Commit Graph
11425 Commits
Author SHA1 Message Date
David Mann 52dc8e09fb Merge PR #9370: fix(ui): Bevel & Emboss Highlight and Shadow Mode are separate dropdowns (#9368) 2026-10-25 15:01:05 +01:00
Khawar Zamman Wani b61622c1e2 fix(ui): Select and Mask's stale-workspace messages are translated
The two reasons stale() gives for closing the workspace were plain English in the status bar
and the control error. They go through tl! now, with rows in every catalogue and a test that
checks each language's status and error (#8896 review).
2026-10-25 19:27:57 +05:30
highperfocused ead45134d3 docs: keep the Crop row's branch-point entry and describe the canvas menu separately (#6561) 2026-10-25 14:56:45 +01:00
David Mann a2ea2bd872 Merge PR #9446: raw: a non-finite black level no longer panics the AHD demosaic 2026-10-25 14:56:18 +01:00
David MannandClaude Opus 5.5 52359ada45 Address CodeRabbit review: refuse a non-regular write target before opening it
An existing FIFO at the write target reached holds_image (and the Unix
xattr copy), whose read-only open blocks until a writer appears, stalling
every request behind the Headless lock. Check the target's type first,
following links, and refuse directories, FIFOs, sockets and devices.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:55:57 +01:00
dcf5c61a64 Integration: cap MapPolynomial opcodes, finite WhiteLevel, RGB-only JPEG XL preview (#8402)
- An opcode list may hold up to 4096 opcodes, and every MapPolynomial is checked for every
  sample, so a crafted DNG with thousands of whole-image polynomials (a few hundred KB) hung the
  open for minutes. Keep at most 64 per list; the rest are named as not applied.
- A DOUBLE WhiteLevel past f32::MAX became an infinite white, which blacked out that channel.
  A non-finite level now falls back to the default, as a zero one does.
- A reduced-resolution LinearRaw IFD (NewSubFileType 1, three samples, JPEG XL) was a JPEG XL
  preview candidate, and its undeveloped samples would show as an sRGB picture. Previews now
  need PhotometricInterpretation RGB or YCbCr.
- README: floating-point JPEG XL raw data and OpcodeList2 MapPolynomial are supported.

Co-authored-by: SBaksa <18648812+SBaksa@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:55:45 +01:00
David Mann 8c89d4ce41 Integration: merge main (#3281) 2026-10-25 14:52:56 +01:00
248972cfa3 Integration: clip VP clone copy-back to the canvas, paste in the stamp preview (#3281)
A newLayer clone copied back the union of every dab's box: one point far
off the canvas (e.g. [9e6, 9e6], accepted by the params) made it read a
region of petabytes and abort. The command and the stamp preview now copy
back and report damage only inside the canvas.

With an image on the clipboard (Paste is on by default), OK pastes first
and the strokes paint over and sample the paste, but the preview painted
the strokes on the unpasted layer and drew the paste on top. The preview
now pastes first too (shared paste_onto), repaints when the paste moves,
and the overlay is only drawn when no current preview shows the paste.
While the paste is dragged the baked preview is hidden so only one copy
follows the pointer; a plane drag freezes preview and paste until release.

Deleting the last plane now drops the stroke preview.

Co-authored-by: Valentin Klein <271991477+vangel76@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:52:51 +01:00
vangel76andClaude Sonnet 5.5 9e2426f6c1 Format the AHD bit-for-bit test with rustfmt
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:52:26 +01:00
David Mann b9df5c727a Merge PR #6620: Crop tool: right-click on a pending crop opens its menu: presets, overlays, Cancel and Crop (#6561)
# Conflicts:
#	docs/gaps.md
2026-10-25 14:51:12 +01:00
vangel76 258ce7c104 Merge origin/perf/9434-fused-develop (stacked) 2026-10-25 14:51:08 +01:00
vangel76 2863c3281c Merge upstream/main into perf/9435-raw-tone-stage 2026-10-25 14:49:50 +01:00
vangel76andClaude Sonnet 5.5 d90ae8386a raw: faster tone stage (Shadows, Highlights, Clarity, Texture, Dehaze), same bytes
The full-size part of the tone stage cost as much as the whole AHD
development at 45 MP. Everything changed here keeps the arithmetic and
its order, so the developed output is bit for bit the same:

