Progress docs: target-app parity, gaps, alpha gate, readiness by audience (#2696)

Progress docs per craftrules standards/progress-docs.md: target-app parity (merges parity-estimate.md), gaps, roadmap with alpha gate, readiness by audience (full / mainstream / essentials) with Opus 5.5 hours, and UI, brush, format, hardware and localization parity docs. docs/parity.md is now docs/parity-checklist.md. The Windows test failure (file_dialog::tests::save_dialogs_remember_where_each_document_was_last_saved) is pre-existing on main: the test hard-codes POSIX paths. Co-Authored-By: Claude Opus 5.5 <[email protected]>
2026-10-14 11:16:53 -04:00 · 2026-10-10 15:09:04 -04:00
1 parent 5ab78e8ce1
commit 939c8f2c70
@@ -11,8 +11,12 @@ PhotoCraft is an open-source, native, Photoshop-comparable image editor written
 | `docs/contributing.md` | Rules: clean-room, tests, layering, style, commits; the "add a command" checklist |
 | `docs/control-protocol.md` | JSON control channel: how agents drive and screenshot the running app |
 | `docs/ui-design.md` | Design tokens, themes, widgets, and how to match Photoshop's look |
-| `docs/roadmap.md` | Honest parity assessment (where we're lacking, where we're going), milestones, **current focus** |
-| `docs/parity.md` | Generated list of every Photoshop menu item, live or missing |
+| `ROADMAP.md` | One-page summary: stage, the two parity numbers, every dimension, languages, what's next |
+| `docs/target-app-parity.md` | The authoritative parity assessment against Photoshop 2026: numbers, weights, evidence, hours |
+| `docs/gaps.md` | Every known shortfall, ranked: the work list to pick from |
+| `docs/roadmap.md` | Milestones, **current focus**, what's next with estimates |
+| `docs/ui-parity.md`, `docs/brush-parity.md`, `docs/context-menu-parity.md`, `docs/file-format-parity.md`, `docs/hardware-parity.md`, `docs/localization-parity.md` | Per-dimension parity checklists |
+| `docs/parity-checklist.md` | Generated list of every Photoshop menu item, live or missing |
 | `docs/scorecard.md` | Generated scorecard: performance budgets and numbers, corpus floors, per-area checklists (tools, files, UI, type, automation, reliability, distribution), settings that do nothing |
 | `crates/<name>/README.md` (where present) | Public API of that crate |
 | [photocraft-corpus](https://github.com/storytold/photocraft-corpus) + `docs/development.md` › Test corpora | Real-file test oracles (our Photoshop-authored PSDs); with psd-tools, ag-psd and PngSuite fetched into `corpus/` by `cargo xtask corpus --all` at the pins in `xtask/src/corpus_pins.rs` |
@@ -70,15 +74,16 @@ People trust PhotoCraft with their work, and a crash loses it. A malformed file,

 Priorities: important infrastructure first, then low-hanging parity, then the long tail.

-0. **Read `docs/roadmap.md` → "Honest parity assessment" first.** It says, dimension by dimension,
-   where PhotoCraft is lacking and the priority order of where we're going. `docs/parity.md`
-   (menu wiring) is not a measure of behaviour. When your work moves a measured number (PSD oracle,
-   round trips, workflow tests, performance), update that section with the dated figure.
+0. **Read `ROADMAP.md`, then `docs/gaps.md`.** `docs/target-app-parity.md` says, dimension by
+   dimension, where PhotoCraft is lacking; `docs/gaps.md` ranks the work. `docs/parity-checklist.md`
+   (menu wiring) is not a measure of behaviour. When your work moves a number or closes a gap,
+   update the parity doc and gap entry it belongs to, with the date, and bump their status line
+   and revision history (craftrules `standards/progress-docs.md`).
 1. **Check `docs/scorecard.md`** before picking work: each area's `missing` and `partial` rows,
   the performance scenarios that are over budget or not measurable yet, and the count of
   settings that do nothing. Its numbers are measured; prefer them to estimates.
 2. `docs/roadmap.md` → **Current focus**.
-3. `cargo xtask parity` → `docs/parity.md` lists every missing menu item, grouped by menu. Low-hanging fruit is usually a missing command whose algorithm already exists in `algo`, `paint`, `vector` or `text`.
+3. `cargo xtask parity` → `docs/parity-checklist.md` lists every missing menu item, grouped by menu. Low-hanging fruit is usually a missing command whose algorithm already exists in `algo`, `paint`, `vector` or `text`.
 4. `log/devlog.md` → the "Still open" bullets of recent entries.

 When parity rises, raise `FLOOR` in `crates/ui-egui/src/parity.rs` (never lower it).
@@ -90,7 +95,7 @@ cargo test -p <crates you touched>
 cargo clippy -p <crates> --all-targets -- -D warnings
 cargo xtask layers
 cargo xtask wasm            # if you touched L0–L6
-cargo xtask parity          # if you added commands; commit the regenerated docs/parity.md
+cargo xtask parity          # if you added commands; commit the regenerated docs/parity-checklist.md
 cargo test -p photocraft-engine --test panic_hunt -- --ignored   # if you added/changed commands: no panic on adversarial input (Rule 9)
 cargo xtask scorecard       # if you moved a number: flip the checklist row in scorecard/*.toml, raise a
                            # corpus floor, fix a dead preference, or meet a budget (then set enforce = true
@@ -119,7 +124,8 @@ Then append a terse entry to `log/devlog.md` (what landed, numbers, what's still


 - `docs/roadmap.md`: milestones M0–M12, status and the current focus.
- `docs/parity.md`: generated Photoshop menu coverage.
+- `ROADMAP.md`, `docs/target-app-parity.md`, `docs/gaps.md` and the `docs/*-parity.md` checklists: where we stand against Photoshop.
+- `docs/parity-checklist.md`: generated Photoshop menu coverage.
 - `docs/scorecard.md`: generated scorecard (sources: `scorecard/*.toml`, `perf/budgets.toml`, `perf/baseline.json`, corpus floors, prefs audit).
 - `docs/releasing.md`: cutting a release (`cargo xtask version`, the `release` branch), signing secrets, packaging scripts in `packaging/`.
 - `../craftrules/release/playbook.md`: how every storytold app builds signed release binaries (the canonical recipe; `docs/release-playbook.md` just points there); `docs/releasing.md` is PhotoCraft's specifics.
@@ -298,7 +298,7 @@ photocraft
 Maintainers: [`docs/releasing.md`](docs/releasing.md) explains how releases are built, signed and published.

