premises
Pricing
unknown
Target user
creatorresearcher

Evidence

6 claims
  • The checkpoint supports 2× decoding of MiniMax H3 video latents. [1]
    The example workflow demonstrates both roles of the same model: it is loaded as the X2 video VAE and decoded through `MiniMax H3 VAE Decode (fast)`, while the same checkpoint is also selected inside the Detailed Upscale node for optional reference enhancement.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • Its optional reference-enhancement mode uses an existing RGB reference image. [1]
    Its architectural limitation remains important: the B32 path uses early encoder features from an existing RGB reference and should not be described as adding B32 detail to arbitrary generated H3 latents.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The release includes a ComfyUI custom node and an example workflow. [1]
    It contains a **2-in-1 MiniMax H3 X2 VAE**, an optional ComfyUI detail-enhancement node, an example workflow, and the full research article below.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The 2× decode workflow requires the MiniMax H3 VAE Decode (fast) node. [1]
    The 2X decode workflow requires [TripleHeadedMonkey/ComfyUI-MiniMaxH3_LatentUpscaler](https://github.com/TripleHeadedMonkey/ComfyUI-MiniMaxH3_LatentUpscaler), specifically its **`MiniMax H3 VAE Decode (fast)`** node.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The checkpoint is installed in ComfyUI's models/vae/ directory. [1]
    It contains a **2-in-1 MiniMax H3 X2 VAE**, an optional ComfyUI detail-enhancement node, an example workflow, and the full research article below.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The release is distributed under the MiniMax H3 Community License Agreement. [1]
    This release is derived from MiniMax H3 and is distributed subject to the **MiniMax H3 Community License Agreement**.
    Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.

Sources

1
1
Hugging FaceOwner-controlled
https://huggingface.co/speach1sdef178/MiniMax-H3-X2-Detail-VAE Retrieved 2026-10-27 01:40 UTC
Verified

Versions

2
VersionEffectiveChangeChanged fieldsState
22026-10-02 13:39 UTCmaterial changeavailability, deploymentCurrent
12026-10-01 14:34 UTCpublicationNoneCurrent

In briefs

2

Open in the catalog

Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.