Since ed7df56a, list_recovery also lists a committed bundle whose sidecar
is not yet published (revision 0). The worker commits the manifest before
it writes the sidecar, so the retry test could catch the bundle between
the two and read revision 0 (seen on windows-latest; 215/300 local runs
on Linux under load). Wait until the entry carries the expected revision.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Discover committed autosave bundles whose JSON sidecar was never published. Rebuild recovery metadata from the readable manifest and retain these bundles during orphan cleanup.
Fixes#4967
* Tests: bad-layer-id checks use an id that can't exist (#3343)
Layer ids come from one process-wide counter (shared with document and
smart-object ids, and bumped by ensure_ids_above on file loads). A full
`cargo test` run passes ids like 9999, so a layer a test just created can
get the "bad" id, the command succeeds and the assertion fails at random
(seen on CI in type_extra_cmds::tests::vertical_orientation_renders_columns_inside_the_canvas).
Use u64::MAX, which the counter never reaches, for every "no such layer"
sentinel in the engine and ui-egui tests. Test-only change.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Tests: use u64::MAX for bad-layer-id sentinels added on main since (#3343)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
fit() computed (y + 1) * h in u32, which passes u32::MAX for a plane
over 2^32 / max rows; the ABR and pattern parsers' edge caps keep real
inputs below that, but the guard lived in another crate. Spans are now
computed in u64. gradient_bake::eval indexed stops[0] and stops[i + 1]
relying on bake's two-stop check; it now handles an empty or lone stop
itself. Same results for every input that worked before.
Co-authored-by: Claude Opus 5.5 <[email protected]>
The merged image carries its transparency in the first extra channel
when the layer count is stored negative, or when a "Saving Merged
Transparency" global block (Mtrn, Mt16, Mt32) is present. Photoshop
saves 32-bit documents the second way: a positive layer count, Mt32,
and a fourth channel named "Transparency" (psd-tools transparentbg).
merged_has_alpha only knew the first, so opening such a file added a
"Transparency" alpha channel to the document, and the oracle compared
it as if it were opaque white (83.7 % of pixels off; now 0.9 %, all on
the shape's edge).
Co-authored-by: Claude Opus 5.5 <[email protected]>
A mask made from the selection takes the selection with it in Photoshop: after
Layer > Layer Mask > Reveal Selection (the Layers panel's mask button with a
selection) the marching ants are gone. PhotoCraft kept the selection, so the
usual next step, Invert (Cmd/Ctrl+I) on the new mask, inverted only the part
inside the old selection and left the whole mask black, instead of flipping
the mask. Hide Selection (the Alt-click variant) already dropped it.
Reveal Selection now clears the selection in the same history step as the
mask: one undo brings the selection back and removes the mask.
Tests: adjust_cmds::tests::reveal_selection_drops_the_selection_so_invert_flips_the_whole_mask
(8/16/32-bit). Tests that deselected right after Reveal Selection no longer
need to (Deselect is disabled without a selection); layer_mask_depth and the
automation agent tasks now expect the selection to be gone.
Co-authored-by: abc2cb <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
Layer Knocks Out Drop Shadow only knocked the shadow out through see-through fill: at
100 % fill the layer was taken to cover it. That holds in Normal mode, but a layer in
Exclusion, Difference, Multiply, Screen and the rest blends with what's beneath it,
shadow and all: a white Exclusion bar with a drop shadow over black type came out as a
blank white bar instead of white type on black (Photoshop's merged image of the same
file shows the shadow kept out from under the bar), and a Multiply layer was darkened by
its own shadow. shadow_see_through() now knocks it out wholly in those modes, for the CPU
compositor and the GPU plan alike.
Co-authored-by: Daniel van Lith <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
Session::execute catches a command's panic and returns an ordinary Err (the
last-resort guard in jobs.rs), so the unwind never reached panic_hunt's own
catch_unwind and a panicking command counted as a pass.
The harness now installs a panic hook that counts every panic in the process,
including on rayon workers, and a run fails if the count moved. Runs are
serialised so a panic is charged to the run that caused it. No production
code changes.
Test: contained_panics_count_and_graceful_errors_do_not. A contained panic
fails the run and a graceful Err passes; the first assertion fails without
the hook. The full --ignored fuzz stays green, so no command panics today.