 > [!IMPORTANT]
-> **Status:** PhotoCraft is in early alpha, and we want to be straight about where it stands: much of Photoshop's feature surface exists in some form, but **it is not yet a Photoshop replacement for daily professional work**. The biggest gaps are AI/generative features, about twenty missing tools, depth in typography and pro workflows, and plug-in compatibility. Every Photoshop menu item is wired to a command ([`docs/parity.md`](docs/parity.md)), but that measures wiring, not behaviour. The honest, dimension-by-dimension picture and where we're going next are in the [roadmap's parity assessment](docs/roadmap.md#honest-parity-assessment-2026-10-05). Expect rough edges, and please file issues (include your OS, document size, layer count and a screenshot). You can also tell us what broke on [Discord](https://discord.gg/artcraft).
+> **Status:** PhotoCraft is in early alpha, and we want to be straight about where it stands: much of Photoshop's feature surface exists in some form, but **it is not yet a Photoshop replacement for daily professional work**. The biggest gaps are AI/generative features, 15 missing tools, depth in typography and pro workflows, and plug-in compatibility. Every Photoshop menu item is wired to a command ([`docs/parity-checklist.md`](docs/parity-checklist.md)), but that measures wiring, not behaviour. The honest, dimension-by-dimension picture and where we're going next are in the [roadmap](ROADMAP.md) and [parity assessment](docs/target-app-parity.md). Expect rough edges, and please file issues (include your OS, document size, layer count and a screenshot). You can also tell us what broke on [Discord](https://discord.gg/artcraft).

 ## Documentation

@@ -0,0 +1,141 @@
+# PhotoCraft roadmap
+
+**Stage: alpha** · next: beta, ~30 points (ready for real work ~45% → ~75%, with reliable PSD) and ~1,100–1,800 h away
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: full re-measure against Photoshop 2026 27.11.0) · **Target:** Adobe Photoshop 2026
+
+PhotoCraft's one-page summary of how close it is to Adobe Photoshop 2026 (27.11.0, measured from
+the installed app on 2026-10-10) at version 0.6.0. Detail: [target-app parity](docs/target-app-parity.md)
+(methodology, weights, evidence), [gaps](docs/gaps.md) (the ranked work list),
+[roadmap](docs/roadmap.md) (milestones, current focus), [scorecard](docs/scorecard.md) and
+[menu checklist](docs/parity-checklist.md) (both generated and measured),
+[architecture](docs/architecture.md).
+
+Hours are Opus 5.5 agent wall-clock hours for one agent working sequentially, calibrated from this
+repository's 1,004 merged PRs (see [calibration](docs/target-app-parity.md#remaining-effort-and-calibration)).
+About 80% of the work parallelizes.
+
+## Headline numbers
+
+| | Value | Kind |
+|---|---|---|
+| **Feature breadth** | **~76%** | partly measured: menus 628/628, tools 53/68, panels 30/35, formats 13+4 of 31 |
+| **Ready for real work** | **~45%** (40–50%) | estimated: weighted by dimension below |
+| Mainstream practitioner | ~43% (38–48%) | estimated: weekly areas of a typical pro, with discounts ([method](docs/target-app-parity.md#mainstream-practitioner-43)) |
+| Essentials user | ~61% (55–67%) | estimated: core features only ([method](docs/target-app-parity.md#essentials-user-61)) |
+| Remaining to **beta** | **~1,100–1,800 h** | estimated |
+| Remaining to **full parity** | **~2,300–4,100 h** | estimated; AI and plug-ins need owner decisions |
+
+### Readiness by audience
+
+| Audience | Ready % | Opus 5.5 agent-hours to ~95% | Work that dominates |
+|---|---:|---:|---|
+| Full target (ready for real work) | ~45% (additive weighted sum: 46%) | 2,300–4,100 | Everything below, plus AI tier, plug-in/scripting ecosystem, video, localization, hardware |
+| Mainstream practitioner | ~43% | 1,700–2,800 | Depth in the 11 weekly areas (1,180–2,000), performance budgets, stability on real machines, PSD exchange with Photoshop |
+| Essentials user | ~61% | 400–700 | Startup and GPU robustness, brush and transform responsiveness, Type tool, discoverability (flyouts, commit buttons), AVIF/PDF open |
+
+Hours calibrated as in [Remaining effort](docs/target-app-parity.md#remaining-effort-and-calibration) (2–9 h per checklist
+item, 8–20 h per format or subsystem, from the repo's 1,004 merged PRs); each tier is a subset of
+the one above. About 80% parallelizes (independent crates and command modules); performance
+architecture and the PSD writer are the serial parts.
+
+**Why alpha.** Core workflows (open, layer, select, paint, adjust, filter, type, save) work end to
+end, and breadth is high, but users of 0.5/0.6 have 579 open issues, 22 of 25 performance budgets
+are missed, and PSDs we write are not yet reliably accepted by Photoshop (#1281, #2469). Beta
+needs ~75% ready and no blocking PSD gap. The [alpha gate](docs/roadmap.md#alpha-gate) passes: all six core workflows
+(retouch, composite, cut-out, design, paint, PSD exchange) complete end to end and save and reopen;
+four are partial in depth, none blocked.
+
+## By dimension
+
+| Dimension | % ready | Hours to full | Doc |
+|---|---:|---:|---|
+| Features (depth) | 55% (breadth 76%) | 600–1,000 | [target-app-parity](docs/target-app-parity.md#feature-areas) |
+| UI/UX fidelity (tools, handles, modifiers, brush feel, panels) | 40% | 350–600 | [ui-parity](docs/ui-parity.md), [brush-parity](docs/brush-parity.md), [context-menu-parity](docs/context-menu-parity.md) |
+| File formats | 60% (presence ~80%) | 300–500 | [file-format-parity](docs/file-format-parity.md) |
+| Hardware (GPU, pen, displays) | 50% | 80–150 | [hardware-parity](docs/hardware-parity.md) |
+| Localization | 55% | 80–140 | [localization-parity](docs/localization-parity.md) |
+| Performance | 25% (3/25 budgets met, measured) | 200–350 | [scorecard](docs/scorecard.md#performance) |
+| Stability | 45% | 150–250 | [gaps](docs/gaps.md#g3-user-reported-bug-backlog) |
+| Platforms | 80% | 30–60 | [scorecard](docs/scorecard.md#distribution) |
+| Ecosystem / plug-ins / scripting | 15% | 250–450 | [gaps](docs/gaps.md#g16-plug-in-compatibility) |
+| AI features | 10% | 300–600 | [gaps](docs/gaps.md#g15-ai-and-generative-features) |
+
+## Features
+
+| Area | % ready | Hours |
+|---|---:|---:|
+| Layers, masks, blend modes, smart objects, styles | 65% | 150–250 |
+| Selections, Select and Mask, channels | 55% | 120–200 |
+| Painting and brush engine | 55% | 150–250 |
+| Retouching (healing, clone, patch, remove, content-aware) | 50% | 100–180 |
+| Transform, warp, puppet, liquify | 55% | 100–180 |
+| Adjustments and filters | 70% | 120–200 |
+| Type | 40% | 120–200 |
+| Vector shapes and paths | 55% | 80–140 |
+| Colour management and modes | 70% | 60–100 |
+| Camera Raw and computational photo | 45% | 120–200 |
+| Automation, actions, scripting | 45% | 120–200 |
+| File I/O | 60% | 300–500 |
+| Print, export, Save for Web | 50% | 60–100 |
+| Video / timeline, data-driven graphics | 30% | 100–200 |
+| Workspace, panels, preferences | 40% | 150–250 |
+| AI / generative | 10% | 300–600 |
+
+Rows and evidence: [target-app-parity › Feature areas](docs/target-app-parity.md#feature-areas).
+
+## Languages
+
+Measured with the `cargo xtask i18n-coverage` key universe (1,608 keys). `partial` everywhere
+because engine messages are still English. Detail: [localization-parity](docs/localization-parity.md).
+
+| Language | Code | UI strings | Status |
+|---|---|---:|---|
+| English | `en` | 100% | full |
+| Simplified Chinese | `zh-hans` | 100% | partial |
+| Spanish | `es` | 100% | partial |
+| Hindi | `hi` | 0% | none |
+| Arabic | `ar` | 0% (no RTL UI) | none |
+| French | `fr` | 100% | partial |
+| Portuguese (Brazil) | `pt-br` | 100% | partial |
+| Indonesian | `id` | 100% | partial |
+| Japanese | `ja` | 100% | partial |
+| German | `de` | 100% | partial |
+| Korean | `ko` | 100% | partial |
+| Vietnamese | `vi` | 0% | none |
+
+Also shipped: Traditional Chinese, Russian, Ukrainian, Czech, Polish, Greek, Dutch, Italian (8).
+
+## Upcoming
+
+| Rank | Milestone | Estimate |
+|---|---|---:|
+| 1 | PSD that Photoshop trusts: Save As corruption, Photoshop re-open check, Photoshop-oracle floor 133 → 200+ of 256 | 150–250 h |
+| 2 | Performance budgets: the 22 over-budget scenarios and the 150-layer GPU crash | 200–350 h |
+| 3 | User bug backlog to under 100 open | 250–400 h |
+| 4 | Brush and transform feel (brush-parity, ui-parity rows) | 160–290 h |
+| 5 | The 15 missing tools | 60–100 h |
+
+Detail and the next releases: [docs/roadmap.md](docs/roadmap.md).
+
+## Progress log
+
+| Date | Entry |
+|---|---|
+| 2026-10-10 | Full re-measure against Photoshop 2026 27.11.0 (installed bundle inspected): breadth ~76%, ready ~45%, alpha. Progress docs reorganized to the craftrules standard (`parity.md` → `parity-checklist.md`, `parity-estimate.md` merged into `target-app-parity.md`). 0.6.0 released. |
+| 2026-10-09 | Remove tool, Rotate View, crop rotation; OpenRaster and Paint.NET; 39 Affinity documents in the corpus |
+| 2026-10-08 | 0.5.0. Affinity import; Magnetic Lasso, Pattern Stamp, Mixer Brush, Pencil, Patch, Vertical Type on the toolbar |
+| 2026-10-07 | 0.3.0. 10 UI languages; shortcut audit 214 → 0 failures; TIFF decode 10× faster |
+| 2026-10-05 | 0.2.0, first signed and notarized release. Honest assessment: ready for real work 25–35% |
+| 2026-10-03 | Menu breadth 625/625 |
+| 2026-10-01 | Menu breadth 224 → 532 of 625 in a day |
+| 2026-09-30 | First commit |
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | minor | Readiness by audience with hours per tier |
+| 2026-10-10 | minor | Mainstream-practitioner and essentials-user numbers added |
+| 2026-10-10 | minor | Alpha gate result (passes; stays alpha) in the stage explanation |
+| 2026-10-10 | major | First version, per craftrules `standards/progress-docs.md` |
@@ -1,6 +1,6 @@
 # Roadmap

