Commit Graph
63 Commits
Author SHA1 Message Date
FredApps c93c0bb36d Merge main into trimmed integration; reconcile blur, pattern and Liquify behavior 2026-10-29 23:47:53 +01:00
FredApps ce77d3def1 Merge approved PhotoCraft PR #8931: feat(algo): describe the Gaussian blur as data for other devices 2026-10-29 03:29:42 +01:00
FredApps 07ec24727a Merge approved PhotoCraft PR #7903: Filter Gallery › Cutout 4/4: trace the image into flat paper shapes 2026-10-29 03:29:36 +01:00
FredApps de85ba7b07 Merge approved PhotoCraft PR #8191: Filter Gallery › Cutout 1/4: colour and palette stage 2026-10-29 03:29:34 +01:00
Finn NewickandClaude Opus 5.5 d42dd4c630 Merge origin/main into feat/operator-separable
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-27 22:51:51 +00:00
aiongg 2be8a2e5a8 refactor(algo): move tests_selection_mix.rs to tests/selection_mix.rs 2026-10-26 22:30:15 +00:00
aiongg dd784da5cc refactor(algo): move tests_mosaic.rs to tests/mosaic.rs 2026-10-26 22:30:15 +00:00
aiongg 1559d287d6 refactor(algo): move tests_ext.rs to tests/ext.rs 2026-10-26 22:30:15 +00:00
David MannandClaude Opus 5.5 d73c91133b Integration: merge main, resolve conflicts (#1610)
Translation catalogs keep main's strings plus the Art History Brush ones
(fr: main already has Area; hi: added the brush strings). docs/scorecard.md
regenerated.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 10:19:42 +01:00
Ayan 47d9266fa1 Make Crystallize 4x and Pointillize 3x faster on 24 MP layers (#6200)
* Speed up Crystallize and Pointillize with cell-space rasterization

* Measure pixelate commands on unlocked layers and Background separately

* Record pixel-exact 24 MP filter and command benchmarks
2026-10-24 11:13:03 -04:00
aiongg 7aaa59712a Merge main into integration-batch-7 2026-10-23 20:37:05 +00:00
Dylan M. Taylor ecd1a692a3 security(algo): bound working sets sized by caller-controlled extents (#8946)
* security(algo): bound working sets sized by caller-controlled extents

Zero-width inpaint grids, content-aware fill mosaics of turned thin images,
Liquify fields over huge canvases, tiny-scale reduction weights, warp tile
bookkeeping and per-tile source footprints, and windowed filters whose halo
follows an unclamped radius or a stacked Filter Gallery reach all asked for
memory sized by file or parameter input. Each is now bounded; halo
arithmetic saturates instead of overflowing.

* security(algo): budget variants by dimensions, size-check footprints first, widen the tile-halo maths

CodeRabbit follow-up:
- fill_cow settles the mosaic budget on variant dimensions only, so a
  variant's image and mask buffers are not built when the mosaic would
  exceed MAX_MOSAIC_PIXELS (complete falls back to the image itself).
- warp validates the footprint's pixel area against MAX_FOOTPRINT_PIXELS
  before src.has_tiles_in, so an oversized footprint returns without the
  tile-grid scan.
- auto_tile widens the input-derived halo arithmetic to i64 and clamps
  before the cast, so an i32::MAX radius cannot overflow in debug builds
  and the 2048 tile cap stands.

* fix(algo): drop the duplicate mosaic_size call left behind in fill_cow

The variant-budget rework kept both the new dimensions-based budget and
the old variants-based one; the old call no longer compiled after
mosaic_size started taking dimensions. Keep the comment that explains
the fallback.
2026-10-23 15:22:24 -04:00
Finn NewickandClaude Opus 5.5 603c5ac5b0 feat(algo): describe the Gaussian blur as data for other devices
Separable (an exact kernel, or box widths) is the blur Gaussian Blur
runs, taken from blur.rs's own functions; the CPU Gaussian is now built
from it, so a GPU operator running the same description can't drift.
OperatorRegion says which source pixels an output area reads, and
edge_extent is the edge-repeat extent the engine already used for layer
filters (now shared). Tests pin the reach to the CPU's real output and
check an output depends only on its input region.

