H3 Continuum EZ Studio – Easy Video Timeline for ComfyUI

H3 Continuum EZ Studio

is an easy video timeline for ComfyUI, powered by the H3 Continuum generation engine. Arrange shots, add reference images or audio, and generate a video from one timeline.

GitHub — setup guide, updates, and workflow downloads

EZ Studio and H3 Continuum

EZ Studio keeps the controls focused on shot setup and video generation. Set the prompt and duration for each shot, then queue the whole timeline. Long shots split automatically during generation.

For more control over individual chunks, including reviewing or regenerating a chunk, resuming from a saved take, or running a second pass, use H3 Continuum.

This workflow is designed for very simple timeline-based video; it cannot handle complex tasks, and the issue of artifacts between chunks remains unresolved.

Install or update

First-time installation

Open a terminal in your ComfyUI folder and run:

cd custom_nodes
git clone https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum-EZ.git

Update an existing Git installation

Open a terminal in your ComfyUI folder and run:

cd custom_nodes/ComfyUI-H3-Continuum-EZ
git pull --ff-only

Restart ComfyUI and refresh your browser after installing or updating. ComfyUI 0.44.0 or newer is recommended for the included workflows.

Quick start

1. Open either workflow JSON in ComfyUI.

2. Open the workflow’s subgraph. Follow the note for model download links and folder locations, then select the models you installed.

3. Open Timeline Studio. Set each shot’s prompt and duration, and add reference images or audio as needed.

4. Click Apply and Queue to generate and save the video.

Workflow files

• FastH3 OmniRef (8 steps) — H3_Continuum_EZ_Studio_FastH3_OmniRef.json

• MiniMax H3 REF2VA + Pika Flex — H3_Continuum_EZ_Studio_REF2VA.json

Both workflows include setup notes and links to the models they use. Download the model weights separately.

Model downloads

Download the files for your chosen workflow and the three shared files. The folders below are relative to ComfyUI/models/.

FastH3 OmniRef only

• Diffusion model in diffusion_models/ · fasth3_omniref_8step_pruned_int8_convrot.safetensors

MiniMax H3 REF2VA only

• Diffusion model in diffusion_models/ · minimax_h3_ref2va_pruned_int8_convrot.safetensors

• Pika Flex LoRA in loras/ · pika_h3_flex_ref2va_v1_bf16.safetensors

Shared by both workflows

• H3 text encoder in text_encoders/ · qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors

• Video VAE in vae/ · minimax_h3_video_vae_int8_convrot.safetensors

• Audio VAE in vae/ · minimax_h3_audio_vae_fp32.safetensors

Optional files in text_encoders/

• For Generate AI Text, add the P-H3 prompt upsampler. Video generation with original prompts works without it.

• Alternative H3 text encoders · BF16 or INT8 ConvRot. Choose one H3 encoder in the subgraph.

Make your first video

1. Drag the workflow JSON into ComfyUI and select your installed model files in the subgraph.

2. Open Timeline Studio. Set each shot's duration and prompt, then assign reference images and add audio if you want a soundtrack.

3. Choose Apply and Queue. Play the generated video in Studio or open the saved output from ComfyUI.