-The live product roadmap, milestone definitions, parity measurements, and current focus are maintained in [`docs/roadmap.md`](https://github.com/storytold/photocraft/blob/main/docs/roadmap.md). Menu coverage is generated into [`docs/parity.md`](https://github.com/storytold/photocraft/blob/main/docs/parity.md) by `cargo xtask parity`.
+The live product roadmap, milestone definitions, parity measurements, and current focus are maintained in [`docs/roadmap.md`](https://github.com/storytold/photocraft/blob/main/docs/roadmap.md). Menu coverage is generated into [`docs/parity-checklist.md`](https://github.com/storytold/photocraft/blob/main/docs/parity-checklist.md) by `cargo xtask parity`.

 Security hardening is cross-cutting rather than a replacement for those product milestones. The current security sequence is:

@@ -1,7 +1,7 @@
 //! Print Photoshop menu parity and optionally write the Markdown report.
 //!
 //! ```sh
-//! cargo run -p photocraft-ui-egui --example parity -- [--write docs/parity.md] [--json]
+//! cargo run -p photocraft-ui-egui --example parity -- [--write docs/parity-checklist.md] [--json]
 //! ```
 //! Usually run through `cargo xtask parity`.

@@ -1920,7 +1920,7 @@ mod reveal_label_tests {
            _ => "Im Ordner anzeigen",
        };
        assert_eq!(translated_menu_label(de, reveal[0]), expected);
-        // The menu catalogue (and docs/parity.md) keeps Photoshop's macOS name.
+        // The menu catalogue (and docs/parity-checklist.md) keeps Photoshop's macOS name.
        assert!(crate::menu_catalog::CATALOG.iter().any(|(_, label, _, i)| *i == id && *label == "Reveal in Finder"));
    }
 }
@@ -1,6 +1,6 @@
 //! Photoshop menu parity: how much of [`crate::menu_catalog::CATALOG`] is live.
 //!
-//! `cargo xtask parity` prints this report and writes `docs/parity.md`. A test keeps the live
+//! `cargo xtask parity` prints this report and writes `docs/parity-checklist.md`. A test keeps the live
 //! count from regressing below [`FLOOR`].

 use crate::menu_catalog::CATALOG;
@@ -31,7 +31,7 @@ impl Parity {
        if self.total == 0 { 0.0 } else { 100.0 * self.live as f64 / self.total as f64 }
    }

-    /// The report as Markdown (the body of `docs/parity.md`).
+    /// The report as Markdown (the body of `docs/parity-checklist.md`).
    pub fn to_markdown(&self) -> String {
        let mut s = String::new();
        s.push_str("# Photoshop menu parity\n\n");
@@ -1,6 +1,35 @@
-# Photocraft Architecture
+# PhotoCraft architecture