Part of #5772 (Phase 1, step 1.1).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 18:24:15 +01:00
David MannandClaude Opus 5.5 7a54fa9c77 Integration: merge main into batch 6
Conflicts, both sides kept: main's .act swatches next to the batch's .asl
styles (Open filters, preset extensions), main's Render smart filters next
to the batch's Distort ones (53 mapped), parity doc status lines.
docs/scorecard.md regenerated.

Follow-ups the merge needed: the Distort smart filter readers pass their
enum type to main's stricter enum_name (the type their writers use), and
the CMYK effect tests set main's new StrokeFx overprint and Bevel
gloss_anti_alias.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-22 12:09:25 +02:00
Matt Van Horn 4571843c27 Add Art History Brush to the Y-group flyout
paint.artHistoryBrush stamps oriented dabs sampled from a history
state. Styles cover Tight/Loose Short, Medium, Long, Dab, Tight Curl
and Loose Curl. Area sets spacing; Tolerance skips dabs whose
footprint already matches the source. Scatter is seeded so a stroke
replays bit-for-bit.

The Y flyout lists History Brush then Art History Brush. The options
bar exposes Style, Area and Tolerance. Includes a 24 MP release bench.
2026-10-21 22:48:10 -07:00
Andee Collard 347991f890 Merge remote-tracking branch 'origin/main' into gallery-cutout-trace
# Conflicts:
#	crates/algo/src/artistic_fx.rs
2026-10-22 04:58:19 +01:00
David MannandClaude Opus 5.5 1ff7487333 Integration: merge main, resolve conflicts (#5249)
docs/file-format-parity.md: kept main's status line and PSD row, re-applied
the PR's smart filter additions.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-21 14:00:56 +02:00
Andee CollardandClaude Opus 5.5 1d09845cc8 Cutout 4/4: fill stage and Filter Gallery hookup
Last piece of the Cutout trace: Filter Gallery › Cutout now traces the
image into flat paper shapes (stages in the earlier three commits).

- `cutout::fill`: exact-area scan conversion, so edges are anti-aliased
  and shapes sharing an edge sum to full coverage (no seams)
- `cutout::trace` runs the stages; `Traced::render` fills any tile
- `artistic_fx`: a stack containing a Cutout declares `Halo::Bounds` and
  is prepared once (`artistic_fx::Prepared`); tiles only fill it in
- the module's `allow(dead_code)` goes now that it is used

Output is unchanged from the single-file version in #7903 (render hashes
over 54 input/setting combinations match).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-21 08:18:10 +01:00
Andee CollardandClaude Opus 5.5 91e4cec435 Cutout 1/4: colour and palette stage
First piece of the Filter Gallery Cutout trace, split out of #7903 so each
stage can be reviewed on its own. Nothing calls it yet.

- new `cutout` module: the filter's settings, and the Oklab conversion,
  bilateral smoothing, palette (median cut + k-means, with a tone of two
  clearly different hues split in two and noise-apart colours merged) and
  nearest-colour labelling
- `cutout::util`: the transposes, union-find `find` and `FxHasher` the
  later stages share
- `quantize`: `median_cut` and `kmeans` are crate-visible

