9 ms·
Does this mean we can basically render any WebGL code server side and effectively stream the output?
by antihero 10y ago
Does this mean we can basically render any WebGL code server side and effectively stream the output?
- jpambrun 10y agoAnd, if so, is it hardware accelerated or a software implementation? Edit: And, how many concurrent context could it support?
- roblabla 10y agoIt looks like it uses whatever native implementation of EGL there is behindt he scenes. So it should be hardware accelerated if there is a GPU on your server. As for whether you can stream the output, looks like you can with `gl.readPixels`, but you'll probably want to compress it somehow. Sending a raw video stream is heavy.
- vvanders 10y agoThat's a really bad idea, glReadPixels is usually a full pipeline flush + read back and stall which kills perf for all sorts of reasons.
- 33a 10y agoYou could, but it would be slow, or at least pretty heavy per client. That said if you have a beefy enough server I suppose it might work.
- nomel 10y agoI don't really understand what's going on here. Is there a practical use for this? When would rendering, server side, ever be beneficial (assuming that's what's going on here)?
- 33a 10y agoLots of places. For example: * You want to generate preview images of WebGL content, such as in a gallery * You are using WebGL to accelerate some computation in nodejs with a library like https://github.com/waylonflinn/weblas https://github.com/waylonflinn/weblas * You want to test your WebGL code using continuous integration etc.
- jpambrun 10y agoI know that some medical image viewers use server side rendering for segmented volumes. Such tasks are computationally intensive and usually involves very large datasets. With server side rendering, you can easily and efficiently stream videos or even JPEGs to thin clients such as mobile devices.