Guide ids are session metadata that playback ignores; changeBackground 1 or
absent applies the background type and colour, any other value gives white.
Background types 1..4 = white, black, transparent, other. The recorded step
carries the last used background. Properties panel background changes record
editArtboardEvent.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Preserve the target-only call and exactly reconstruct empty-guide settings. Apply preset tags without resizing, retain fractional RGB, and refuse unresolved guide state before mutation. Nonempty associations and unproved flag values remain unfinished.
Preserve the original reference history and measurements, then document the 22 newly supported forms and their strict mapping boundaries. Artboard from Group remains kept; the Crop Photos parent does not cover generated helper recordings.
perf_scenarios rows that time a command and then the canvas refresh now also record each part
(cmd_samples_ms, refresh_samples_ms; samples_ms stays their sum, so budgets and the baseline are
unaffected), the refresh kinds (gpu-rect, gpu-full, rect/full on the CPU compositor, none) and the
CpuReason kind of any fallback. The console line ends with the medians (Summary::of's p50, as
p50_ms uses).
Bench::refresh follows the app's damage_since: it tracks the (document, revision) the canvas
shows and redraws nothing at the same revision, the last edit's rect one revision on, else
everything; drag_frame invalidates it. A timed refresh that also drew untimed setup edits, or
redrew everything because the canvas showed nothing or another document, is reported on stderr.
Rows that timed setup:
- P8-P11 (duplicate/paste a 12 MP layer): the fresh layer and the copied clipboard are refreshed
before the timer; the "first" rows delete what each sample added, so the stack no longer grows
by two 12 MP layers per sample; the repeat rows get their own source and clipboard, warm up
first, and are set up only when --only/--group selects them
- Spot Healing 10 MP refreshes before its timer, so sample 1 no longer redraws the new document
and the Content-Aware Fill row's edits
- Open PSD drops the previous sample's session before the timer and frees its GPU textures and
compositor pages (GpuCanvas::retain, as closing a document does); gpu_bytes 293 -> 112 MB
Ported from perf/fix-measurement, which is too far behind main to merge; its xtask baseline
merge, regressions_fail setting, 40-sample rows and nightly workflow change are left out.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every remaining event recorded in the reference application through its
menus: Scale Effects, Create Layers, Frame and Artboards, all eleven
Stack Modes with their plug-in codes, video layer and frame commands,
Create Work Path, Find and Replace, Check Spelling, measurement scale
and Scale Marker, Close All, data sets (apply, import, export), Package,
Video Frames to Layers, Import Notes, Photomerge, Crop and Straighten
Photos, Conditional Mode Change and Fit Image. Close Others and Define
Variables record nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Measured in the reference application: Batch source and destination keys
(open documents, Save and Close, override Open and Save As, suppress open
options), file naming output (the extension component only gets a dot as
the last component), processing order, flat output with subfolders, name
clashes, stop and log error modes with the log file format, and Esc.
What a normal recording names (layers by name, merged nudges and typing,
document offsets) and how it plays on another document. Trace Contour's
recorded keys and its 0..255 level; Quick Export records no step.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Measured in the reference application through its dialogs and recorded
actions: Shear point ranges and what ShrS/ShrE mean, the classic and the
new Gradient tool steps, Content-Aware Fill, Perspective Warp, Print,
Create Droplet and Sky Replacement descriptors, the mmdd/ddmm Batch date
components, and every layer comp reference form. Export As records no
step; Convert Stroke to Fill was not available in the measured version.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Stream PSD user and real masks into tile rows
* Correct streamed-mask fixture metadata and test lint issues
* Budget PSD pattern expansion and bound multichannel conversion
---------
Co-authored-by: FredApps <FredApps@users.noreply.github.com>
Addresses CodeRabbit on #10603: the Group A intro said every Filter Gallery effect loads as a
kept step, but Spatter is mapped; Custom's scale, offset and matrix cells now have their
measured ranges and the matrix has no default (a step carries all 25 cells).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Recorded in the reference application through its menus and read back from the saved Actions
panel, or returned by the dialog on OK: Custom, Fibers, Displace (with ranges), Color Lookup,
Shape Blur, Lighting Effects, Blur Gallery, Liquify, Camera Raw Filter, Vanishing Point,
Calculations; how Image Size, Canvas Size, Convert Mode and New Group are recorded. The Batch
descriptor with its file naming components, and the Actions panel menu and Playback Options.
Known gaps gain four rows: Convert Mode is saved with an enum the reference application
rejects, adjustment layer opacity on Make, Canvas Size in percent, Image Size in cm/inches.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Known gaps: every exported step of the round-trip test was played in the reference
application (128 of 151 ran); six codes the mapping writes are wrong there (Vibrance and
Gradient Map events, Stroke Outside, Pass Through, canvas extension colour, Absolute
Colorimetric), and a document offset wraps around there.
- Reference forms: layers and documents by index, id, offset and ordinal, measured.
- Newer events: 63 candidates from the menu catalogue run in the reference application;
which need no settings, which were measured with settings, which need their dialog.
- Version 12 files: the older descriptor layout, checked on 325 real sets.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The contract tests no longer compiled on main (AtnItem::dictionary is now
Vec<u8>, so it is compared with b"Future event"), sat outside main's
tests/it layout, and the new atn.yml added six legs (two macOS) to every
PR against ci.yml's policy. Move the tests into tests/it so the existing
CI runs them, drop atn.yml and docs/atn-testing.md, keep main's ci.yml and
fuzz target, and note the committed fixtures in ATTRIBUTION.md.
Co-authored-by: denstu473 <63150090+denstu473@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Bound PSD import pixel storage before decoding
* Include ZIP inflater workspace in PSD preflight
* PSD import: budget unknown-mode layers, stop after a fatal layer error, keep best-effort for non-allocation limits (#8593 review)
* Stream PSD user and real masks into tile rows
* Correct streamed-mask fixture metadata and test lint issues
---------
Co-authored-by: FredApps <FredApps@users.noreply.github.com>
Every mapped event was checked against the reference application: the corpus sets'
5 296 recorded steps and factory descriptors went through Load and Save Actions and were
compared with the original, and the forms PhotoCraft writes were played there. The new
section lists the settings mapped events still lose (transform reference point, show
only this layer, group names, clipping and colour labels on Make, shape layers, layer
via copy options, load selection from another layer, Background by property, Close that
asks, Lab Photo Filter colours) and the settings the commands do not have yet.
The supported table showed 15 direct adjustment events with shifted columns.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>