What resolution should I use for best speed/quality?

Use ImageScaleToTotalPixels to target around 0.5–1.0 megapixels on mid-range GPUs (e.g., ~720p equivalent). Higher pixel counts improve detail but increase VRAM and render time. The node preserves aspect ratio, so you can scale without distortion.

Does the output include audio, and can I disable it?

MiniMaxH3ImageToVideo can produce a native audio track alongside the video. SaveVideo writes it into the MP4 if present. If you prefer a silent clip, generate as usual and then mute/replace the audio in a video editor—this avoids extra inference.

How fast is the 8-step model compared to the base?

The FastH3 8-Step V2 checkpoint is a DMD2-distilled variant that renders in just 8 sampling steps, significantly reducing inference time versus the base H3 model. Actual speed depends on your GPU, resolution, and clip length.

How do I get consistent results between runs?

Set a fixed seed in the MiniMaxH3ImageToVideo node and keep the same input image, total pixel setting, and clip parameters (duration/fps). This will make the output reproducible across runs.

FastVideo FastH3: Image to Video

This ComfyUI tutorial workflow turns a single still image into a short video with synchronized, auto-generated audio using the MiniMax H3 family via the FastVideo FastH3 8-Step V2 checkpoint. The pipeline is simple and fast: LoadImage reads your source, ImageScaleToTotalPixels (assisted by GetImageSize) rescales it to a GPU-friendly pixel budget while preserving aspect ratio, and the MiniMaxH3ImageToVideo node (id: 4c314f31-ecda-4b08-ae98-faaba1bf613f) renders both video frames and a native audio track in one pass. SaveVideo then writes an MP4 output that includes the model’s audio when available.

FastH3 8-Step V2 is a DMD2-distilled variant of MiniMax H3 that cuts inference to just 8 sampling steps, delivering much faster turnaround than the base model while maintaining audio/video synchronization. The workflow uses the minimax_h3_video_vae_fp16.safetensors VAE and the FastVideo/FastVideo-FastH3-8-Step-V2 checkpoint, giving you a practical, single-image-to-video path that’s ideal for animating portraits, product shots, and quick social clips without extra audio post-processing.

API

Use this workflow from code

Every Comfy workflow is a JSON graph. The payload below is this workflow, exactly as ComfyUI runs it — fetch it from the URL, keep it in version control, or load it in ComfyUI and run it node by node.

0edd9b3fffb3.json

Fetching workflow JSON…

Run it from TypeScript or Python with the Comfy SDK. The same code targets Comfy Cloud or a ComfyUI you host yourself — only the base URL changes.

// Install (beta)
npm i @comfyorg/sdk

// Run this workflow (TypeScript)
import { Comfy } from "@comfyorg/sdk";

const client = new Comfy({ apiKey: "comfyui-..." });
const wf = await client.workflows.fromFile("workflow_api.json");
const job = await client.run(wf);
await job.getOutputs("<output-node-id>")[0].toFile("output.png");

The SDK takes a workflow in API format: open this workflow in ComfyUI and use File → Export Workflow (API).

Comfy Cloud API access requires a plan with an API key.

SDK docs Get an API key

FAQ

Frequently Asked Questions

View all workflows