Tests: Oklab round trip and gamut, out-of-gamut colours keep their hue,
smoothing flattens noise but keeps a step, the palette finds the image's
own colours and splits a two-hue tone, transposes and `find`.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-21 08:10:29 +01:00
Matthias Schlich d104f7015f fix(filters): Clouds and Fibers move with their smart object
Their noise was sampled in document coordinates, so moving a smart object re-rendered a
different texture with the same params. The pattern is now anchored at the filter's bounds:
the object for a smart filter, the canvas on a layer. Fibers, like Clouds, takes the canvas
as its bounds even with a selection, which only masks it, so layer results stay as before
(CodeRabbit review on #6160).
2026-10-20 22:49:04 +02:00
Matthias Schlich 0a7325af6e Merge origin/main into feat/psd-render-smart-filters
# Conflicts:
#	docs/file-format-parity.md
2026-10-20 22:20:37 +02:00
David MannandClaude Opus 5.5 aa8621788e Minimum / Maximum: CMYK spreads paper white with Maximum, ink with Minimum (#7074) (#7075)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-19 13:04:02 -04:00
Korg1ch f3fb5781ef Lens Blur: an alpha channel as the depth map source, as in Photoshop (#7400) (#7508)
* fix(filters): Lens Blur reads its depth from an alpha channel, listed in Depth Map › Source as in Photoshop (#7400)

* test(filters): Lens Blur depth from an alpha channel by name or index, its errors, and the Source menu (#7400)
2026-10-18 17:23:32 -04:00
JGF b48a916b13 fix(algo): normalize nonpositive filter tile sizes (#7554) 2026-10-18 16:57:08 -04:00
abc2cb 978410ea66 Merge remote-tracking branch 'upstream/main' into feat/psd-distort-invert-smart-filters 2026-10-18 17:02:23 +03:00
abc2cbandClaude Opus 5.5 85760b1c36 PSD Wave smart filters: keep more than 64 generators verbatim
The Wave renderer clamps the generator count to 64 (the command allows up
to 999), but a Photoshop Wave with Scale 100% mapped to filter.distort.wave
whatever its Number of Generators. A file with 332 generators opened as a
supported, editable Wave and, once its cached pixels were re-rendered,
drew 64.

Treat it like a scaled Wave: a Wave with fewer than 1 or more than 64
generators stays Photoshop's, descriptor kept verbatim and written back
unchanged. The limit is photocraft_algo::MAX_WAVE_GENERATORS, which the
renderer now clamps to as well, so the two cannot drift.

Tests: the 332-generator Wave stays unsupported and round-trips its
descriptor; a new boundary test maps 1 and 64 generators and keeps 65, 999,
0 and -1 verbatim.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-17 16:24:48 +03:00
Matthias Schlich 875ec69fce fix(filters): Clouds size their pattern by the canvas, not the selection
Since Clouds became a filter, its cloud size came from the filter's reference bounds, which are
the selection's when there is one: clouds in a selection were finer than the canvas's. Before,
they were sized by the document. Clouds now take the canvas as their bounds; the selection only
masks them, as before (review on #6160).
2026-10-17 12:28:10 +02:00
Matthias Schlich c6fff8a3a8 Merge origin/main (up to a1e19a0b) into feat/psd-render-smart-filters
# Conflicts:
#	crates/engine/src/smart_cmds.rs
2026-10-17 12:10:28 +02:00
Gert Pellin - GPE a6a1d2ed22 feat(algo): face detection with libfacedetection's CNN in pure Rust, and image.detectFaces (part of #3603) (#4513)
photocraft_algo::faces::detect finds faces in an RGBA image with a score, a box and five landmarks, through a pure-Rust port of libfacedetection's YuNet network (Shiqi Yu, BSD-3-Clause; licence, NOTICE and ATTRIBUTION rows included). The weights are the library's, converted to half precision (107 KB), parsed defensively. image.detectFaces reports the document's faces to scripts and agents; it is read-only with no history step and no menu entry. The network is compiled out of the browser build, where the command reports itself unavailable. Head outputs are pinned against the C++ reference.

Part of #3603.
2026-10-16 17:31:02 -07:00
Matthias Schlich 50a011ae9e feat(filters): Clouds and Difference Clouds as smart filters; render filters work within a smart object
Clouds and Difference Clouds ran outside the filter pipeline (map_pixels with the session's
colours), so on a smart object they failed instead of adding a smart filter. They are now a
regular render filter (FilterParams::Clouds, same noise as before) that records its colours,
so a smart filter re-renders the same clouds after the colours change.