Co-authored-by: prekabreki <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
* test(engine): add Color Range fuzziness 0 exact-match test (#214)
Adds an explicit test for Sampled Colors at Fuzziness 0 (point sampling
and hex colors) asserting that exact matches are fully selected and
adjacent shades are rejected. Updates scorecard TOOL-214-3 to done.
* docs: regenerate scorecard (TOOL-214-3 done)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017WeVmjnnPPKdxVTpr2Vjtw
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
Opacity and Fill are fractions in layer.setProps but percentages in the
Layer Style commands, and a value above 1 was clamped silently: 55 gave
an opaque layer. Out-of-range or non-numeric values now fail with an
error that names the range, and the layer is left alone.
select.object samples the active layer by default. On an empty layer
(scripts often add one above the photo) it found nothing and returned
without an error; the failure only surfaced at the next command that
needed a selection. It now fails with an error that names the cause and
the sampleAllLayers option, also when the layer's pixels lie outside the
rectangle or point.
Fibers renders from the foreground and background colours without
reading the layer, but its output area was the layer's content grown by
a zero halo: on an empty layer that is empty, so nothing was drawn (and
a following transform found nothing to transform). On a partly filled
layer only the content's bounding box was filled. It now writes the
reference bounds (the selection's, else the canvas) like the global
filters, without reading the whole layer as they do.
The Background has none of them. layer.setProps applied them to it as it
was, and Select and Mask's Layer Mask output gave it a mask. Like
Photoshop (and like PhotoCraft already does for moving, transforming,
erasing, styles and Layer > Layer Mask), it now becomes a normal layer,
"Layer 0", first. Values that leave its look alone (opacity and fill
100 %, Normal) keep it the Background.
Fixes#4836
* Reject clipboard origins that cannot be moved into a new document.
* Fill Clip.resolution in the unrepresentable-origin clipboard test.
main added resolution to Clip; the synthetic overflow fixture omitted it and failed engine lib tests.
* Format the clipboard origin test (cargo fmt)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Brandon Thomas <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
Reds/Yellows/Greens/Cyans/Blues/Magentas selected hue closeness times
chroma. Measured against Photoshop 2025 on 126 colours per preset
(Reds, Yellows, Blues; hue +-40 degrees, saturation 5..100 %,
brightness 33..100 %), Photoshop also multiplies by (2 - max): bright
colours were already exact, darker ones came out up to 61 levels short.
With the factor every sample matches within one level. The muted navy
#3d3f54 from the issue now gets 35/255, as in Photoshop (it was 21):
Photoshop also leaves it below 50 %.
hRes and vRes are stored as pixels per inch. The unit that follows only
controls how that number is shown, so a 300 ppi file with unit cm opened
at 762 ppi. Export already writes the document ppi and does not divide
by 2.54.
Fixes#5201
Co-authored-by: Cursor Agent <[email protected]>
- ROADMAP.md: the oracle floor is 145 of 258 (crates/io/tests/corpus.rs,
gaps.md G1), not 133 of 256; 953 open issues, not 579.
- docs/architecture.md: ~539,000 lines in ~1,310 files and ~8,040
#[test] functions, recounted 2026-10-13.
- README.md: "a layer kind filter" instead of "layer filters"; only the
Kind filter exists.
- AGENTS.md: drop an empty bullet at the end of section 7.
Co-authored-by: Claude Opus 5.5 <[email protected]>
Testers report ~50 ms input-to-pixel while painting on Linux and macOS, and asked for a
per-stage breakdown. With PHOTOCRAFT_STROKE_TRACE=1 the app prints, per stroke, p50/p95 of
the frame interval, pointer moves per frame, the live stroke's render, the canvas refresh
(composite + upload), the UI build, eframe's CPU time (UI, tessellation, submission) and the
remaining vsync/swap-chain wait, with the GPU backend; =<file.csv> also appends every frame.
The last summary is in the control channel's perf.strokeTrace.
RTX 5090, Vulkan, 120 Hz: interval 8.33, cpu 2.2 (ui 0.9, stroke 0.8, canvas 0.4), wait 6.0 ms.