-Status: draft v1 (2026-09-30). This plan does not assume a final UI toolkit. The candidates are discussed in [`rust-framework-options.md`](rust-framework-options.md).
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** minor (current-state summary added; the design plan below is kept, with its deviations listed) · **Target:** Adobe Photoshop 2026
+
+## Current state (2026-10-10)
+
+What exists on `main` today. The numbered sections below are the original design plan of
+2026-09-30; where the code differs, this summary wins.
+
+- **Workspace:** 25 library crates under `crates/`, 3 apps (`photocraft` desktop, `photocraft-cli`,
+  `photocraft-web`), `xtask`, an example plug-in and 5 fuzz crates; ~393,000 lines of Rust in
+  ~1,020 files; ~5,460 `#[test]` functions (ui-egui 1,677, engine 1,237, algo 408, io 377,
+  codecs 325, psd 263). Crate map and layers: [`AGENTS.md`](../AGENTS.md) §2 and
+  `xtask/src/layers.rs` (enforced by `cargo xtask layers`).
+- **UI toolkit:** decided: egui/eframe on wgpu (no webview, no JavaScript). The same Rust builds
+  to WebAssembly for the browser (trunk + wasm-bindgen).
+- **Engine:** ~700 commands registered in `crates/engine` (`CommandSpec` in `commands.rs`); the
+  UI, CLI, JSON control channel and MCP server all dispatch by command id. 628 Photoshop menu items
+  map to them (`crates/ui-egui/src/menu_catalog.rs`).
+- **Rendering:** CPU compositor (`compose`, the oracle) and wgpu compositor (`gpu`, ≤ 1/255 of the
+  CPU); colour through our own ICC engine (`cms`), display transform folded into a GPU 3D LUT.
+- **Files:** own PSD/PSB crate (`psd`), flat codecs (`codecs`), camera raw (`raw`), HEIF (`heif`),
+  Affinity (`affinity`), document mapping (`io`), native `.pcraft` (`format`, atomic saves).
+  Coverage: [`file-format-parity.md`](file-format-parity.md).
+- **Not built (yet) from the plan:** the `ml` crate (no ML runtime; see gap G15), plug-in ABI v2.
+- **Deviations from the plan below:** `photocraft-tablet` is the one crate allowed `unsafe`
+  (AppKit pen events); `affinity`, `heif` and `raw` crates were added; the phased roadmap in §15
+  is superseded by [`roadmap.md`](roadmap.md).
+
+---
+
+# Design plan (2026-09-30)

 This plan sets out:
 - the Cargo workspace,
@@ -529,7 +558,7 @@ Versions get pinned when we scaffold, using the numbers in `rust-framework-optio

 ## 15. Phased roadmap

-Each phase ends with a demoable build and green CI on all targets. Phases 8–10 (colour/print pro, photographer/designer pro, frontier) and the per-feature phase assignments are in [`photoshop-parity.md`](photoshop-parity.md).
+Each phase ends with a demoable build and green CI on all targets. Phases 8–10 (colour/print pro, photographer/designer pro, frontier) and the per-feature phase assignments are in [`target-app-parity.md`](target-app-parity.md).

 | Phase | Scope | Done when |
 |---|---|---|
@@ -552,3 +581,10 @@ Each phase ends with a demoable build and green CI on all targets. Phases 8–10
 4. **PSD crate.** Build our own, or adopt or fork `ag-psd`/`psd`. Decide after the week-1 spike.
 5. **Default working depth and blending.** 8-bit + gamma-space blending matches Photoshop. 16-bit/float + linear is the modern default. The proposal is to follow Photoshop by default and make the other a per-document option.
 6. **Scope of Photoshop plugin compatibility.** Decided: no native `.8BF` hosting; a sandboxed WebAssembly plug-in API instead (see [`plugins.md`](plugins.md)).
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | minor | Current-state summary and deviations from the plan; status line; broken link to `photoshop-parity.md` repointed to `target-app-parity.md` |
+| 2026-09-30 | major | Design plan v1 |
@@ -0,0 +1,113 @@
+# Brush and painting parity
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: brush engine, dynamics, pressure, smoothing, cursors, retouch tools and feel audited from source and open issues) · **Target:** Adobe Photoshop 2026
+
+Painting is where Photoshop users notice differences first, so it gets its own checklist: the
+brush engine (`crates/paint/src/brush.rs`), the Brush Settings UI
+(`crates/ui-egui/src/brush_sections.rs`), the live stroke renderer, pen input
+([`hardware-parity.md`](hardware-parity.md)), and the brush-based tools. Status words as in
+[`ui-parity.md`](ui-parity.md).
+
+**Summary (estimated):** breadth ~90% (every Brush Settings section and option exists), ready for
+real work ~55%. The engine models what Photoshop models; the gaps are feel (latency, pressure
+response, edge quality, smoothing), a few missing options, and live previews for retouch tools.
+Remaining: 100–180 h (gap G4 in [`gaps.md`](gaps.md)). Human: feel must be checked on real
+tablets against Photoshop side by side.
+
+## Brush Settings sections
+
+| Section | Status | Evidence |
+|---|---|---|
+| Brush Tip Shape: size, angle, roundness, hardness, spacing, flip X/Y | done | `brush.rs` |
+| Shape Dynamics (size/angle/roundness jitter, pen pressure/tilt/rotation/wheel controls, minimums, flip jitter, brush projection) | done | #174 |
+| Scattering (both axes, count, count jitter) | done | |
+| Texture (pattern, scale, brightness, contrast, mode, depth, each tip) | done | |
+| Dual Brush | done | |
+| Color Dynamics (FG/BG, hue/sat/brightness jitter, purity, per tip) | done | |
+| Transfer (opacity/flow/wetness/mix jitter and controls) | done | |
+| Brush Pose (tilt, rotation, pressure overrides) | done | |
+| Noise, Wet Edges, Build-up, Protect Texture | done | |
+| Smoothing: amount, Pulled String, Stroke Catch-up, Catch-up on End, Adjust for Zoom | done | `brush.rs:428-438` |
+| Section locks | done | #174 |
+| Brush tip from an image file | missing | TOOL-213-12 |
+| Erodible, Airbrush and Bristle tips (Photoshop's special tips) | partial | not audited in detail; "3D-based brushes don't work" (#1660) |
+| Anti-aliasing off for brushes | missing | #1666 (Pencil is the aliased path) |
+
+## Pressure and pen
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Pressure drives size/opacity/flow per sample (no shared pressure within a frame) | done | #2112 |
+| Global pressure curve in Preferences | done | #2195 |
+| Per-control pressure curves | missing | TOOL-213-11 |
+| Quick taps keep their pressure | partial | #1967 fixed one case; tap paints a full blob (#1798), tap sensitivity (#1933) |
+| Smooth pressure on Windows | partial | "jumpy" (#2046) |
+| Tilt and rotation drive angle/roundness | partial | macOS, X11, web; not Windows; rotation not applied (#1301) |
+| Eraser end of the pen switches to Eraser | partial | macOS; flipped Wacom on Windows not (#2285) |
+
+## Stroke quality and feel
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| No seams, hard edges stay hard, zero-scatter defaults, ABR tips never black | done | TOOL-213-13 |
+| Smooth line weight along a pressure ramp | partial | stair-steps (#1776) |
+| Interpolation and smoothing match Photoshop | partial | #2138, #2057 |
+| Soft-edge falloff matches Photoshop's curve | partial | #2149, #1619 |
+| Brush preview and cursor crisp, not pixelated | partial | #1433 |
+| Cursor outline equals the painted footprint at every zoom | partial | #1761, #2364 |
+| Latency: dab to screen within a frame | partial | budgets P13/P16/P33 over (scorecard); big-brush lag #2619, #1994; cursor lag #2566, #1343 |
+| No hitch at mouse-up on a long stroke | not measurable | P32 needs a frame harness |
+| Live stroke for Brush and Eraser | done | |
+| Live stroke for every brush-based retouch tool | missing | TOOL-213-7 |
+| No OS arrow cursor over the canvas with a pen | partial | TOOL-213-14 |
+
+## Gestures and cursors
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| ⌃⌥-drag (macOS) / Alt+right-drag (Windows) resizes and changes hardness with a readout | done | TOOL-213-10 |
+| `[` `]` size, `⇧[` `⇧]` hardness, number keys opacity, ⇧+numbers flow | done | 2026-10-07 |
+| ⌥-click samples a colour with painting tools; Eyedropper comparison ring | done | TOOL-213-9 |
+| ⇧-click draws a straight line from the last point | done | |
+| Right-click opens the Brush Preset picker | done | [`context-menu-parity.md`](context-menu-parity.md); XP-Pen right-click not (#1980), macOS issue #2360 |
+| Precise / normal / full-size tip cursors, crosshair in the tip | partial | Caps Lock precise cursor not audited; round Pencil cursor requested (#2662, #1057) |
+| Brush choice kept per tool (Brush, Eraser, Blur each their own) | partial | #2542 |
+
+## Symmetry
+
+| Mode | Status |
+|---|---|
+| Vertical, Horizontal, Dual axis, Diagonal | done |
+| From a path (Make Symmetry Path) | done |
+| Wavy, Circle, Spiral, Parallel Lines, Radial, Mandala | missing (#2659) |
+
+## Brush-based tools
+
+| Tool | Status | Notes |
+|---|---|---|
+| Brush, Pencil (aliased, Auto Erase), Eraser (Brush/Pencil/Block) | done | Pencil 1 px quality #2139 |
+| Mixer Brush (Wet, Load, Mix, Flow, Sample All Layers) | partial | "incomplete" (#2247) |
+| Clone Stamp (opacity, flow, aligned, sample, overlay) | partial | Clone Source W/H link and Alt target cursor (TOOL-213-8) |
+| Pattern Stamp (aligned, impressionist) | done | |
+| History Brush | done | Art History Brush missing |
+| Spot Healing, Healing | partial | grain not matched on flat colour (#2104), "horrible in general" (#1092), stops responding after switching (#1751) |
+| Blur, Sharpen, Smudge (premultiplied, Sample All Layers, spacing) | done | BUG-207-2/3, #2171 |
+| Dodge, Burn, Sponge | done | |
+| Background Eraser, Magic Eraser | done | |
+| Color Replacement | missing as a tool | engine command exists |
+| Quick Selection (brush-driven) | done | |
+
+## Presets
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| .abr import (v1/v2/v6) and export, persistent presets, folders | done | `psd/src/abr.rs` |
+| Brushes panel: drag a folder into a folder | missing | #2014 |
+| Tool presets (.tpl) | partial | Tool Presets panel exists; .tpl import not found |
+| Import 500 brushes without blocking the UI | not run | P24 |
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | major | First version, from a source audit and the open issues |
@@ -1,5 +1,7 @@
 # Context-menu inventory and parity contract

