benjiyaya

benjiyaya / ComfyUI_Dreamid-Omni

Public

Generate talking-head videos with custom character faces and voices directly within ComfyUI.

19
1
100% credibility
Found Mar 29, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A ComfyUI custom node that generates identity-preserving talking-head videos from reference face images, voice audio, and descriptive text prompts.

How It Works

1
🔍 Discover ComfyUI Add-on

You find this fun ComfyUI tool that turns your photos and voice clips into talking videos where people look and sound just like themselves.

2
📥 Add to ComfyUI

Drag the tool into your ComfyUI folder and restart – now it's ready to use like any other creative block.

3
📦 Grab the Magic Files

Download a few special picture and sound packs and drop them in the right spots so your videos come out crystal clear.

4
🎨 Load a Sample Scene

Open an example layout in ComfyUI to see how it all connects – it's like a ready-made recipe.

5
🖼️ Upload Your Face & Voice

Pick a clear photo of someone's face and a short voice clip – the tool remembers exactly who they are.

6
💬 Describe the Action

Type what happens next, like 'the person smiles and says hello while stirring coffee' – keep it lively!

7
▶️ Press Play & Watch Magic

Click generate and in minutes, your custom talking video appears, perfectly matching the face, voice, and scene.

🎉 Share Your Talking Video

Download the smooth MP4 of your lifelike talking character – ready for social media or fun projects!

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ComfyUI_Dreamid-Omni?

This Python custom node brings DreamID-Omni to ComfyUI, letting you generate AI talking head videos with custom character faces and voices directly in workflows. Feed it reference images and audio clips plus a tagged prompt—like for identity and speech for dialogue—and it spits out MP4s preserving facial and vocal traits for single or two-person scenes. Solves the hassle of stitching separate audio-video tools by handling identity-preserving generation end-to-end.

Why is it gaining traction?

It runs on consumer 24GB GPUs via BF16/FP8 models and PyTorch SDPA (no Flash Attention builds needed), plus Sage Attention support for speed. Native VIDEO output plugs into ComfyUI pipelines, with example workflows for quick starts and prompts controlling scenes, actions, and speech. Stands out for multi-person convos and memory tricks like text encoder offload.

Who should use this?

ComfyUI power users building AI character animations or talking head videos for demos, ads, or avatars. Video AI experimenters needing custom faces/voices without server farms. Indie devs prototyping generative talking head AI in workflows.

Verdict

Grab it if you're deep in ComfyUI and have 24GB VRAM—solid for quick identity-locked clips despite 19 stars and 1.0% credibility signaling early maturity. Docs and examples are clear, but expect tweaks for edge cases; test on your rig first.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.