Render filters on a smart object (Clouds, Difference Clouds, Fibers, Lens Flare) now use the
placed content's bounds, as Photoshop does: a flare's centre is a fraction of the object, not
of the canvas.
2026-10-15 16:05:34 +02:00
Gert Pellin - GPE 4bbcede572 Fibers: fill an empty layer, as Photoshop does (#5241) (#5289)
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.
2026-10-14 01:59:00 -04:00
optimalbyteeandoptimalbytee 50c4f72af7 Speed up large-radius Unsharp Mask with guarded FFT convolution (#2987)
* Speed up large-radius Unsharp Mask with guarded FFT convolution

* Address Unsharp Mask review feedback

---------

Co-authored-by: optimalbytee <optimalbytee@users.noreply.github.com>
2026-10-13 13:23:55 -04:00
ragtuxandClaude Fable 5.1 995bc89c40 Auto-Blend Stack Images: a depth map by depth from focus, for Lens Blur (#2073)
* 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 <noreply@anthropic.com>

* 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 <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-13 13:23:35 -04:00
29f3656fe6 perf(algo): filters read the selection once per tile when mixing it back (#4533) (#4754)
mix_selection looked up the selection one pixel at a time
(`sel.pixel(x, y)`, which allocates a Vec and walks the tile map) for
every output pixel of every tile of a filter applied through a
selection: about 24 million allocations and map lookups at 24 MP.

It now reads the selection over the tile once (`read_region`) and
indexes it. The result is bit-identical: the old per-pixel version is
the test oracle, with fully selected, unselected and partial pixels,
with and without alpha, on a tile crossing the selection's tiles.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
2026-10-13 02:47:53 -04:00
David MannandClaude Opus 5.5 95663451c0 Find Edges and Solarize: treat CMYK values as ink (#4222) (#4223)
CMYK samples are ink (1 = full ink), but both filters assumed light
polarity. Find Edges wrote 1 - edge, so flat CMYK areas became 100% ink
(black) instead of white; Solarize inverted channels above 50%, i.e. the
shadows. Both now use fxutil::subtractive(ctx), like Trace Contour: Find
Edges writes the edge strength as ink, Solarize inverts channels below
50% ink. Alpha stays untouched. RGB, Grayscale, Lab and other modes are
unchanged.

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-12 21:17:54 -04:00
a58b6f3d9e Mix filtered selection edges with premultiplied alpha (#3895)
* Mix filtered selection edges with premultiplied alpha

Co-authored-by: Codex <noreply@openai.com>

* Update stylize kernel comparison for alpha-aware selection mixing

Co-authored-by: Codex <noreply@openai.com>

---------

Co-authored-by: jonathnwxh <258340642+jonathanwxh-cell@users.noreply.github.com>
Co-authored-by: Codex <noreply@openai.com>
2026-10-12 09:52:59 -04:00
UnemarsandClaude Opus 5.5 6ecad20132 Content-Aware tools: faster, cancellable at once, same pixels (P21, P22, P49) (#3395)
* Content-Aware tools: faster, cancellable at once, same pixels

#2202 matched Photoshop's Content-Aware Fill but made the perf nightly slower (#2575): P21
Content-Aware Fill 10 MP 5x, P22 Spot Healing 2.4x, P49 cancel on 24 MP 4x and over its 200 ms
budget. This keeps every result bit for bit and takes the time back:

- The solver works over the hole's bounds grown by a patch: match field, costs, votes and the
  per-target tables are kept there (nothing outside changes or is matched), the vote computes
  each target's weight once and writes into a buffer instead of a Vec per pixel.
- Colour adaptation reuses each patch's own moments (target sums kept per EM step, source sums
  per level where they pay off), summed in the same order as before.
- Exact early outs: SSD per pixel instead of per row (callers only ask "better?"), and the onion
  peel's exhaustive search starts from where the filled neighbours took their sources and stops a
  source once it can only lose (first-of-equals kept).
- PatchMatch sweeps a band in wavefronts of blocks when there are fewer bands than threads: each
  cell sees the same neighbours and random numbers (SplitMix64 skip) as in raster order.
- Parallel pyramid, valid-source and target setup; no copy of the window when there are no
  variants (fill_owned / complete_masked_owned); the job reads the window in parallel bands and
  keeps only the selection's bounds for the apply step.
- Cancellation is checked per row everywhere, also in Auto sampling and in its max flow
  (maxflow_with / grid_cut_cancellable): a 24 MP job now stops in tens of milliseconds.
- algo::par: rayon on native, sequential on wasm.