* fix(compose): a wide stroke effect keeps a round outer edge (#3646)
A stroke's outside distance came from the 5x5 chamfer, whose level sets are polygons: at a
245 px outside stroke the boundary wanders ~10 px into visible flat facets. Use the exact
Euclidean distance (edt_nearest, already used for glows and chiselled bevels) for the outer
edge; the inside edge and spread/choke keep the chamfer, whose sub-pixel edge seeding is
fitted to Photoshop at small offsets. A point source dilated by 60 px now stays within 0.5 px
of a circle (was 2.5 px).
Tests: a_wide_outside_stroke_is_round_not_a_polygon; stroke_edges_use_the_exact_euclidean_distance
(a 3 px outside stroke's (1,3) corner coverage is 0.838, the exact distance; it was 0.764).
Full photocraft-compose suite: 206 pass; fmt and clippy clean.
* fix(gpu): a wide stroke keeps its exact outer edge on the GPU (#3646)
1080008a fixed the CPU compositor's stroke outer edge to use the exact
Euclidean distance instead of the 5x5 chamfer, whose level sets are
polygons and facet a wide stroke. The GPU compositor builds the outer
stroke field from FieldKind::StrokeOutside, which still maps to the
chamfer, and the canvas is drawn on the GPU, so the fix did not reach the
app.
Add FieldKind::StrokeOutsideExact (dist_outside_by Euclidean) used only by
the stroke. The spread of drop shadows and softer outer glows keeps the
chamfer on purpose (fitted to Photoshop, #224), so StrokeOutside keeps its
meaning. StrokeOutsideVector is exact too, matching the CPU vector outline
path.
Tests: crates/gpu/src/fx.rs a_raster_stroke_uses_the_exact_outer_field
(stroke uses the exact field, shadow spread keeps the chamfer), and the
banded/whole field test covers the new kind.
* fix(stroke): a stroke keeps the chamfer narrow and the exact edge wide (#3646)
The exact Euclidean outer edge (#3646) fixed a wide stroke's faceting but
regressed Photoshop's own composites against the PSD flatten oracle: the
fitted 5 × 5 chamfer matches Photoshop at small offsets, and only at a large
outside width do its octagon facets show. Pick the metric by the stroke's
outside width instead: below `STROKE_CHAMFER_MAX` (20 px) keep the chamfer and
its sub-pixel edge seeding, at or above it use the exact Euclidean transform.
The measured chamfer deviation from a point source is 0.027 · r px — 0.48 px at
r = 18, the widest outside stroke any Photoshop oracle matches with the chamfer,
and 1.6 px at r = 60, where the boundary visibly facets. Half a pixel is reached
at r ≈ 19, so the switch sits at 20. Inside stroke edges keep the chamfer.
The CPU and GPU compositors pick the same field by the same width
(`stroke_outside_exact`); `FieldKind` gains `StrokeOutsideVectorChamfer` for a
narrow shape-layer outline, while the shadow / glow spread keeps its chamfer.
io corpus: 140 → 148 (floor 146) and psd-tools 231 → 237 (floor 236), both back
to main's counts with no file newly differing; a 60 px outside stroke stays
round (`a_wide_outside_stroke_is_round_not_a_polygon`).
A renderer's data window is its crop of the frame; the composition lives
on the whole display window (our MY27 renders: data 1782x1267 at
(1109,1100) inside a 4000x3000 stage). decode_deep now validates the
display window against the limits and reports it as stage + origin on
DeepImage; flatten pads the composed region to the stage at its offset
(transparent elsewhere) and tags the image with a DisplayWindowStage
warning. Files without a usable display window open as before.
Includes a cropped-deep generator and a stage/offset oracle test.
* fix(gpu): contain errors in unwatched compositor work
* test(gpu): gate recovery fixtures on format capabilities
---------
Co-authored-by: haranishi <[email protected]>
A flat TIFF dropped EXIF on export ("EXIF not supported") and ignored it
on import. TIFF is EXIF's own container: IFD0 points to the Exif (34665)
and GPS (34853) directories and holds the descriptive tags.
- Export: the EXIF block's Exif and GPS directories are written as
directories outside the page chain, IFD0 points to them, and IFD0's
descriptive tags (ImageDescription, Make, Model, Software, DateTime,
Artist, Copyright) are added unless the text metadata already wrote
them. Values are copied as bytes, reordered to the file's byte order,
so unknown tags survive.
