Reset All Tools cleared every tool's remembered brush, and a tool
without one takes the live brush when it is picked. With the Brush's
starting tip now differing from the other tools', that leaked: Reset All
from the Brush left the Pencil and Eraser on a 13 px soft tip, and from
another tool it left the Brush on a 20 px hard one. Reset All now gives
each painting tool other than the live one its own starting brush.
Also names the Brush tool's reset tip and updates tool_reset's docs.
Reset Tool and Reset All Tools gave the Brush tool the default 20 px
hard tip at 10 % spacing. The reference resets it to a 13 px soft round
(hardness 0 %, spacing 25 %). `fresh_brush` now takes the tool and
gives the Brush that tip; the other painting tools and
`BrushSettings::default()` keep today's tip.
A row recorded from the Count tool was labelled "Measurement N" like a
selection row. It is now "Count N", as in the reference application,
with the same shared number; selection and ruler rows keep their labels
and every other column is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
count.deleteGroup kept the same index current after deleting the
current group, so with three groups and the second current the third
became current. The first remaining group now becomes current, as the
reference application does. Deleting another group by index still keeps
the current one (#5654), and deleting the last group still leaves none.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Escape closed the picker like Enter, after the frame had already sent a
size or hardness typed in it, so the typed values stayed. The picker now
remembers the brush it opened with (or last picked a preset into), and
Escape goes back to it. Enter still applies the typed values, Tab keeps
the picker open, a preset picked before Escape stays picked, and Escape
that ends a rename still only ends the rename.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Alt-click (Subtract) and Shift+Alt-click (Intersect) with the Magic Wand
took the seed's whole region, wherever the seed was, so a click outside
the selection still cut into it. With Contiguous on, the region now
grows from the seed through selected pixels only: a seed outside the
selection subtracts nothing and intersects to nothing, and a seed
inside gives the same results as before. Contiguous off is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
BrushSettings::default() turned Pressure for Size on, so a fresh brush -
and the brush Reset Tool hands back, which is that default plus
Photoshop's 10 % smoothing - reported pressureSize: true from brush.get
and drew the options bar's "Always use pressure for size" button
highlighted, while the reference application leaves both pressure
toggles off after Reset Tool. Neither is on now.
The toggle itself is untouched: it still switches both ways, Shape
Dynamics' Pen Pressure control is unaffected, and path.stroke's Simulate
Pressure still tapers a brush that follows pen pressure. A brush JSON
that omits pressureSize now reads as off, as the reference's default is.
Two engine tests took their expectation from the old default and set
pressureSize explicitly instead: cmd::brush_cmds::tests::stroke_backwards_compatible
and cmd::vector_cmds::tests::stroke_path_simulate_pressure_tapers_both_ends.
The new crates/engine/src/brush_pressure_size_default_tests.rs covers the
default, brush.get on a fresh session, an omitted field, switching the
toggle, and that a pressure ramp with the default brush paints like
pressureSize: false.
Fixes#11809
With Show > Count off, a click added the marker but left it invisible,
so the click looked like it did nothing. It now switches the flag on,
without adding a history state. Alt+click and hidden groups are
unchanged.
Fixes#11770
With the Note tool and a note selected, Delete and Backspace went the
usual way and deleted the active layer, leaving the note. They now ask
"Delete this note?" (Yes / No, "Don't show again"); Yes deletes the note
as one "Delete Note" step and selects nothing, No changes nothing. The
Notes panel's trash asks the same, and "Don't show again" skips the
question for both.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review finding on this PR (#11675, discussion_r4499346688): with the warp now cleared as part of
the reset, a smart object whose stack holds a visible filter this build cannot apply took the
Free Transform path's forgiving branch - `refresh_layer` returns false for that stack,
`refresh_text` drops it on the floor, and `restore_unrendered` put the OLD cache back warped by
the relative map. The step recorded a placement at 100 % with no warp whose pixels still showed
the warp.
The step now insists on the guarded refresh the other smart-object edits use:
- `unapplied_filters` is checked before anything changes, so an unsupported Photoshop filter kept
verbatim from a PSD names itself in the error and the document is never touched.
- `refresh_or_fail(doc, id)?` runs at the reset boundary, right after `transform_by_map` and
before the exact map is pinned. It is a cache-warm call in the normal case (same document state
as the render `transform_by_map` just did), and it refuses an unreadable source or a render that
fails outright. Because `Session::edit` runs its closure on a copy and returns early on `Err`,
a refused reset leaves the document, the placement, the warp and the history list exactly as
they were - the fallback pixels are never accepted.