Proof: the new `content_aware_bench` example hashes every tool's output (Fill with each sampling
and adaptation option, Edit > Fill, Delete and Fill, Spot Healing x3, Patch, Content-Aware Move,
8/16/32-bit, Gray/CMYK/Lab, and the benchmarks' 10 and 24 MP fills): 117 hashes identical to
main, also with 1, 2 and 4 threads. New tests compare the onion peel with a plain scan, full and
clipped patch distances and moments, the valid-source runs with the summed-area test, thread
counts 1/3/16, in-place and copied fills, and cancellation of Auto sampling and max flow.

perf_scenarios --cpu, Ryzen 7 5800X shared with other work (load 4-7), 3 interleaved rounds:
P21 p50 295-306 -> 167-171 ms (max 342-362 -> 180-184); P22 p50 67-68 -> 38-40 ms (max 69-71 ->
40-41); P49 max 378-389 -> 54-77 ms (p50 94-110 -> 47). retouch_bench: Content-Aware Move
-47..-64 %, Spot Healing Content-Aware -40..-61 %.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* Content-Aware: faster Spot Healing dabs than main with #2805, same pixels

Against main with #2805 the branch was slower for small Spot Healing dabs (size 50:
+25-30 %, single-threaded +60 %): its onion peel walked all 49 offsets of every candidate patch,
checking each against the known pixels, and had no bound at the hole's edge.

- Onion peel: the target's known pixels are gathered once (#2805), the search starts from the
  filled neighbours' shifted sources or every 16th source at the hole's edge (#2805), the limit
  a candidate must stay under is a precomputed sum (no division per pixel), and candidates are
  summed four at a time in step: their sums are independent, so the processor overlaps them.
  Each is still given up only against the best from before the four, which can only improve.
- Vote: each target's weight and match centre are taken once; a pixel's patches are visited
  over the clipped offset range, with the channel count and the colour adaptation fixed at
  compile time. Same sums, same order.
- Valid sources: sliding counts per row and stacked runs per column instead of a 7-row check
  per pixel; the k-th source is found through per-row and per-64-column counts, so no level
  but the coarsest lists them all.
- `integral` stays out of line: inlined into `synthesize` it made Create Texture slower.

Same results: `content_aware_bench -- hash` (112 cases) and `-- hash-big` (the benchmarks' 10 and
24 MP fills) equal main's, also with 1 and 3 threads; the onion peel tests (#2805's and ours)
compare with the plain scan. `content_aware_bench bench spot` runs retouch_bench's Spot Healing
dabs; with `CAF_SPOT_TYPE` (comma-separated) other types, with page faults per dab.

Ryzen 7 5800X, shared (load 5-20), interleaved with main (3af4c46a), min of rounds,
retouch_bench, all threads / one thread:
Spot Healing Content-Aware 50: 49 -> 31 / 204 -> 117 ms; 100: 62 -> 35 / 164 -> 98 ms;
200: 118 -> 67 / 250 -> 158 ms; 300: 212 -> 119 / 411 -> 285 ms. Content-Aware Move -47..-62 %
(one thread -23..-45 %). Rows that don't run this code stay within noise, except Proximity
Match 200 after the Content-Aware dabs in the same process (+8..23 %): it takes ~5800 page
faults there against ~300 on main, because main's heavier dabs left more of the heap faulted
in; alone, or after Content-Aware dabs in `bench spot`, it is equal.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-12 00:11:10 -04:00
ragtuxandClaude Fable 5.1 ec60f43b5b Auto-Align Layers: register focus and exposure brackets by intensity when there is nothing to match (#2065)
* Auto-Align Layers: register focus and exposure brackets by intensity when there is nothing to match

