Animated live portrait text-to-video (ComfyUI)

Updated Categorized as Workflow Tagged , , , , , , No Comments on Animated live portrait text-to-video (ComfyUI)

Flux AI is the latest image model that generates the highest quality images. Live Portrait adds facial expression by copying it from a reference video.

This workflow combines both techniques to generate a live portrait from text. All files to reproduce this animated video will be provided.

Reference video for facial expression:

Live portrait output from this workflow:

Software

We will use ComfyUI, an alternative to AUTOMATIC1111.

Read the ComfyUI installation guide and ComfyUI beginner’s guide if you are new to ComfyUI.

Take the ComfyUI course to learn ComfyUI step-by-step.

Step-by-step guide

Step 1: Download the Flux AI model

Download the Flux1 dev FP8 checkpoint.

Put the model file in the folder ComfyUI > models > checkpoints.

Step 2: Update and install missing nodes ComfyUI

ComfyUI has native support for Flux starting August 2024. You need to update your ComfyUI if you haven’t already since then.

The easiest way to update ComfyUI is through the ComfyUI Manager. Click Manager > Update All.

Make sure to reload the ComfyUI page after the update — Clicking the restart button is not enough.

Step 3: Load the Live Portrait Flux workflow

Download the workflow below.

Become a member of this site to see this content

Already a member? Log in here.

Drag and drop the JSON file to ComfyUI.

Step 4: Install missing nodes

If this is your first time using Live Portrait, you will see some nodes missing as red blocks.

Click Manager > Install missing custom nodes.

Install the Live Portrait custom nodes.

Restart ComfyUI.

Step 5: Upload a reference video

The workflow copies the facial expression of a reference video. You can use the video below for testing.

Find the Load Video (Upload) node. Upload the video by clicking choose video to upload.

Step 6: Generate the video

Click Queue Prompt to generate the video. You should get this:

Usage tips

Generate a different image

Change the seed value in the KSampler node to generate a different image.

Prompt

You can customize the image by changing the prompt in the CLIP text encode (Positive Prompt) node.

Reference videos

I find my reference videos on the royalty-free video site Pexel.

Live Portrait is not good with large movements. Find a video that is mainly static with slight changes in the face, such as blinking, winking, and starting to smile.

Andrew

By Andrew

Andrew is an experienced software engineer with a specialization in Machine Learning and Artificial Intelligence. He is passionate about programming, art, and education. He has a doctorate degree in engineering.

Leave a comment

Your email address will not be published. Required fields are marked *