H3 Editor GitHub Install

A video editor for MiniMax H3, inside ComfyUI.

Write shots, render them on your own GPU, and cut picture and sound together. Do it by hand, or hand the same timeline to an agent.

git clone https://github.com/igorls/comfyui-h3-editor

00:01:12 of 00:15:00

This kitchen scene was written and rendered by an agent through the editor's API, and exported from the editor as you see it: four shots of HyperFlow H3 Flash with their own sound, 15 seconds. The chef is one cast member, made from a single picture.

The launch film

The three shots in it are the ones above, rendered in the editor with their sound. The graphics around them are code, built by a coding agent with the open-source motion-graphics skill; the music was generated with YuE2.

What it does

The editor with the café teaser open: clips and cast on the left, the viewer in the middle, the selected shot's fields on the right, and the three clips with their sound on the timeline below.
The whole editor, on one screen: clips and cast, the viewer, the shot you are writing, and the timeline. It opens from ComfyUI's sidebar and needs no other custom node pack.

Install

  1. Add the pack to ComfyUI

    You need a recent ComfyUI with its built-in MiniMax H3 support. In ComfyUI's custom_nodes folder:

    git clone https://github.com/igorls/comfyui-h3-editor
    pip install -r comfyui-h3-editor/requirements.txt

    On the Windows portable build, run the second step as ..\python_embeded\python.exe -m pip install -r comfyui-h3-editor\requirements.txt. The requirements bring ffmpeg, for importing and exporting video, and the gguf reader for HyperFlow's GGUF models.

    Restart ComfyUI. H3 Editor appears in the sidebar.

  2. Download the models

    The pack does not download models. These are the ones a first video needs, about 54 GB:

    FileFolderSizeWhere to get it
    qwen3vl_32b_minimax_h3_nvfp4_awq.safetensorsmodels/text_encoders15.7 GBComfy-Org/MiniMax-H3, text_encoders/
    minimax_h3_video_vae_fp16.safetensorsmodels/vae5.2 GBComfy-Org/MiniMax-H3, vae/
    minimax_h3_audio_vae_fp32.safetensorsmodels/vae0.6 GBComfy-Org/MiniMax-H3, vae/
    minimax_h3_ref2va_hyperflow_flash_14b-int8_convrot.safetensorsmodels/diffusion_models/hyperflow16.4 GBVideoRebirthAI/hyperflow-h3-flash, ref2va_int8_convrot_comfyui/
    minimax_h3_fl2va_hyperflow_flash_14b-int8_convrot.safetensorsmodels/diffusion_models/hyperflow16.0 GBVideoRebirthAI/hyperflow-h3-flash, fl2va_int8_convrot_comfyui/

    HyperFlow's model page recommends these INT8 files on RTX GPUs with 24 GB or more. On a smaller GPU, take its GGUF files in their place, from Q8_0 at 15.2 GB down to Q3_K_M at 7.2 GB, anywhere in models/diffusion_models; see models and quality.

    These are optional, and worth having:

    FileFolderSizeWhere to get it
    minimax_h3_latent_upscaler_3d_conv_v1_bf16.safetensorsFinals at up to 1920 pixelsmodels/latent_upscale_models0.7 GBLBH-123-AI/Minimax_h3_latent_Upscaler
    MMH3-RealismSlider-V1.safetensorsNatural skin in finalsmodels/loras3 MBCivitai model 3014078
    taeh3.safetensorsThe moving live previewmodels/vae_approx23 MBOzzyGT/taeh3

    Other H3 checkpoints work as well. Models and quality lists every one the editor can render with.

  3. Make your first video

    1. Open H3 Editor from the ComfyUI sidebar. It starts you on a first project.
    2. In the Scene panel, choose the format: 16:9, 9:16 or 1:1.
    3. Press New clip, write what happens in What happens, and press Render draft. The viewer shows the clip taking shape.
    4. When the draft lands, press Space to play it. Not right? Change the text and render again: each render becomes another take.
    5. Press Extend forward to carry the shot on, or add another clip.
    6. Press Render final for the takes you keep, then Export. The video is saved in ComfyUI's output folder, under h3take_projects/<project>/exports.

For agents

Everything the editor does is an HTTP call. The h3-editor skill teaches an agent the whole workflow: plan the shots, render drafts, review them, join them, render finals and export. Its client, scripts/h3.py, needs only Python.

To install it, copy skills/h3-editor from the repo into your agent's skills folder.

The editor is a shared cutting room. An agent's changes appear in your open editor as it works, yours merge with its, and every change is a revision you can go back to.

Read the skillThe full API

$ python h3.py check
$ python h3.py create "The pass"
The_pass
$ python h3.py ops The_pass shots.json
$ python h3.py wait The_pass
$ python h3.py look The_pass
$ python h3.py render The_pass c822cf892 cc2eecf3a caa4b5e11 c23560c00 --final
$ python h3.py export The_pass
Exported to ComfyUI's output folder:
h3take_projects/The_pass/exports/The_pass_001.mp4