Repository navigation
perf(rendering): convert YUV420P frames in 2x2 blocks - #2631
Merged
richiemcilroy merged 6 commits intoOct 18, 2026
Merged
Conversation
Signed-off-by: BestCodes <106822363+The-Best-Codes@users.noreply.github.com>
Contributor
|
@The-Best-Codes is attempting to deploy a commit to the Cap Software Inc Team on Vercel. A member of the Team first needs to authorize it. |
Signed-off-by: BestCodes <106822363+The-Best-Codes@users.noreply.github.com>
The-Best-Codes
commented
Oct 15, 2026
The-Best-Codes
left a comment
The-Best-Codes
left a comment
Contributor
Author
There was a problem hiding this comment.
Ready for review. I just did some benchmarks in editor playback too. I had Claude Opus 5.5 test with a 3024x1964 software-decoded recording (UHD 620), 3 alternating runs per build:
| Mode | FPS main -> branch | Render p50 |
|---|---|---|
| Full 1920x1080 | 22.1 -> 25.0 (+13%) | 25.7 -> 21.7 ms |
| Half 1248x702 | 27.0 -> 31.2 (+16%) | 20.6 -> 17.2 ms |
| Quarter 480x270 | 32.3 -> 36.7 (+14%) | 17.4 -> 14.8 ms |
| Scrub (half) | 9.7 -> 10.1 (+4%) | 24.2 -> 19.8 ms |
Every run on this branch beat every run on main! According to Opus, scrubbing gains less because it's mostly keyframe decoding. Export barely moves because CPU encoding hides the GPU saving there; playback has no encoder, so it shows up directly.
The speedup applies wherever frames are decoded in software (Linux, and Windows without hardware decode). macOS is unaffected.
Signed-off-by: BestCodes <106822363+The-Best-Codes@users.noreply.github.com>
…t-blocks # Conflicts: # crates/rendering/src/shaders/yuv420p_to_rgba.wgsl # crates/rendering/src/yuv_converter.rs
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
2 participants
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.
The YUV420P to RGBA shader ran one thread per pixel. In 4:2:0 video every 2x2 block of pixels shares one U and one V sample, so each colour sample was read four times.
Each thread now converts one 2x2 block and reads its U and V once. The per-pixel math is untouched. It just moved into a helper function.
This shader handles software-decoded frames on Linux and Windows, in both editor playback and export.
Numbers from an i7-8650U with UHD 620, using a 31s 3024x1964 recording with camera:
Without #2602, export time on my machine stays the same, because decoding is the bottleneck there. The GPU work drops either way.
Testing:
cargo test -p cap-renderingpasses.
The PR appears safe to merge.Summary
Converts YUV420P frames in 2×2 blocks to share U and V reads.
Reviews (1) · Last reviewed commit: "perf(rendering): convert YUV420P frames ..." · Reviewed by Greptile