- Import: a classic TIFF with an Exif or GPS directory gets an EXIF
block built from them and IFD0's descriptive tags.
- TIFF's format caps now include EXIF. Layered TIFFs keep EXIF as
resource 1058 as before (read back unchanged); the layered importer
already preferred the TIFF's own EXIF when there is one.
exiftool and Photoshop 2025 read the Exif and GPS tags of such a TIFF.
Fixes#4875
Drag out of the rulers' upper-left corner to move the zero point: cross hairs mark where it
lands, Shift snaps it to the ruler ticks and View > Snap To targets apply otherwise; dropping it
back on the corner cancels. Double-clicking the corner puts it back at the image's top-left.
The rulers then read 0 at the zero point (negative left of and above it) and the Info panel's
X/Y measure from it.
The zero point is a document field (`ruler_origin`, document pixels), saved in .pcraft with
`#[serde(default)]` so older files open at the top-left. It is set by two undoable engine
commands, `view.setRulerOrigin {x, y}` and `view.resetRulerOrigin`, so agents can drive it.
It is not written to PSD: the public PSD spec documents no resource for it.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
The compressed RW2 layouts of earlier Lumix bodies (e.g. the DMC-GX7)
decode at full sensor resolution instead of falling back to the camera's
embedded JPEG preview. The layout was established by observation of
sample files (fixed 16 bytes per 14 pixels, predictor groups with
per-parity predictors, page-permuted downward bit order); the black
level of these files reads 15 counts low.
Adds round-trip tests through a mirrored synthetic encoder, cursor and
predictor-group unit vectors, and updates the support matrix.
Co-authored-by: Gotya-OFFCIAL <[email protected]>
Saving a 36 MP document spent most of its time packing the merged image's
RLE rows one after another on one thread (packbits over ~144 MB), and
importing it decoded each layer's channels sequentially. The merged-image
row packer now runs in parallel chunks that concatenate in order (rows are
independent; errors surface like the sequential path's), the layer gather
runs in row bands on all cores before the per-channel RLE, the layer's
channels decode side by side, and the full-surface clone per layer is
gone.
Measured on a 6000×4000 8-bit document with a photo, a paint layer and a
Levels adjustment: save 2.86 s -> 2.55-2.64 s, open's import stage
1.12 s -> 1.07 s (psd_to_document 745 -> 564 ms). The remaining save time
is dominated by the merged composite render itself (~0.7 s).
* Auto-Blend Stack Images: a depth map by depth from focus, for Lens Blur
`edit.autoBlendLayers` with `method: stack` and `depthMap: true` adds a hidden gray Depth layer, the layer each pixel is sharpest in (black = the bottom layer, white = the top), from `photocraft_algo::dff`: the ring difference filter as focus measure, guided-filter aggregation with the merged picture as guide, a streaming peak search with Gaussian interpolation and a confidence, edge-aware weighted least squares solved by multigrid-preconditioned conjugate gradients, and guided upsampling. Once on a layer mask it drives Lens Blur's depth map.
Closes#2067.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
* Depth from focus: a guide or plane of the wrong length is the empty map
Review follow-up on #2073: `depth_from_focus` is public and indexed
`guide[i]` without checking that `guide` and every `luma` plane are
`w × h` long. It now returns the empty map (depth 0, confidence 0) for a
mismatch, before anything is read, and the test covers a short guide, a
short plane and a size that matches neither.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
---------
Co-authored-by: Claude Fable 5.1 <[email protected]>
* i18n(de): translate the labels that were still English
A walk through the German UI with a lookup recorder (every menu item, every
tool's options bar, every dialog it opens) found 344 texts that fell through
to English. They were invisible to the existing i18n test, which only scans
tl!("literal") calls: the labels of the schema-driven filter dialogs are built
from the engine's parameter names ("centerX" -> "Center X"), and status and
system-info templates are looked up at run time.
- de.tsv: 344 new entries. Filter, Render (Flame, Picture Frame, Tree, Lighting
Effects), Blur Gallery, Lens Correction, Color Lookup, Shadows/Highlights,
Apply Image / Calculations, Layer Style blending options, Find and Replace,
Document Info, the keyboard shortcut editor's commands, system info and the
tool names that were missing (Color Sampler, Vertical Type, Patch).
