Home / Tools / EMA-VFI

EMA-VFI

Efficient AI frame interpolation model for smoother video.

Visit EMA-VFI
EMA-VFI preview
github.com Open Source

Upvotes

0

Pricing

Open Source

Category

Video Enhancers

In the database since

Oct 2026

Overview

About EMA-VFI

EMA-VFI is an open-source video frame interpolation model from a CVPR 2023 paper. It creates new frames between existing ones, which makes video smoother or turns clips into slow motion. The method uses inter-frame attention to track both motion and appearance, keeping results sharp. It is efficient enough to run on modest hardware compared to older models. It is made for researchers and video makers who need quality frame interpolation. The code is open under the Apache 2.0 license. Check the official GitHub repo for the paper and code.

Key Features

1. Inter-frame attention

It watches motion and appearance across frames, so the new frames look natural and sharp.

2. Efficient design

The model needs less compute than older methods while keeping the final visual quality high.

3. Slow-motion output

The added frames turn normal clips into smooth slow-motion video with clean, natural, smooth motion.

4. Open code

The GitHub repo shares the full training code and trained weights for open research use.

5. ComfyUI use

Many video makers use EMA-VFI inside ComfyUI video workflows for their own AI video projects.

Frequently Asked Questions (FAQ)

What is EMA-VFI used for?

It adds frames to video for smoother playback or slow motion. Researchers and video makers use it where quality matters.

What is frame interpolation?

It means creating new frames between existing video frames. More frames make motion look smoother or slower.

How is it different from older methods?

It uses inter-frame attention on motion and appearance together. That keeps results sharp while needing less compute.

What hardware do I need?

It is more efficient than older models, but you still need a GPU. Check the README for the exact setup it was tested on.

Can I use it in ComfyUI?

Yes. The community uses it in ComfyUI video workflows. Check the ComfyUI guides for the current node setup.

Conclusion

EMA-VFI brings research-grade frame interpolation to open use. Its efficient design and sharp results suit researchers and video makers who need smooth slow motion. The open code invites study and reuse. Try it on a short clip first to judge the quality yourself. Check the official GitHub repo for the paper, code, and setup steps.

Newsletter

Tools like this change fast

New Video Enhancers tools launch every week — the newsletter keeps you ahead of what's worth trying.

Keep exploring

Similar AI tools

Ranked by the job they do, their capabilities, and how closely they fit this workflow.