What is the current state of headless rendering with WebGPU?

Questions
,
1

The commented flags here suggest that there’s been some experimentation with headless WebGPU rendering in the past but that that work is still incomplete.

I’ve tried headless rendering with puppeteer/WebGPU but only see a blank screen where the canvas should be, even when a GPU is available from chrome://gpu. Is the three.js test suite able to do this successfully?

Example code that results in a blank space in place of a canvas:

'use strict';

const puppeteer = require('puppeteer');

(async () => {
  const browser = await puppeteer.launch({
    headless: 'new',
    args: [
      '--no-sandbox',
      '--headless=new',
      '--use-angle=vulkan',
      '--enable-features=Vulkan',
      '--disable-vulkan-surface',
      '--enable-unsafe-webgpu',
      '--enable-webgpu',
    ]
  });
  const page = await browser.newPage();

  page.on('console', msg => {
    console.log('console text:', msg.text());
    console.log('console location:', msg.location());
  });

  await page.goto('https://webgpu.github.io/webgpu-samples/?sample=rotatingCube');
  await page.screenshot({ path: 'screen.png' });

  await browser.close();
})();