Feature matching fails outright on layers that nearly coincide (a focus bracket: sub-pixel shifts, up to a few percent of scale from focus breathing, much of every frame defocused). `photocraft_algo::direct` registers the target to the reference coarse-to-fine on a Gaussian pyramid of the luminance, the cost the mean-removed RMS of the warped difference over the pixels both cover, by a bounded Nelder-Mead search with a per-level schedule; the pose is a similarity about the centre, affine or projective per the projection, and resolution independent, so the search stops short of full resolution. `edit.autoAlignLayers` gets `registration`: auto (features, then intensity when nothing matches), features, intensity; layers register outward from the reference, each to the previous one aligned.

Closes #2064.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* Intensity registration: reject a no-fit, search on coarsened planes

Review follow-up on #2065:
- `direct::cost` also returns the correlation of the reference with the
  warped target over the overlap; `Registration.correlation` carries it.
  Unlike the mean-removed RMS it is exposure independent, so it tells a
  bracket (near 1) from unrelated pictures (near 0).
- `register_direct` fails with the existing "couldn't find enough matching
  detail" error when a layer's best pose overlaps its neighbour by less than
  a quarter of the plane or correlates below 0.5, instead of placing it at
  whatever pose the search box ended at.
- The luminance planes are built at the search scale (`luma_alpha` with the
  reduction factor, ≤ 2048 px across) and the warp of each registered layer
  runs there too; the pose is resolution independent and its matrix at full
  resolution places the layer. Nothing full-resolution is held.
- The reference index is checked once with `get`; no unclamped indexing.
- Tests: correlation at an exposure step, for unrelated textures and a flat
  plane; the engine rejects unrelated layers and a mostly empty one.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-11 04:02:26 -04:00
JohnnYoruandBrandon Thomas 897f9f37ad Add the Remove tool with non-local patch completion (#2193)
* Add the Remove tool with non-local patch completion

Tool::Remove (J flyout) paints a hole that paint.remove fills from its
surroundings as a background job, in one undo step. A stroke around an
object also removes what it encloses; the selection limits the hole.

The fill (photocraft_algo::nonlocal) follows Newson, Almansa, Gousseau and
Pérez (IPOL 2017): texture features in the patch distance, onion-peel
initialisation, match fields upsampled between levels and a best-patch
final copy, plus gradient-domain hiding of edge and interior seams.
Sparse holes are filled in tiles. inpaint::complete (Content-Aware Fill,
Spot Healing, Content-Aware Move) is unchanged.

On 20 holes in public-domain photos (remove_quality) the fill keeps 0.89
of the texture it replaces, against 0.54 with the Wexler/PatchMatch
completion. 24 MP (remove_bench): 100 px ring 120 ms, 1000 px ring 2.5 s.

* Translate the Remove tool's label and tighten comments

Adds "Remove Tool" to every UI catalog, following each one's pattern for
tool names and its existing word for Remove. Corrects comments that no
longer matched the code: the benchmark's description, the tile and margin
minimums, and the reasons given for two parameters.

* Put the Remove tool first in the J group

The J flyout's slot now shows Remove until another J tool is picked, and
J from another group selects it; repeated J goes on to Spot Healing Brush,
Healing Brush, Patch, Content-Aware Move and Red Eye.