+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** trivial (status line added; the inventory is the 2026-10-06 branch-point snapshot, current status lives in context-menu-parity.md) · **Target:** Adobe Photoshop 2026
+
 This is the source-code inventory of PhotoCraft's right-click behavior at the branch point, captured 2026-10-06. It distinguishes a menu from another secondary-button gesture. The initial selection, Paths row, and document-tab menus added by this PR are called out separately. Photoshop comparison should be checked against an identified Photoshop version and platform before claiming exact parity; the current Layers menu explicitly targets Photoshop 2026 order.

 ## Canvas: every tool
@@ -50,3 +52,10 @@ Canvas invariant: secondary click must not accidentally start a paint stroke, co
 3. Prioritize shared canvas menus, then high-value tool-specific interactions (path, selection, type, crop/zoom), then panel and tab gaps. Reuse Layers/Channels implementations where their command contract already exists. Preserve the existing Move/layer pick, brush picker, erase preference, Alt/Option brush resize, and toolbar flyout gestures.
 4. Test each descriptor's item order and enablement, dispatch with expected params, empty/invalid states, modifiers, and action result. Add UI gesture tests for right-click opening, target selection, Escape/outside dismissal, and the absence of accidental edits. Validate native and web builds.
 5. Capture screenshots from the running app for representative tool groups and panel states, with test fixture documents rather than personal photos. A screenshot proves visible menu composition; command tests prove its behavior.
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | trivial | Status line, revision history; marked as the branch-point snapshot |
+| 2026-10-06 | major | Source inventory of right-click behaviour |
@@ -1,5 +1,7 @@
 # Context-menu parity contract

+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** trivial (status line and revision history added; linked from ui-parity.md) · **Target:** Adobe Photoshop 2026
+
 Status: implementation contract, updated 2026-10-09. Scope: canvas, document tabs, and panels. “Right-click” means a secondary pointer click; macOS Control-click should also open the same menu where the platform sends a context-menu gesture. Photoshop changes a context menu with the active tool, selection, and panel item. Adobe documents that principle but does not publish a complete current menu catalog. **Documented** means an Adobe source explicitly names the gesture or action. **Inferred** means a PhotoCraft design target or an item known from a main menu whose presence in a Photoshop context menu still needs a current-product capture. Do not present inferred rows as exact Photoshop menus. [Adobe: panels and menus](https://helpx.adobe.com/photoshop/using/panels-menus.html), [Adobe: shortcuts and context menus](https://helpx.adobe.com/sa_en/photoshop/using/customizing-keyboard-shortcuts.html).

 ## Dispatch contract
@@ -93,3 +95,11 @@ Photoshop's Generative Fill and Delete and Fill Selection rows are left out: Pho
 4. Long tail: shape/path, crop, metadata and specialist tools. Keep a per-tool fixture with no selection, active selection, masked/locked layer, and active operation where applicable.

