FSR1Node: Add FSR 1 port for WebGPURenderer.#33339
Merged
Mugen87 merged 5 commits intomrdoob:devfrom Apr 6, 2026
Merged
Conversation
This was referenced Apr 6, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related issue: -
Description
The PR adds a TSL Port of FSR 1 for
WebGPURenderer. I've experimented with Spatial Upscaler recently and I think it's good to have at least one upscaling option in three.js.I've also tested a Bicubic upscaler but the differences to the default Bilinear are so minor that it makes more sense to use FSR 1. FSR 1 internally works with two passes, one that upscales and other one that sharpens the image. I've used the "Little Tokyo" asset in the new
webgpu_upscaling_fsr1demo since it has a variety of features that are good for comparison.Vegetation
Text
Details
You can see that FSR 1 produce a better result that Bilinear although it is not comparable with an image rendered in full native resolution which is as expected though. But because of it's easy setup and it's good results, I think FSR 1 is still a worthwhile addition.
That said, one should keep in mind that FSR 1 has it's own overhead and not every scene will benefit from it. Many three.js scenes are so simple that they will be rendered faster in full native resolution instead of e.g. half resolution and upscale with FSR 1. Only use FSR 1 if the app is actually fragment shader bound and lowering the internal resolution brings you back to your target framerate.
Compared to the original FSR 1 source of AMD, the port has not the same performance level in the shader because a) we don't make any use of textureGather() (not available in WebGL 2) and b) we can't make use of packed math (e.g. two 16-bit ops in parallel) which would significantly lower arithmetic costs. This is a general restriction of the web platform though. Besides, a render pass is in general more expensive in the web than compared to native apps. Since FSR 1 requires two passes, the related overhead is more noticeable than in native environments.
Live Link
https://rawcdn.githack.com/Mugen87/three.js/9ce5a585f373cfbe2bd55f009c539f480b45fbe3/examples/webgpu_upscaling_fsr1.html