- The bilinear lookup of the reduced change maps (and of Dehaze's
  transmission) worked out the source position with divisions for every
  pixel and did it twice, once for each of the two planes. The taps of
  every column and row are now computed once per image, and both planes
  of a change are stored together.
- Texture's Gaussian bases ran their column pass down strided columns,
  one tap at a time with a mirror per tap. Both passes now accumulate
  neighbouring samples side by side, in the same tap order.
- The box mean's column pass runs row-wise the same way, blocks in
  parallel.
- The luminance reduced for the work copy and for Texture's copy is
  worked out once per pixel instead of twice, and the work copy is not
  reduced at all when only Texture is set and there is no black point.
- log2 of the luminance is taken once for both changes.

Unit tests keep the replaced code and compare the bits.

Part of #9435.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:49:45 +01:00
24ef10df4d Integration: no crop menu over a live crop drag, an Enter test (#6620)
A right-click while the left button still dragged a crop frame or a
Straighten line opened the crop menu over the live gesture (or, outside
the image, the pasteboard menu). The canvas now opens no menu with the
Crop tool while a crop drag or Straighten line is under way and the
primary button is still down. It needs both: egui ends the drag on the
right button's release without an Up, so the drag flag can stay stale,
and a lost release (Alt+Tab, a cancelled touch) leaves the button
"down". A finger held still (egui's press-and-hold right-click) is
excepted. The agent path (ui.pointer) opens no menu while its own crop
or Straighten drag is down, also the first one on the default frame.
Tests: no_crop_menu_while_the_left_button_drags_the_frame (real pointer
events), a_left_button_down_without_a_crop_drag_keeps_the_crop_menu,
no_crop_menu_while_the_left_button_draws_a_straighten_line,
a_finger_held_on_a_pending_crop_opens_the_crop_menu,
agent_right_click_during_its_crop_drag_opens_no_menu, and
enter_under_the_crop_menu_does_not_crop (only Esc was tested). The
parity row notes Photoshop's Reset Crop and Rotate Crop Box, both still
missing.

Co-authored-by: Marco Arnulfo <103679419+marcoarnulfo@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:48:29 +01:00
David Mann fe07776735 Merge PR #8179: Crop: Content-Aware Fill keeps the pixels a non-deleting crop hid (#8060) 2026-10-25 14:47:33 +01:00
Brandon ThomasandClaude Opus 5.5 b1794c6e06 Release 0.16.0
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
v0.16.0
2026-10-25 09:45:03 -04:00
highperfocused b68d153f03 test(ui): Shift+Alt+right on Quick Selection opens only the selection menu (#6098) 2026-10-25 14:37:36 +01:00
VoiceLessQ 26c218f722 retouch: wrapped Dodge and Burn tone each pixel once; Spot Healing skips a window with nothing to heal from
Dodge and Burn remember each pixel's coverage so one stroke tones it once. With wrap on, that
memory was keyed by the unwrapped position. A stroke longer than the canvas came back to a pixel
under a new key and toned it again from its toned value: mid grey went to 0.816 instead of 0.651
under Dodge at 60%. Coverage is now kept per canvas pixel. A dab bigger than the canvas is folded
first, and each pixel takes the most coverage any of its copies gets (the copy WorkTiles keeps).

Spot Healing returns early when the stroke covers its whole sampling window, as a wrapped dab
bigger than the canvas does. The fill had nothing to sample from. It left the pixels alone but
reported the whole canvas as damage.