 Each slice needs pure model tests (including every item dispatching a known command), UI pointer tests for secondary click and modifier precedence, command/undo tests, a disabled-state test, and screenshots from PhotoCraft with the menu actually open. Compare labels, order, and state against dated Photoshop screenshots; a screenshot of a different tool or menu bar is insufficient evidence. Maintain a checklist beside each captured reference: Photoshop version/platform, target description, tool, selection/document state, modifiers, and exact items observed.
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | trivial | Status line, revision history |
+| 2026-10-09 | minor | Selection canvas reference and tool matrix updates |
+| 2026-10-07 | minor | Pen canvas reference capture |
@@ -23,4 +23,4 @@
 4. **Respect the context:** active selection (feathered), layer vs mask target, locks, the colour model and depth.
 5. **Tests** in the module: behaviour, undo/redo, disabled states, bad params, several depths.
 6. **Dialog** (if it has parameters): filters and adjustments get schema-driven dialogs with live preview (`ui-egui/src/filter_dialog.rs`, `dialogs.rs`).
-7. Run `cargo xtask parity` and commit the updated `docs/parity.md`.
+7. Run `cargo xtask parity` and commit the updated `docs/parity-checklist.md`.
@@ -41,7 +41,7 @@ cargo run --release -p photocraft -- --control 7878 --control-token-file .privat
 cargo test --workspace                                     # everything
 cargo xtask ci                                             # fmt + clippy + tests + layers + wasm
 cargo xtask stats                                          # tests and lines per crate
-cargo xtask parity                                         # Photoshop menu coverage -> docs/parity.md
+cargo xtask parity                                         # Photoshop menu coverage -> docs/parity-checklist.md
 ```

 Image code is slow at `opt-level 0`, so the workspace profile builds dependencies at `opt-level 2`. Use `--release` for anything interactive.
@@ -253,7 +253,7 @@ intents, black point compensation). It ships CC0 built-in profiles, including a
 ## Menu parity

 `cargo xtask parity` compares Photoshop's menu tree (`crates/ui-egui/src/menu_catalog.rs`) with
-the live command registry (`menus::is_live`) and rewrites [`docs/parity.md`](parity.md). The test
+the live command registry (`menus::is_live`) and rewrites [`docs/parity-checklist.md`](parity-checklist.md). The test
 `parity::tests::parity_does_not_regress` fails if the live count drops below `parity::FLOOR`.

 `cargo xtask i18n-coverage` prints the UI translation coverage for each language in the
@@ -315,7 +315,7 @@ cargo xtask perf --update-baseline   # also write perf/baseline.json from this r
  class, commit, date and load average.
 - `crates/io/tests/corpus.rs`: the corpus floors (every `Source { .. }` constant).
 - The prefs audit: `Preferences` fields that no code reads (target 0, #204), plus counts taken
-  from the tree (never-crash attribute coverage, `docs/parity.md`).
+  from the tree (never-crash attribute coverage, `docs/parity-checklist.md`).

 **Perf runs.** `xtask perf` builds the benches in release (`perf_scenarios`, `interactive_bench`,
 `fx_bench`, `bevel_bench`, `type_bench`, `large_image_bench`, and `layout_bench` once it exists), runs each with
@@ -0,0 +1,136 @@
+# File-format parity
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: every Photoshop 2026 format from its bundle against our readers and writers; Affinity notes moved here from `roadmap.md`) · **Target:** Adobe Photoshop 2026
+
+Every format Photoshop 2026 (27.11.0) reads or writes, with PhotoCraft's support, fidelity and
+tests. Photoshop's list comes from its `Info.plist` `CFBundleDocumentTypes`, the PiPL read/write
+flags of `Contents/PlugIns/Required/File Formats`, and format names in the main binary (Open/Save
+As behaviour from Adobe's documentation where the bundle doesn't say). Ours comes from
+`crates/codecs/src/format.rs` (capability table), `crates/io/src/lib.rs` (import/export
+dispatch), `crates/psd`, `crates/raw`, `crates/heif`, `crates/affinity`. More detail on our side:
+[raster formats](../book/src/formats/raster-formats.md), [PSD/PSB](../book/src/formats/psd-psb.md),
+[OpenRaster](ora.md), [Paint.NET](pdn.md).
+
+**Summary.** By count, 13 of Photoshop's 31 formats are fully supported in the directions Photoshop supports them, 4 partially (48%,
+measured). Weighted by use (PSD/PSB 40%, JPEG 15%, PNG 15%, TIFF 10%, camera raw 8%, WebP/AVIF/
+HEIF/JXL 6%, PDF 3%, everything else 3%), presence is ~80% and ready-for-real-work ~60%: PSD
+is broad but not yet reliable against Photoshop (gap G1), and Photoshop PDF, AVIF read and JPEG
+XL are missing (gap G10). Remaining: 300–500 h including PSD fidelity.
+
+## Photoshop's formats
+
+| Format | Photoshop | PhotoCraft read | PhotoCraft write | Fidelity and tests |
+|---|---|---|---|---|
+| PSD | R/W | yes | yes | 1/8/16/32-bit; Bitmap, Gray, Indexed, RGB, CMYK, Multichannel, Lab, Duotone; layers, groups, masks, vector masks, adjustment and fill layers, effects, text, shapes, smart objects (SoLd/lnk2), smart filters (14 mapped, rest verbatim), slices, comps, Advanced Blending. Oracle floors: Photoshop-authored 133/256, io 146/170, psd-tools 236/309; round trip ≥ 169/170, 308/309, 258/256. Known failures: #1281, #2469, #2465 (#2374, text layers, closed 2026-10-10) |
+| PSB | R/W | yes | yes | Auto-switches above 2 GB (REL-212-2); 4–5 GB fixtures generated (FILE-216-2); > 4 GB fails on Windows (#375) |
+| Cloud PSD (.psdc) | R/W | no | no | Out of scope (Adobe cloud) |
+| TIFF | R/W | yes | yes | 8/16/32f, Gray/RGB/CMYK, BigTIFF, every IFD, banded parallel decode; layered TIFF both byte orders (FILE-215-7: extra alpha channels not written) |
+| JPEG | R/W | yes | yes | 8-bit Gray/RGB/CMYK, ICC/EXIF/XMP; no JPEG Options dialog (quality 0–12, baseline/progressive) (FILE-215-19) |
+| PNG | R/W | yes | yes | 8/16-bit, ICC/text; APNG first frame only; alpha channels on open (#2225) |
+| GIF | R/W | yes | yes | First frame only |
+| BMP | R/W | yes | yes | 8-bit |
+| WebP | R/W | yes | yes | Lossless and our own lossy VP8 encoder; ICC/EXIF/XMP |
+| AVIF | R/W | no (#463) | optional feature, off by default | |
+| JPEG XL | R/W | inside DNG only | no | |
+| HEIF/HEIC | R/W(HEIF) | yes (pure-Rust heic-rs, `heif` feature, on in releases) | no | 8/16-bit |
+| OpenEXR | R/W | yes | yes | f16/f32; deep EXR and Cryptomatte readers (`corpus/exr`) |
+| Radiance HDR | R/W | yes | yes | f32 |
+| Portable Bit Map (PBM/PGM/PPM/PNM/PFM/PAM) | R/W | yes | yes | 8/16/32f |
