Prevent stale temporal history after camera cuts - #163
Draft
proggeramlug wants to merge 1 commit into
Draft
Conversation
|
Important Draft PR not reviewedDraft PRs are not automatically reviewed by default.
To automatically review draft PRs, update your CodeRabbit configuration: reviews:
auto_review:
drafts: trueThanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
This was referenced Sep 11, 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
After a camera cut, TAA still sampled the previous history epoch even when the CPU requested current-only output. Stale confidence could activate settled-history locks. A 40-frame old-camera control leaves visible residue at roughly 16,900 pixels on Vulkan and DX12, with maximum channel error 232; the original eight-frame DXC case also has a one-byte mismatch.
Apply the existing history-usable predicate before reading old color or provenance. Reset and bootstrap frames retain current-frame defaults and zero confidence; valid accumulation keeps its existing policy. Extend the exact-equality camera-cut regression to both eight and 40 history frames, preserving its later FOV/rotation/pan checks and every threshold. The golden device helper now honors explicit shader compiler options.
Validation: unchanged-shader controls retain their failures. With the guard, all six fresh/reset capture pairs are byte-identical across Vulkan, DX12/FXC and DX12/DXC; each backend's initial fresh image is unchanged. The complete local shared CI component passes under DX12/DXC and Vulkan: 489 library tests, device negotiation, 93 goldens and the remaining integration checks. Formatting, strict Clippy policy, repository contracts and diff checks pass. Existing ignored tests remain ignored. See the report.
Stacked on #162; advances #135/#149/#140. Hosted Tests run 34550069181 completes with 21 successful jobs and one failed Windows shared job. The actual Windows native build and all 20 native example links pass; macOS shared tests pass all 93 goldens, including the expanded camera-cut test. Canonical Metal captures and quality contracts pass. Windows shared tests still crash with
0xc0000005in the library stage before goldens; this remains a separate compatibility investigation. Named full-hardware quality jobs are skipped.Published evidence contains the original failures, all six exact RGBA comparisons, complete local and hosted logs, source-match receipts and example link evidence. ZIP SHA-256:
c03b55eafd7c8d2cf17c08c958440395d4aa65fd37f9beeef1241a53e247d4d7. No full hardware/performance acceptance or merge is claimed.