Wrap off, Dodge in bench_live_dab_release (6000x4000, 8-bit, medians of 4 interleaved runs)
stays within 0.8% of before at every brush size.
2026-10-25 11:36:15 -02:00
4a17c64139 Integration: a straight Content-Aware crop fills over hidden pixels (#8179)
Photoshop keeps the pixels an earlier Delete Cropped Pixels off crop hid
only under a turned frame. A straight crop pulled back out over them fills
the whole area past the old canvas from the canvas, with Delete Cropped
Pixels off or on (checked: 64x48 green with a red left strip, crop to x=16
keeping pixels, then a Content-Aware crop to x=-24 w=72: all green). The
PR took the snapshot for straight crops too, so the hidden red came back.
The straight path is main's again; the turned path keeps the snapshot.

extending_crop_keeps_the_hidden_pixels becomes
extending_crop_fills_over_the_hidden_pixels (transparent and white
Background, Delete off and on); it fails with the straight snapshot back.

Co-authored-by: Michael Higgins <55243596+Higgs32584@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:35:33 +01:00
vangel76andClaude Sonnet 5.5 dc791a43de raw: develop the default rendering band by band without the f32 RGB frame
When neither Dehaze, a local tone slider nor a camera black point is in play, no
statistic of the whole frame is needed, so the demosaic output no longer has to be
kept. Every band is demosaiced into a small band buffer and encoded straight into the
u16 output. The f32 RGB frame (12 bytes a pixel, 545 MB at 45 MP) and the pass that
read it back are gone. A Bayer sensor also skips the padded CFA plane: each band
normalizes the rows it reads (with the mirrored borders) into a small window. X-Trans
keeps its plane and green plane, LinearRaw has none. Tone, Dehaze and black points
keep the full-frame path.

The bands are the ones the full-frame path computes, with the same arithmetic per
pixel, so the output is bit-identical (tests compare both paths for Bayer, X-Trans and
LinearRaw, odd sizes, band borders, orientations and 1, 3 and 16 threads).
working_memory now charges what each path really holds.

Fixes #9434

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:35:29 +01:00
David Mann 14946918cd Integration: merge main (#8179) 2026-10-25 14:33:03 +01:00
vangel76andClaude Sonnet 5.5 132af3807b raw: add develop_half, a fused half-size preview develop
develop_half bins every 2x2 block of a Bayer mosaic (3x3 for X-Trans,
2x2 box for LinearRaw and monochrome) and fuses black/white scaling,
white balance with its clip, gain maps, the colour matrix and the gamma
encode into one pass, with no demosaic and no full-size float frame.
The result is a Developed in the same colour space, type and
orientation as develop_sensor's, at half size, so a viewer can show it
first and swap the full develop in later. Tone sliders and the camera
black point are not applied (a warning says so). Part of #9433.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:32:10 +01:00
David MannandClaude Opus 5.5 d0c0b93942 Integration: keep main's note on P28's stale baseline in the seam carving doc (#5236)
Folding #5236's performance notes into docs/seam-carving-performance.md
dropped main's "P28's recorded number is stale" section (the 2026-10-05
baseline commit at 266 s, the loaded machine behind the 954 s). None of
it depends on #8526's code, so it is back, dated and attributed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:31:20 +01:00
David Mann d2f9aeb88d Merge PR #7203: feature: gradient paint to the shape tools, any shape can have a gradient stroke. 2026-10-25 14:29:48 +01:00
David Mann d576113233 Merge PR #3627: Mixer Brush: render mixed paint during the stroke and commit the live result
# Conflicts:
#	crates/engine/src/float_cmds.rs
2026-10-25 14:29:47 +01:00
David Mann 8fe70c95d2 Merge PR #5236: perf(algo): Content-Aware Scale 2x faster, within its P28 budget (#211)
# Conflicts:
#	crates/algo/examples/bench_seam.rs
#	crates/algo/src/seam.rs
#	docs/gaps.md
#	docs/target-app-parity.md
#	perf/budgets.toml
2026-10-25 14:28:59 +01:00
David Mann b28073389c Merge PR #3316: Add intermediate UI scale presets between 100% and 125%
# Conflicts:
#	contributors/people.toml
#	docs/gaps.md
#	docs/scorecard.md
#	docs/ui-parity.md
2026-10-25 14:27:10 +01:00
David Mann 5c51830482 Merge PR #8877: PSD: recognise Photoshop's MTrn merged-transparency marker
# Conflicts:
#	docs/file-format-parity.md
2026-10-25 14:26:41 +01:00
David Mann c8847e84b2 Merge PR #9128: fix(#9098): preserve PSD byte order in layered TIFF text descriptors
# Conflicts:
#	docs/file-format-parity.md
2026-10-25 14:26:21 +01:00
David Mann 927ca567e0 Merge PR #9148: Layered TIFF: write Photoshop's global layer mask block (#9095) 2026-10-25 14:26:10 +01:00
e8894aa9f9 Integration: Preferences › Integrations shows the local model rows only where models run (#1199, #4097)
#1199 hid the whole Integrations section in builds without local models, so its
model rows never drew there. #4097 makes the section visible for Allow Agent
Control and Control Port, and every default build then showed "Local selection
models", "Local models aren't available in this build or session." and two
disabled Download buttons under it. The rows now draw only with the local-ml
feature, like the model choices (LOCAL_ML_ONLY).
integrations_shows_local_model_rows_only_where_models_can_run fails without it.