---------

Co-authored-by: Brandon Thomas <echelon@gmail.com>
2026-10-10 05:19:11 -04:00
optimalbyteeandClaude Opus 5.5 6ae2a8498c Optimize Radial Blur sampling without changing output (#1499)
Radial Blur prepares its sample transforms once per filter (blur::radial::Plan) and shares them across output tiles, reading each bilinear tap once with channels kept together. Output is identical to the previous sampler at Draft, Good and Best, across colour models and depths (tests compare against the unchanged reference path); a failed table allocation falls back to that sampler. Measured on 24 MP RGBA8, Good, amount 1: Spin 66.1 -> 16.4 s, Zoom 19.2 -> 5.4 s (bench_radial example, docs/radial-blur-performance.md).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 14:41:52 -04:00
Gert Pellin - GPEandClaude Opus 5.5 15737e122b Filter > Other > Color to Alpha (#1683)
A new filter that removes a chosen colour by unmixing it into transparency, as GIMP's and
Krita's Color to Alpha do: per pixel the most transparent opacity that keeps the unmixed colour
in range, shaped by transparency and opacity thresholds. It runs through run_filter (selection,
smart filters, Last Filter, one history step), floats the Background first and refuses a
transparency-locked layer; the filter dialog gains a colour parameter. Fixes #1576.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 10:13:42 -04:00
optimalbyteeandClaude Opus 5.5 f5a86628ed Speed up exact Shape Blur convolution (#1566)
Shape Blur now builds its span kernel once per operation instead of per tile. It folds
premultiplication into the f64 row prefix sums and adds whole output rows at a time, in the
same order as before. Output is bit-identical to the previous implementation, which stays in
the tests as an exact oracle. A 24 MP Circle r25 runs in 0.65 s instead of 1.04 s.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 01:19:09 -04:00
optimalbyteeandoptimalbytee c7dbb37b3c Speed up wide Motion Blur with adaptive FFT convolution (#1471)
Keep the merged upstream row kernel for shorter streaks and unsupported inputs.
Use bounded overlap-save convolution of its merged bilinear coefficients when
channel pairs, parallelism and memory estimates favor FFT. Preserve alpha cutoff
behavior, cancellation, selections and depth/model-neutral filtering.

Validate 303 algorithm tests in debug and release, plus native checks and full
image comparisons. Repeated current-upstream comparisons show 2.31x at 24 MP
and 4.15x at 1.5 MP for angle 30 / distance 1000 on the measured host.

Co-authored-by: optimalbytee <optimalbytee@users.noreply.github.com>
2026-10-09 09:02:00 +07:00
869ecf3cff Red Eye tool in the J-group flyout (#1084)
* Add the Red Eye tool to the J-group

Click a red pupil to neutralize the flash reflection. Pupil Size
sets the search window around the click; Darken Amount controls how
far corrected pixels go toward a darker, desaturated pupil.
Catchlights stay white. A miss reports an error instead of a silent
no-op. The kernel only reads a window plus a 5 px mask-filter halo.

* paint.redEye: write back only changed pixels, so CMYK/Lab stay untouched

write_rgba re-encoded every pixel of the click window through the naive
RGBA round trip, which regenerates K in CMYK and shifts Lab values. Now
untouched pixels keep their original channel data; only pixels whose RGB
changed are re-encoded. Adds a CMYK test checking untouched pixels inside
the window stay bit-exact.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* algo::redeye: cargo fmt

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Polish catalog rows for the Red Eye tool; regenerate docs/scorecard.md

pl.tsv landed on main and the coverage tests need every catalog to translate new UI strings.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-08 19:57:14 -04:00
ef8610adeb Fix radial blur sampling density (#515) (#658)
* Fix radial blur sampling density (#515)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot-Session: e77cd34a-ce3e-4981-996f-4d37d49df90d

* Bound radial blur quality and parallelize sampling

Add Draft, Good, and Best radial sampling tiers, defaulting to Good, and process sampled blur rows in parallel. Preserve compatibility with serialized filters that omit quality.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot-Session: e85c024b-ace5-4415-9208-b8d4e4684505

* i18n(ko): Radial Blur quality choices Good and Best

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Copilot-Session: e77cd34a-ce3e-4981-996f-4d37d49df90d
Copilot-Session: e85c024b-ace5-4415-9208-b8d4e4684505
2026-10-08 18:28:01 -04:00
optimalbyteeandoptimalbytee 0ab30f0c9e Optimize Mosaic by averaging each cell once per tile (#1294)
Co-authored-by: optimalbytee <optimalbytee@users.noreply.github.com>
2026-10-08 17:04:45 -04:00
Matt Van Horn 649417af1d Relight: size the tile halo from σ, keep large images, speed up the blur (#895)
The neighbourhood radius was a 97 px constant (3σ+1 at σ=32) while softness
can drive σ to 96, so tiles seamed once the short side reached about 1600 px.
Halo is now 3σ+1 plus the finite-difference step, taken from the document's
σ (or the 96 cap when size is unknown).