+| Targa | R/W | yes | yes | 8-bit |
+| Photoshop PDF | R/W | no (#2541) | raster only (print, artboards to PDF) | No vector/text PDF, no multi-page save, no PDF presentation (FILE-215-10) |
+| Photoshop EPS, DCS 1.0/2.0, EPS TIFF/PICT preview | R/W | no | no | |
+| Generic EPS / AI | R (rasterize) | no | no | |
+| SVG | R (rasterize) | yes, as editable shape layers or a vector smart object | no | Better than Photoshop for editing; gradients approximated (FILE-215-9) |
+| JPEG 2000 | R/W | no (#1047) | no | |
+| Cineon / DPX | R (DPX), R/W (CIN) | no (#1049) | no | |
+| DICOM | R/W | no | no | |
+| PCX | R/W | no | no | |
+| Pixar | R/W | no | no | |
+| Amiga IFF | R/W | no | no | |
+| Wireless Bitmap | R/W | no | no | |
+| MPO / JPS (stereo) | R/W | no | no | |
+| PICT | R | no | no | Legacy |
+| Scitex CT | R/W | no | no | Legacy |
+| Photoshop Raw (.raw) | R/W | no | no | |
+| Camera raw (via Camera Raw 18.7) | R | partial | – | See below |
+
+## Camera raw
+
+Photoshop opens raws through Camera Raw 18.7 (the bundle registers cr2, crw, dcr, dng, erf, mos,
+fff, 3fr, mef, mfw, iiq, nrw, rwl, srw, mrw, nef, orf, pef, raf, srf, x3f; Camera Raw itself
+supports many more, including CR3).
+
+| Camera format | PhotoCraft | Evidence |
+|---|---|---|
+| DNG (incl. LJPEG and JPEG XL tiles) | yes; lossy DNG no | FILE-215-11 |
+| Canon CR2 | yes | |
+| Canon CR3 | detected, unsupported | `crates/raw/src/lib.rs:219` |
+| Nikon NEF/NRW (lossless, lossy) | yes; "lossy after split" no | #50 |
+| Sony ARW/SR2/SRF (incl. compressed) | yes | |
+| Panasonic RW2, Pentax PEF | yes | |
+| Olympus ORF | uncompressed only | |
+| Fujifilm RAF (Bayer and X-Trans) | uncompressed only; no Fuji colour calibration | |
+| Others (CRW, 3FR, IIQ, X3F, SRW, MRW, ERF…) | embedded preview only | |
+| .xmp sidecars, highlight modes, raw preferences | missing | FILE-215-11 |
+| Camera Raw as a PSD smart filter | partial | 2026-10-07: WB, Light/Presence, curves, HSL, grading, detail, grain, vignette mapped (FILE-215-12) |
+
+## Formats Photoshop doesn't have (ours only)
+
+| Format | Read | Write | Notes |
+|---|---|---|---|
+| `.pcraft` (native) | yes | yes | Incremental, autosave, crash recovery, atomic save |
+| Affinity `.af`, `.afphoto`, `.afdesign`, `.afpub` | yes | no | See below |
+| Paint.NET `.pdn` | yes | no | [pdn.md](pdn.md) |
+| OpenRaster `.ora` | yes | yes | [ora.md](ora.md) |
+| ICO, QOI | yes | yes | |
+| Krita `.kra`, GIMP `.xcf`, Procreate | no | no | Not in Photoshop either |
+
+### Affinity import notes (moved from `roadmap.md`)
+
+2026-10-08: Affinity `.af`, `.afdesign`, `.afphoto` and `.afpub` documents (container versions
+8–12) open **natively** ([format notes](../crates/affinity/README.md)): pages and artboards, layers
+and groups, curves and geometric shapes with fills, gradients and strokes, text as type layers,
+placed images as smart objects, pixel layers and masks, as an 8-bit RGB document without a source
+save path. Measured against the thumbnail Affinity embeds in each file, 21 pinned public documents
+(`cargo xtask corpus --affinity`) differ by 0–4.6 of 255 on average. Layer effects, adjustments,
+live filters, brush strokes, special shapes, master pages and CMYK/Lab/16-bit document colour are
+approximated or left out, each with a warning; damaged or unknown files fall back to the embedded
+preview. Affinity writing is not implemented: no Affinity installation was available to check
+written files, so `.af` export stays unsupported.
+
+2026-10-10: current `.af` artboard properties (`phrp`/`aprp`) are recognized alongside the
+legacy flag, including converted curve boards. Synthetic two-board regressions check separate
+bounds, overflow clipping, `.pcraft`/PSD board and pixel round trips, and moving one board with its
+children through undo/redo. Nested boards remain masked groups; rotated/curved board outlines
+still import at their bounding rectangle.
+
+2026-10-09: `corpus/affinity/` now has 39 pinned documents and 21 PNGs (20 rendered references and
+one bitmap-fill texture): the prior 21 public documents plus 18 CC0 samples for #1606. The new
+samples compare against their exported PNGs;
+mean differences range from 0.44/255 (conical gradients) to 86.05/255 (RGB/32 reduced to 8-bit).
+Clouds, hearts, cogs, callouts, arrows, double stars, tears, crescents, diamonds and circular segments
+now import as parametric shapes; the special-shapes sample measures 5.68/255 against its PNG. The
+remaining gaps in #1606 stay tracked with per-sample corpus ceilings.
+
+## Presets and data files
+
+| Format | Photoshop | PhotoCraft | Notes |
+|---|---|---|---|
+| Brushes `.abr` | R/W | R/W | `psd/src/abr.rs` (v6/v12) |
+| Gradients `.grd` | R/W | R/W | |
+| Patterns `.pat` | R/W | R/W | |
+| Swatches `.aco`, `.ase` | R/W | R/W | FILE-215-14 |
+| LUTs `.cube` / `.3dl` / `.look` | R/W (Export Color Lookup) | R/W / R / R | `cms/src/lutfile.rs` |
+| Keyboard shortcuts `.kys` | R/W | R | `engine/src/kys.rs` |
+| ICC profiles | R | R | Own CMS |
+| Actions `.atn` | R/W | no | AUTO-219-2 |
+| Styles `.asl` | R/W | no | #1693 |
+| Contours `.shc`, custom shapes `.csh`, tool presets `.tpl` | R/W | no | FILE-215-15 |
+| Curves `.acv`, Levels `.alv`, other adjustment presets | R/W | not audited | |
+| C2PA Content Credentials | R/W | no | #1050 |
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | major | First version: Photoshop 2026 27.11.0's formats from its bundle, against our codecs; Affinity notes moved from `roadmap.md` |
@@ -0,0 +1,215 @@
+# Where PhotoCraft falls short of Photoshop
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: every known shortfall ranked, from the 2026-10-10 re-measure) · **Target:** Adobe Photoshop 2026
+
+Every known shortfall against Photoshop 2026 (27.11.0), one entry each, ranked by user impact
+on the way to beta. This is the work list agents pick from: take the highest-ranked entry that
+nobody is working on, and when you close part of one, update its evidence and estimate here and
+the matching parity doc. Numbers come from [`target-app-parity.md`](target-app-parity.md);
+checklist ids (`TOOL-…`, `FILE-…`, `UI-…`) are rows in [`scorecard.md`](scorecard.md); `#n` are
+GitHub issues.
+
+No gap blocks alpha: the [alpha gate](roadmap.md#alpha-gate) passes. "Blocks beta" marks what
+stands between alpha and beta.
+
+Hours are Opus 5.5 agent-hours, one agent working sequentially, including tests. **Human** marks
+what an agent can't finish alone.
+
+## Summary
+
+| # | Gap | Kind | Impact | Estimate | Parity doc |
+|---|---|---|---|---:|---|
+| G1 | PSD fidelity and Photoshop acceptance | file format | blocks beta | 150–250 h | [file-format-parity](file-format-parity.md) |
