Skip to main content

Rendering Videos

As mentioned in the quickstart guide, you can render videos directly through the user interface of the editor, or through a function call.

Rendering with renderVideo() function​

To render videos through a function call, you can use the renderVideo() function (API reference):

import {renderVideo} from '@revideo/renderer';

async function render() {
  console.log('Rendering video...');

  // This is the main function that renders the video
  const file = await renderVideo({
    projectFile: './src/project.ts',
    settings: {logProgress: true},
  });

  console.log(`Rendered video to ${file}`);
}

render();

Rendering from the browser​

To render videos from the editor, simply click the "Render" Button after opening the editor with npm start:

Render Button

How does Rendering work?​

The rendering process mostly runs in the browser. The only part that runs in a separate "backend" nodejs process is everything related to audio processing.

The code in the browser loops through all frames in the defined video, draws the defined frames onto an HTML Canvas and feeds them into the VideoEncoder of the WebCodecs API. This turns all frames into a mute mp4 file.

In the "backend", ffmpeg is used to extract all audio from the <Video/> and <Audio/> elements of the project and to merge them into a single audio file. This file finally gets merged with the mute video produced in the browser to obtain the resulting mp4 video.

Rendering Speeds​

Since the release of v0.4.6, rendering videos is almost always faster than real-time (meaning that rendering a 1-minute video takes less than one minute). We have created a guide that describes what affects rendering speeds and how to make rendering faster.

Parallelized Rendering​

If you want to speed up rendering, you can parallelize the rendering process of your videos. This means that instead of rendering your full video through a single process, you can use multiple processes to render only small parts of your video and subsequently merge the single parts together (for example, you can use 10 processes to render 1 minute of a 10-minute video each).

The renderVideo() function (API reference) provides a settings.worker argument that you can use to parallelize rendering in a single process. This is useful when you have a lot of RAM available.

Alternatively, the best way to set up parallelized rendering is to use serverless function providers like AWS Lambda. Revideo provides a renderPartialVideo() function (API reference) that you can use to render a partial video - you can use it together with concatenateMedia() (API reference) and mergeAudioWithVideo() (API reference) to parallelize rendering across serverless functions. We also provide guides and example projects for setting up parallelized rendering on AWS Lambda (recommended) and Google Cloud Functions.