Written from the plain meaning of each label, in the vocabulary the catalog
already uses (see its header: no proprietary localisation resource consulted).
- "Light" was translated as "Light": the appearance mode in Preferences and the
toolbar tooltip read "Light" next to "Dunkel". It is "Hell" now (Camera Raw's
Light section and the font weight have their own contexts).
- Export As: the dialog title was built with format!() and could not be
translated; it goes through the catalog now.
- Names such as "iPhone 14 Pro Max" (the artboard dialog's preset choices) were
split like camelCase keys and read "I Phone 14 Pro Max"; a name with a
space is shown as it is. Test: display_names_are_not_split_like_camel_case_keys.
Left English on purpose: product and file-format names, units, document and
layer names the engine creates ("Untitled", "Background": other code compares
the latter), brush and swatch preset names (data), and words that are the same
in German (Filter, Info, Zoom, Median...).
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* i18n(de): drop the process note and give the new sections English headings
The first heading described how the entries were found (and when), which is not
catalog content, and the other headings were German while the rest of the file
uses short English section comments. Only comment lines change.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
* i18n(de): US spelling in a section heading
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
---------
Co-authored-by: Claude Sonnet 5.5 <[email protected]>
* Blur Gallery: Field Blur pins on the image (#3373)
Field Blur opened a parameter dialog with one centre, though the engine
already blends any number of pins. It now works as Photoshop's Blur
Gallery workspace: the image under the dialog takes the pointer. A
click adds a pin and selects it, a drag moves a pin, and a drag around
the selected pin's ring turns its blur up or down (one turn is the
0-500 px range; the arc shows the amount). Delete or Backspace removes
the selected pin. The dialog shows the selected pin's Blur and Preview,
and opens at the right edge, beside the image, so the pins stay in view.
The pins live in the dialog's fields (`pins`, `__selected`), so the
existing filter preview follows every change, OK applies
filter.blurGallery.fieldBlur with them in one step, and automation
drives it with ui.dialog.set. Pins are fractions of the selection's
bounds when there is one, else the canvas, as the engine reads them.
Proxy previews now scale Blur Gallery pins' pixel blur like other pixel
params.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* i18n(vi): translate the Field Blur pin strings
main added the Vietnamese catalog after this branch was cut, so
every_tl_literal_is_translated failed on the merge for "No pin
selected" and the pin hint. Terms follow the catalog's own: ghim
(pin), Bấm (click), Delete kept as the key name.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_018CiFWEfKibLnwb6oSy6ZBE
* Blur Gallery: Alt+Cancel Reset starts again from one pin
main's Alt+Cancel → Reset (#3871) restores a filter dialog's schema
defaults, and the schema has none for Field Blur's pins, so Reset
left the workspace with no pins. blur_pins::reset now puts back the
opening pin (in the middle, selected), and reset_fields runs it for
the Blur Gallery. init is reset plus capturing the bounds, which
Reset keeps.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_018CiFWEfKibLnwb6oSy6ZBE
* i18n(hu, tr): translate the Field Blur pin strings
main added complete Hungarian and Turkish catalogs after this branch
was cut. Terms follow each catalog: tű / Mezőelmosás (informal
Hungarian), pim / Alan Bulanıklığı (formal Turkish).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_018CiFWEfKibLnwb6oSy6ZBE
* i18n(sr): translate the Field Blur pin strings
Main gained a complete Serbian catalogue (#4436); every_tl_literal_is_translated needs these strings there too.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
* Profile prompts on open: one file at a time, before it opens, like Photoshop
With Ask When Opening on, files opened together put up every Embedded
Profile Mismatch prompt at once, each answer went to the active document
rather than its own file, and no prompt named its file. Checked in
Photoshop 27.11: one prompt at a time, naming the file, before that file
opens; OK opens it and asks about the next; Cancel opens neither it nor the
files still to come.
A file that asks is now held back until answered: the engine's background
open parks the decoded document (ColorState::open_question,
Session::take_held_open) and the shell queues it (open_prompt). The answer
is applied to that document (ColorState::resolve_open) before it opens.