The kernel used to copy the input unchanged above 8 MP of luma. Large rects
are split into overlapping bands instead, so an 8 MP+ layer is actually relit.

Replace the O(nσ) separable kernel with the three-box Gaussian already used
for other smoothing fields in this crate.
2026-10-08 16:39:05 -04:00
304322960e Gaussian and Box Blur: large radii no longer slow down (live preview too) (#464) (#867)
Blurs run as box passes now filter the whole area at once instead of tile by tile: the
horizontal passes run over the source rows only and the vertical passes over column strips,
with the canvas edge repeated virtually instead of materialising a 3-sigma margin around every
tile (at radius 500 that margin was ~25 MP of floats per 4 MP tile, copied and transposed
several times). The results are written to the layer's tile columns in parallel.

Output matches the previous running sums (8-bit: identical except 114 of 2.1 M samples one
level apart; 16-bit: within 2/65535) and a direct-convolution oracle at every depth.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
2026-10-07 21:16:38 -04:00
0f16bce0c5 Add the Magnetic Lasso tool (#864)
* Add the Magnetic Lasso tool

A selection border that snaps to edges as the pointer moves, in the
Lasso (L) flyout after the Polygonal Lasso.

- photocraft-algo `magnetic`: live-wire edge tracing (Mortensen &
  Barrett 1995): Sobel strength, non-maximum suppression and gradient
  direction link costs, a shortest path inside a corridor of the
  detection width around the pointer's path, sub-pixel edge centres.
  Pixels are read lazily in 128 px blocks.
- Engine: `EdgeSource` (the active pixel layer, else the composite) and
  `select.magneticLasso`: rough points in, a border along the edges out;
  `trace: false` selects a given outline, which is how the tool commits.
- UI: click to start, move to trace (button up or down), points fasten
  by themselves (Frequency), click fastens, Alt-click straight, Alt-drag
  freehand, Backspace/Delete removes points, click the first point,
  double-click or Enter closes, Esc cancels, [ ] change the width.
  Options bar: Width, Contrast, Frequency, tablet pressure. `ui.inspect`
  reports the border in progress.
- The Lasso and Polygonal Lasso now apply the options bar's Feather;
  `select.lasso` rejects a feather outside 0..1000.
- Icon adapted from Noun Project's "Magnetic Lasso Tool" by Lil' Seal
  (CC BY 3.0, licence file and ATTRIBUTION.md row).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* it: translate the Magnetic Lasso strings (Italian landed after this PR was written)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
2026-10-07 19:42:45 -04:00
b70bf524ea Filter › Render › Relight: redirect a photo's lighting (#735)
* Add Filter › Relight to redirect a photograph's lighting

Introduce a tiled Retinex and Lambertian kernel with angle, elevation,
intensity, ambient, warmth and softness. Wire it as filter.render.relight
with a live-preview dialog, selection support, and smart-filter recording.

* Tidy Relight kernel loops

Rewrite the per-channel colour multiply and alpha restore so the filter
builds cleanly under clippy.

* Format the Relight kernel and enablement check

Keep rustfmt's wrapping on the alpha restore and collapse the lock
predicate to a single expression.

* Step Relight shading gradients at the blur scale

A one-pixel difference vanishes on large documents, so the lamp
collapsed to a flat multiply. Sample S at ±σ and keep a 256 px
sphere direction check so that does not regress.

* i18n: German entry for Filter › Render › Relight

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* i18n: Brazilian Portuguese entry for Filter › Render › Relight

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-07 16:45:57 -04:00