Regression test `reset_transform_refuses_a_stack_it_cannot_re_render` (red before this commit:
the reset returned Ok) asserts the error names the filter, that transform, warp and filter stack
are unchanged, that no history state was added, and that the same object resets normally once the
unsupported filter is off the stack. The four existing reset tests still pass, so the success
path and its warp clearing are unchanged.
No new dependency; photocraft_io's UNSUPPORTED_FILTER was already re-exported by this module.
Stroke Emboss was drawn as an Inner Bevel of the layer, shading the
layer's own edge. It now bevels the outline of the layer with its top
Stroke effect and shades only the stroke, as Photoshop does; without a
stroke it shows nothing. The GPU plan hands a layer with Stroke Emboss
to the CPU compositor, which builds the stroke's band.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Adds my own entry to contributors/people.toml, as AGENTS.md asks a
contributor to do in their own PR. sync_github_name = true takes the
display name from my public GitHub profile; the About window drops it
until the maintainer's next contributors.py run marks the entry
verified.
The merge of main (d453c7d12) moved the Magnetic Lasso's pointer-only
hover block inside `if buttons.clicked`. `canvas_buttons` sets `clicked`
from `response.clicked() || erase_click`, so on a pointer-only frame it
is false and the block never ran: after the first fastening click, live
edge tracking and automatic fastening froze until the next click. The
base ran this block at a top-level `if`.
Hoist it back above the click-only branch. A menu-dismissal gesture is
still inert: the block is gated on `menu_dismiss_press`, the same state
that starves the rest of the input pass and that the reviewer asked the
dismissal suppression to use.
Regressions, verified by reverting only canvas.rs:
- `on_the_canvas_the_border_follows_the_hovering_pointer_and_keys_work`
and `on_the_canvas_a_double_click_closes_the_border` pass on the merge
base d453c7d12 and fail on the merged head (FredApps [P1], #9841).
- new `on_the_canvas_hover_moves_survive_a_menu_dismissal_and_resume`
covers the dismissal interaction: the hover is suppressed while the
dismissal press is held and resumes on the next pointer move.
The Brush, Pencil and Eraser Opacity fields take whole numbers from 1 to
100, as in the reference application (0 and -10 become 1, 150 becomes 100,
33.7 a whole number). The digit keys "00" set the Brush to 100 % instead of
0 %, and tools.setBrush floors opacity at 0.01, so a brush can no longer be
set to paint nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- A complete import makes a fixed number of cancellation checks; cancelling on the last per-object check and on the public boundary check must both return Cancelled and never a document.
- Six page variants (plain, CropBox offset, rotations 90/180/270, oversized CropBox) compare each stored page raster with a direct PDFium render at 72, 150 and 300 DPI: 18/18 byte-identical, and the sizes prove the crop intersection (190x80) and oversized CropBox (200x100) cases.
- Both tests skip without PDFium and fail under PHOTOCRAFT_REQUIRE_PDFIUM, like the existing end-to-end test.
The ui-parity row, its revision entry and the status line said the Warp survives Reset Transform.
It does not: measured against the reference application 2026-11-06 (report in #11675), the warp is
cleared with the rest of the transform, and the object centres on the average of the placement's
four corners. A 2026-11-07 revision row records the correction; the 2026-11-06 entry stays as the
history of what landed when.
cargo xtask parity and scorecard re-run: unchanged (661 of 661 live), so nothing to regenerate.
Measured against the reference application on 2026-11-06 (report in #11675): a smart object
warped with Arc 50, then rotated 20 degrees and scaled to 80 %, came back from Reset Transform
with no warp at all, at its contents' own size, unrotated, centred on the average of its four
corners. So the Warp goes with the transform, and the centre is the corner average rather than
the point the contents' box centre maps to - the same point for an affine placement, and the
quad's centre once Distort or Perspective bent it.
The warp is cleared before the placement is re-rendered, so the step draws the unwarped
contents once. What still follows the placement is the Smart Filter mask and any linked layer or
vector mask, exactly as under Free Transform: how a painted filter mask should compensate for a
cleared warp is not something the measurement covers, and #11171 has its own answer to it.
Smart filters, sources and history are unchanged: still one "Reset Transform" undo step. The
tests now pin the cleared warp (red before this commit), the corner average for a distorted
placement, and that a second reset changes nothing.
The merge of main into #10730 dropped the first line of the comment above the
filter-preview check in confirm_generic_command; it is back. The Batch test now
compares out/b.png with the hand-typed reference as well as out/a.png, so a
replay that stops after the first file fails it (review on #10730).
Co-Authored-By: Lcoding15 <Lcoding15@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Remove the Azerbaijani menu coverage test (red on main), the crop Fill
label coverage test and the has() loop in the pen path Esc test. A
missing string falls back to English; cargo xtask i18n-coverage lists
the gaps. docs/contributing.md now says tests never require a string to
be translated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>