Co-authored-by: mighypr0grammer <83077551+mighty-programmer@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:23:26 +01:00
David MannandClaude Opus 5.5 5f2a7b9903 Merge PR #1199: Add optional local models for subject matting and object selection
Merged with our fix-up d5139d177 (pr-1199-fixup), which merged main as of 8f09d122d.
Conflicts with main since then and with batch 8:
- photocraft-cli: main split the CLI into clap modules (#7071); the model wiring
  (local_headless, configure_cli_models, batch's shared backend, droplet) is re-applied there.
- Cargo.lock regenerated: crates/ml pins ureq =3.4.2, so the app's ureq goes 3.5.0 -> 3.4.2.
- cutout_cmds: main's subject area (#8235) kept for the classical path beside the model path.
- file_cmds batch: main's per-input closure takes `s` later, so the backend and
  integrations prefs are read before the closure instead of inside it.
- prefs_ui: model-choice filter with #4573's pref_choice_label; Integrations stays visible
  (#4097) with the model choices only where models run; the translation-coverage test main
  deleted (c3aa3f9e5) stays deleted; old section_fields test calls take PreferenceFieldOptions.
- Semantic clashes git did not flag: #4097's Integrations test literals in
  apps/photocraft take ..Default::default() for the new model fields; main's and #4573's
  section_fields test calls pass PreferenceFieldOptions.
- engine lib.rs, prefs.rs, smartselect tests, snapshot example, jobs_ui_tests, people.toml,
  docs and i18n catalogs merged as unions.


Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:23:25 +01:00
David MannandClaude Opus 5.5 9ef4c0a487 Integration: the History panel's snapshot spot uses the document id (#3567, #5940)
Merging #3567 renamed the History panel's `doc` to `doc_id`, while main's #5940
(c7ef3676a) still read `doc.id` there, so photocraft-ui-egui no longer compiled
(E0423 at panels.rs `let spot = (doc.id, ...)`).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:23:12 +01:00
vangel76 c7c5aae4af Merge remote-tracking branch 'origin/perf/9428-raw-encode-stage' into perf/9434-fused-develop 2026-10-25 14:14:18 +01:00
vangel76 0966e14ae0 Merge origin/perf/9427-ahd-demosaic (keep clampf from the R1 fix) 2026-10-25 14:14:18 +01:00
David MannandClaude Opus 5.5 fa4269af98 Automation: refuse to replace a non-image file in the write root (#9439)
`doc.save {"path": "src/main.rs", "format": "png"}` and `doc.render
{"path": "Cargo.toml"}` renamed image bytes over whatever file was there.
`AuthorizedWorkspace::write` now replaces an existing file only when its
extension is a format PhotoCraft writes or its bytes are an image; new
files of any name are still created. docs/mcp.md says what a granted root
allows and recommends a dedicated output folder.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 14:09:13 +01:00
vangel76andClaude Sonnet 5.5 df23873ccf raw: block-wise final encode stage for the default rendering
When neither Dehaze nor a local tone map is active (neutral tone and no
black point; only the Nikon D4 profile sets one), the matrix is applied
into a block of 768 samples and the gamma encode runs over the block:
clamp and square root first (vectorises), then the table with
[value, step] pairs. Same arithmetic, same output; other renderings keep
the old per-sample loop. A test compares the block encode with the
per-sample encode.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:05:24 +01:00
highperfocused 906b44b47d fix(ui): the color HUD picks only after the pointer moves from the press (#6098) 2026-10-25 14:05:02 +01:00
vangel76andClaude Sonnet 5.5 cc7da71b47 raw: faster lossless JPEG scan loop (inline diff, 2-component loop, shared tables)
The predictor 1 row loop called Bits::diff out of line, indexed the
per-component tables with bounds checks, and kept a private copy of
the 48 KiB Huffman lookup pair per component even when the components
share one DHT (CR2 and DNG two-component frames).