Files that finish reading meanwhile wait their turn. A developed raw has no
embedded profile, so it doesn't ask.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Profile prompts on open: the prompt's text is localized in every catalog
The two messages are templates now, one line each so the catalog check covers
them: the sentence naming the file and the mode, "Embedded: {embedded}" and
"Working: {working}". The mode reads as the catalogs name it (French "RVB"),
and the file's name goes in last, so braces in it are never taken for a
placeholder. Rows added to all 17 catalogs; main's two "colour profile"
templates (Czech and Italian only) are removed, nothing uses them.
The Camera Raw control test from main expected a mismatch prompt when a raw
opens. With this change a developed raw is not asked about (it has no embedded
profile), so the test now checks that no prompt stands before Camera Raw.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Profile prompts: rows for the hu, sr and tr catalogs; tests hold for any read order
The prompt's four templates join the Hungarian, Serbian and Turkish catalogs
that landed on main.
Files read in the background ask in the order their reads finish, which is the
order given unless a later read overtakes an earlier one. Three tests assumed
the order given and one failed once under a loaded test run; they now follow
whichever file the prompt names (and still require the order given for files
read on the spot).
New test: a file held for its prompt still reports type in a font that isn't
installed, once.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
Select Subject ran every graph cut (the GrabCut iterations and the
full-resolution band re-cut in `refine_band`, tile by tile on rayon)
through the non-cancellable `grid_cut`, so the Remove Background job only
saw Cancel/Esc before and after the whole "Finding subject" step.
Add `select_subject_with`, `grabcut_with` and `finish_region_with`, which
take an `Interrupt` and return `Err(Cancelled)`: each band tile checks it
before reading pixels and its cut runs through `grid_cut_cancellable`, so
in-flight tiles stop at the max-flow's next poll and unstarted ones never
start. Remove Background runs the step under `JobCtx::stage`. The old
functions keep their signatures and results (they pass `Interrupt::NONE`),
so `select.subject`, Object Selection and the layer menu are unchanged.
Co-authored-by: lloyd <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
* fix(psd): prioritize 16-bit and 32-bit global layer blocks over compatibility section (#4928)
In 16-bit and 32-bit PSD files saved by Photoshop, the Layer Info section
either contains 0 layers or 8-bit compatibility layers, while the actual
16-bit/32-bit layers are stored in the global Lr16/Lr32 tagged blocks.
Previously, candidate_seen was initialized to layer_info.is_some(), which
caused PsdFile::from_bytes to skip the Lr16/Lr32 block whenever the section
was non-empty. This left the document with empty layer lists or 8-bit records
that failed channel decoding at depth 16/32, causing all layers to be empty
and invisible.
Prioritize the depth-specific tagged block (Lr16 for 16-bit, Lr32 for 32-bit)
over the compatibility section.
* test(psd): improve 16-bit multi-layer compatibility regression tests
* PSD: Backgrounds survive a round trip through PhotoCraft
Photoshop opens a PSD's bottom layer as its Background when that layer's
record has no transparency (-1) channel. PhotoCraft always wrote one, so
every file with a Background came back from PhotoCraft with an ordinary
"Layer 0"-style bottom layer instead. Measured by opening PhotoCraft's
re-exports of the 427 corpus files in Photoshop 27.11: 317 lost their
Background this way.
- Export: the bottom top-level layer is written without the -1 channel
when it is a locked (transparency and position), unclipped, unmasked
raster that is opaque over the whole canvas at full opacity and Normal,
as a Background always is; flags as Photoshop writes them (bit 4 marked
meaningful, and "pixel data irrelevant" on adjustment layers).
- Import: like Photoshop, a bottom raster layer whose record has no -1
channel becomes the Background (locked, named "Background") whatever
the file names it: psd-tools' "Фон" and ag-psd's "bg" or "Layer 1" open
as Backgrounds in Photoshop.
