release: 0.2.2 — corrected grain-ladder measurements - #54
Merged
Conversation
v0.2.1 shipped the grain ladder built on an 86-title, single-scene probe. That sample was too small to be trustworthy and several of its published figures were wrong, so the released documentation was misleading. 0.2.2 carries the re-measurement: 752 titles across 2230 scenes, three scenes per title. Individual decade buckets moved by up to 78% and the overall median by 38%; two conclusions were false (the 1960s are not the grainiest decade, and TV does not stay grainier than film into the 2010s). It also surfaces the largest effect in the data, invisible at the smaller sample: 1080p masters are 2.2x grainier than 4K masters, which is where bench material should come from. No library or filter code change -- documentation and bench methodology only. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
v0.2.1 shipped the grain ladder built on an 86-title, single-scene probe. That sample was too small to be trustworthy and several of its published figures were wrong, so the released documentation was actively misleading. This cuts 0.2.2 so the released docs carry the corrected measurement.
What 0.2.2 corrects
Re-measured at 752 titles across 2230 scenes, three scenes per title:
Two conclusions v0.2.1 stated were false: the 1960s are not the grainiest decade (mid-pack; the 1970s-80s peak), and TV does not stay grainier than film into the 2010s (they converge).
Root cause, measured: within-title scene-to-scene spread has a median of 0.71x the title's own median and p90 of 2.02x, so a one-scene probe misestimates by ~36% typically and 100%+ for a tenth of titles.
New finding, invisible at the old sample
1080p masters are 2.2x grainier than 4K masters (median 0.00724 vs 0.00323, like-for-like native crops). 4K releases are denoised hard enough to compress, so the impairment has already been removed. Bench material should be 1080p.
Scope
Documentation and bench methodology only — no library or filter code change. The ladder table and the
netflix-bar/--grain <= 16recommendation are unchanged, since those were measured directly rather than sampled.Gates
21/21 fast suite, clang-format clean, changelog render in sync, meson version-skew assertion passes.
🤖 Generated with Claude Code