- Bits::diff is always inlined.
- Rows with one or two components carry Ra in registers and walk the
  row with chunks of two.
- Components that name the same table share one copy.

The frame size cap, the one-bit-per-sample truncation guard and the
DHT rebuild cap are untouched.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:02:03 +01:00
vangel76andClaude Sonnet 5.5 d9aed18582 raw: AHD demosaic with reused planar scratch, about 1.6x fewer instructions
Same arithmetic in the same order, so the output is bit for bit the same
(the old code stays under cfg(test) as ahd_ref and a test compares them).
Scratch planes are allocated once per group of bands instead of per band
(no per-band zeroing or page faults) and never cleared, since every cell
a stage reads was written earlier in the same band. The layout is planar
RGB/Lab, the candidate pixel is built from values instead of indexed
stores (which stalled store forwarding), the 3x3 homogeneity sum is
separable, and the LabF table holds [value, step] pairs. The 32-row band
is kept, as a larger band costs peak memory.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:01:56 +01:00
vangel76andClaude Sonnet 5.5 e15d68bfbb raw: unpack 10/12/14-bit rows by groups, keep a single strip's buffer
Bit-packed uncompressed raw data was read one sample at a time through
a bit accumulator. Rows of 10, 12 and 14-bit samples are now read a
whole group (4x10, 2x12, 4x14 bits = 5, 3, 7 bytes) at a time; the
row tail still goes through the general reader. A file whose only
segment is the whole image row for row no longer copies it into a
second full-size buffer: the decoded Vec becomes the plane. The
predictor 2 undo works in place instead of on a copy.

The summed-segment charge (#4959) and the per-segment limits are
unchanged. A test compares the grouped reader with the plain bit
reader for 10/12/14 bits, 1 and 3 samples and several widths.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 14:00:24 +01:00
vangel76andClaude Sonnet 5.5 97a719a7a7 raw: a non-finite black level no longer panics the AHD demosaic
A DNG whose BlackLevel is a DOUBLE beyond f32 reads as inf. The periodic
fill then computed (raw - inf) * 0 = NaN, and AHD's f32::clamp with NaN
bounds asserts. The fill now returns (0, 0) for a non-finite black, as
Scale::value does, and AHD uses a compare-and-select clamp that gives the
same result for every non-NaN bound.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 13:58:12 +01:00
vangel76andClaude Sonnet 5.5 ff1d29509e raw: count the saturation histogram per CFA row in parallel bands
lift_clipped looked up the CFA colour of every sample of the active
area (two modulo operations each) in one serial pass, about half of a
DNG decode. Each row now computes the colours of one pattern repeat
once, and row bands are counted in parallel into partial histograms
(at most 16) that are summed. The counts are integers, so the result
does not depend on the worker count. A test compares the new
histogram with the per-sample count for offset Bayer, 6x6 patterns,
inner active areas and 1, 2, 3 and 8 threads.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 13:58:10 +01:00
David Mann 731eae5efb Integration: merge main, resolve conflicts (#6620)
# Conflicts:
#	crates/ui-egui/src/crop_size.rs
#	crates/ui-egui/src/panels.rs
#	docs/context-menu-parity.md
#	docs/gaps.md
2026-10-25 13:57:11 +01:00
highperfocused 05bc75d29a fix(ui): only a frame-changing gesture ends the crop Reset no-op (#6561) 2026-10-25 13:57:09 +01:00
David MannandClaude Opus 5.5 4c66dd45d2 Address CodeRabbit review: the DOCTYPE check is one linear pass
Each <!DOCTYPE started its own scan to the next unquoted '>' or '[', so a
comment full of them cost quadratic time. Scans in the same quote state
go on alike, so one pass with three flags finds the same answer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:50:54 +01:00
05b9477635 Integration: Mixer Brush tips above 1024 px, Alt with a held tool, long strokes (#3627)
A UI Mixer Brush drag with a tip of 1025-5000 px was dropped on press: the live
preview refuses tips above 1024 px, and the canvas cleared the drag on any begin
error, though the status said the stroke would render on release (main painted
it). The drag is now kept when only the preview size refuses it
(`LiveMixer::defers_to_release`), and the release renders it; a stroke that
can't paint at all (a locked layer) is still refused on press.