Of the 227 corpus files whose bottom layer is a Background, PhotoCraft's
export now writes all 227 the way the original does (it was 0), and no
other bottom layer turns into one. Corpus oracles and round trips are
unchanged (io 148/170, psd-tools 237/309, Photoshop 137/258 on this
commit; every round trip renders the same).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* PSD: a flat Bitmap / Indexed file's image opens as the locked Background
Probe 7 (PhotoCraft's re-exports opened in Photoshop 27.11): with the
Background fix, 314 of 318 Backgrounds survived; the other four were
psd-tools' flat 1-bit Bitmap and Indexed files. Photoshop opens a flat
file's image as its Background, but PhotoCraft's import of these modes
left the "Background" layer unlocked, so the export wrote it as an
ordinary layer. An opaque flat image now opens locked (as flat PNG/JPEG
and layered-but-flattened PSDs already did) and saves as a Background.
Corpus unchanged (io 148/170, psd-tools 237/309, Photoshop 137/258;
every round trip renders the same).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
When the wgpu compositor can't draw a refresh (layers over its memory
budget, or a feature it doesn't cover), the CPU compositor redrew the
whole document inside the draw call: minutes of Not Responding on a
multi-gigabyte PSB. With background jobs, a refresh of 4 MP or more is
now composited by a worker in 1024 px blocks, the visible ones first;
each frame uploads the finished blocks and the canvas keeps its previous
contents until they land. An edit while it runs draws its own damage and
restarts the worker on the edited document with the blocks still to
draw. The status bar shows Redrawing N%. Web and tests keep the
synchronous path.
* Brush: remember the last selected brush across restarts (split from #2516)
The persistence half of #2516 on its own, so it can merge and be maintained
apart from the brush-outline half. The tool's brush (settings, tip
references, and the name of the preset it was picked from) is saved once the
pointer is up and restored at launch: options bar, painting cursor and
Brushes panel highlight all start where the session left off.
Review fix from #2516 -- no per-frame work:
- `ToolState::brush_rev` is bumped by every brush/preset setter
(`tools.setBrush`, preset save/delete/rename/setCurrent/deleteGroup,
`brush.defineFromSelection`, `brush.presets.importAbr`, the store's tip
re-linking, and the shell's tool-switch/reset paths); `remember_brush`
serialises only when it moved since `brush_saved_rev`, so an unchanged
brush costs a counter compare per frame, not a clone and a JSON dump of
the whole brush (sampled tip previews included).
- Restoring counts as saved: the launch value is never rewritten.
Bitmaps stay out of the preferences document: remembered tips and patterns
are references (a loaded stored tile drops its full bitmap), and a sampled
bitmap too big to embed is skipped, keeping the previous value so a preview
never enters the prefs file. A malformed remembered brush falls back to the
default.
docs/scorecard.md regenerated with `cargo xtask scorecard` (34 of 154).
Tests: engine 1267 lib (5 new: revision gating, rejected changes, picked
preset, restore round trip, no embedded bitmaps), ui-egui 1982 lib; clippy,
layers, wasm green.
* rustfmt the StoredTile line and regenerate the scorecard after the main merge
* engine tests: destructive Levels keep the signed float overshoot at 32 bits (#2592)
* Scorecard: regenerate after merging upstream/main
* Export As: compare batch paths without the Windows separator
* font fix ?
* fix(ui): only Filter menu dialogs get the in-dialog preview window
#3877 drew the preview window in every schema dialog with a Preview
checkbox, including Image > Rotation > Arbitrary. Photoshop's Rotate
Canvas and adjustment dialogs have no preview window, and the taller
dialog covered the canvas, which broke
arbitrary_rotation_preview::size_changing_rotation_preview_is_visible_and_preserves_the_panned_center
on main. The window and its reserved height now apply to filter.*
commands only.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* cargo fmt --check max_width = 160 fix
---------
Co-authored-by: JJ <[email protected]>
Co-authored-by: Brandon Thomas <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
Clicking a History panel state ran Edit > Undo / Redo once per state. Each
step bumped the document revision, so the canvas could not tell what the
jump changed and recomposited the whole document (re-uploading every
changed layer on the GPU path).
history.selectState {state} returns the document to a state (numbered as
the panel lists them, redo states included) in one command: the history
moves by pointer swaps, the document becomes current once, with one
revision and the damage rect between start and target, and the targeted
layers restored as Undo restores them. The panel runs it once per click;
an open Free Transform refuses it (panel guard kept, and the transform
intercept greys and refuses it for every caller). A floating piece goes
back as with Undo instead of being dropped as a new step.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