Alt sampling tested `app.ui.tool`, so a held temporary tool (macOS Cmd = Move)
with Alt loaded paint; it now uses `active_tool()`, like `alt_eyedropper`.

The 65,536-sample and 250,000-dab caps also applied to the release replay and
the `paint.mixerBrush` command, so a long stroke (or Stroke Path with Simulate
Pressure, up to 200,000 samples) failed entirely; main had no cap. They now
bound the live preview only, which falls back to the replay.

Also: `cargo fmt` on the merged `LiveCommit` literal, and `as_chunks` in a test
for clippy's `chunks_exact_to_as_chunks`.

Co-authored-by: durtyrobbin <76844303+durtyrobbin@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:49:48 +01:00
Hector VillarrealandClaude Opus 5.5 295d739508 fix(ui): Select and Mask starts from its defaults unless Remember Settings is on (#9423)
Select and Mask reopened with the last settings used (Radius, Smooth,
Feather, ...), like the filter dialogs, so each use started from the
previous refinement instead of a neutral mask. Photoshop's Select and
Mask has a Remember Settings checkbox, off by default, and without it
reopens at its defaults.

The dialog gains Remember Settings (off by default). OK with it off
remembers only that choice, so the next Select and Mask starts from the
defaults; with it on, the settings carry over as before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 06:48:16 -06:00
highperfocused 8929175c08 fix(cms): reject malformed Mesh declarations and honour the declared range (#9078)
A `Mesh` line is now a declaration only when it carries exactly two integer
fields after the keyword; tokens like `Mesh invalid 1 12` no longer recover a
declaration from numbers in the wrong positions and fall back to headerless
inference. Under a valid declaration the node codes are integers in the declared
output range, so the declared range is used for every in-range value — including
a small max, where code 1 is 1/declared of full output, not full output itself
(#9073). Only a node beyond the declared range still falls back to max-value
inference.
2026-10-25 13:44:30 +01:00
e13a984042 Integration: refuse non-array stop lists, keep CMYK/Lab stops, stroke width takes the gradient (#7203)
- parse_fill took a present non-array "stops" / "opacityStops" / "midpoints"
  as absent: {"gradient":{"stops":"red"}} silently became black to white. It is
  an error now (CodeRabbit's finding); the hostile-values test also checks that a
  rejected shape.create leaves the document and history alone.
- The swatches sent every gradient stop as sRGB RGBA, so in a CMYK or Lab
  document ticking Reverse or changing the angle turned every stop into an RGB
  colour (written back to PSD as RGBC). Stops now go out as stored Colors, which
  shape.create / shape.edit accept, finite with alpha in 0..1.
- Options bar, strokeless shape selected, stroke gradient chosen: the width
  field and the stroke type added a solid background-colour stroke. They take
  the stroke gradient now, as the Fill checkbox and the next shape do.

Co-authored-by: Denis <179751626+deniskhud@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:37:56 +01:00