+| G2 | Interactive performance | performance | blocks beta | 200–350 h | [scorecard](scorecard.md#performance) |
+| G3 | User-reported bug backlog | stability, UI | blocks beta | 250–400 h | [target-app-parity](target-app-parity.md) |
+| G4 | Brush feel | UI / painting | high | 100–180 h | [brush-parity](brush-parity.md) |
+| G5 | Transform, handles and modifiers | UI | high | 60–110 h | [ui-parity](ui-parity.md) |
+| G6 | Missing tools (15) | features | high | 60–100 h | [ui-parity](ui-parity.md#tools) |
+| G7 | Workspaces shown as parameter dialogs | features | high | 80–140 h | [target-app-parity](target-app-parity.md#feature-areas) |
+| G8 | Type depth | features | high | 120–200 h | [target-app-parity](target-app-parity.md#feature-areas) |
+| G9 | Panels, preferences and workspace | UI | medium-high | 150–250 h | [ui-parity](ui-parity.md#panels-and-workspace) |
+| G10 | Missing and partial file formats | file format | medium | 120–220 h | [file-format-parity](file-format-parity.md) |
+| G11 | Actions and scripting | ecosystem | medium | 120–200 h | [target-app-parity](target-app-parity.md) |
+| G12 | Pen tablets and display hardware | hardware | medium | 80–150 h | [hardware-parity](hardware-parity.md) |
+| G13 | Languages: Hindi, Arabic, Vietnamese, RTL, engine messages | localization | medium | 80–140 h + human | [localization-parity](localization-parity.md) |
+| G14 | Camera Raw depth | features, formats | medium | 120–200 h | [file-format-parity](file-format-parity.md#camera-raw) |
+| G15 | AI and generative features | AI | medium (high for some users) | 300–600 h + human | [target-app-parity](target-app-parity.md) |
+| G16 | Plug-in compatibility (.8BF, UXP) | ecosystem | medium | 250–450 h + human | [target-app-parity](target-app-parity.md) |
+| G17 | Platforms and distribution | platforms | low-medium | 30–60 h + human | [scorecard](scorecard.md#distribution) |
+
+## Feature gaps
+
+### G1. PSD fidelity and Photoshop acceptance
+- **Missing:** PSDs we write are not always accepted by Photoshop, and Photoshop-authored files
+  don't always render as Photoshop renders them.
+- **Evidence:** corpus floors (`crates/io/tests/corpus.rs`, scorecard): Photoshop-authored
+  oracles 133/256 (52%), io corpus 146/170 (86%), psd-tools 236/309 (76%); round trips ~100%
+  within the corpus. Smart-filter re-render matches 5/30. User reports: a PSD with text layers
+  re-saved unmodified would not reopen (#2374, closed 2026-10-10), Photoshop 26.6 cannot reopen our files (#1281), "PSD not
+  opening in Photoshop" (#2469), a CS2-era smart object unavailable (#2465), PSB smart object
+  blank (#1764), perspective placed layers flattened (#579), PSB > 4 GB fails on Windows (#375).
+  No test opens our exports in a third-party editor (FILE-216-4).
+- **Impact:** PSD is the format professionals exchange. Beta requires it to be reliable.
+- **Estimate:** 150–250 h. **Human:** opening exports in Photoshop on the owner's Mac (or a
+  scripted check there) is the only real oracle.
+
+### G6. Missing tools
+- **Missing (15 of 68):** Artboard, Single Row Marquee, Single Column Marquee, Perspective Crop,
+  Frame, Color Sampler (#1046), Color Replacement (engine command exists), Art History Brush,
+  Freeform Pen and Curvature Pen (#1054), Add Anchor Point, Delete Anchor Point, Convert Point
+  (only as Pen + ⌥), Horizontal Type Mask, Vertical Type Mask.
+- **Evidence:** `Tool` enum (`crates/ui-egui/src/state.rs`) against the 67 `<tool>` entries of
+  Photoshop's `Default Keyboard Shortcuts.kys` plus Remove, Triangle and Curvature Pen.
+- **Impact:** users look for them by name and shortcut; several are daily tools for designers.
+- **Estimate:** 60–100 h (2–9 h each, per the calibration).
+
+### G7. Workspaces shown as parameter dialogs
+- **Missing:** Select and Mask workspace with view modes, Refine Edge brush and outputs
+  (TOOL-214-1, Refine Hair #2671); Content-Aware Fill workspace with a paintable sampling area
+  (TOOL-214-15); Vanishing Point plane editor (TOOL-214-14); Content-Aware Scale through the
+  transform box (TOOL-214-7); Puppet Warp mesh following the outline (TOOL-214-6); Content-Aware
+  Move's Transform On Drop (TOOL-213-4).
+- **Evidence:** scorecard rows above, all partial or missing.
+- **Impact:** these are the tools retouchers and compositors use daily; a slider dialog is not a
+  substitute.
+- **Estimate:** 80–140 h.
+
+### G8. Type depth
+- **Missing:** type on a path (TYPE-218-1), paragraph direction UI (TYPE-218-2), complex-script
+  fallback fonts and golden tests (TYPE-218-3), variable-font axes (TYPE-218-4), OpenType panel
+  UI (TYPE-218-5), manual and optical kerning with PSD round trip (TYPE-218-6, #206), live Warp
+  Text preview (TYPE-218-7, #2689), triple/quadruple click (TYPE-218-8), "rasterize and continue"
+  for distort/perspective on type (TYPE-218-9), text free transform (#2630).
+- **Evidence:** scorecard Type: 0 done, 3 partial, 6 missing. The Type tool froze the app for
+  60 s on one macOS machine (#2518).
+- **Estimate:** 120–200 h.
+
+### G11. Actions and scripting
+- **Missing:** document-independent action recording (AUTO-219-1), a full Actions panel with
+  step toggles and .atn import (AUTO-219-2), replay tests (AUTO-219-4), Batch recursion and
+  naming (AUTO-219-5), ExtendScript/UXP compatibility.
+- **Ahead of Photoshop:** CLI, MCP and the JSON control channel drive every command.
+- **Estimate:** 120–200 h. **Human:** whether to support a JavaScript scripting dialect at all
+  (AGENTS.md forbids JS in the app).
+
+### G14. Camera Raw depth
+- **Missing:** CR3 (detected, unsupported), compressed RAF, lossy DNG (FILE-215-11), .xmp
+  sidecars, masking and calibration in the Camera Raw dialog, per-camera colour for non-DNG
+  cameras, lens profiles.
+- **Evidence:** `crates/raw/src/lib.rs:219` (CR3), scorecard FILE-215-11/12, #50.
+- **Estimate:** 120–200 h.
+
+### G15. AI and generative features
+- **Missing:** Generative Fill and Expand, Generative Upscale, Neural Filters, Sky Replacement,
+  Super Zoom, ML Select Subject/Sky. There is no ML runtime (AUTO-219-7). Classical substitutes
+  exist (GrabCut/max-flow selection, Remove tool by non-local patch completion).
+- **Evidence:** Photoshop's `Required/UXP` ships the Neural Filters gallery and its `.kys` has
+  a Neural Filters taskspace; users ask for local models (#2313, #1996, #2682).
+- **Estimate:** 300–600 h. **Human:** owner decision #41 (model licences, local vs remote).
+
+### G16. Plug-in compatibility
+- **Missing:** hosting Photoshop .8BF filters (#2647), UXP panels, CEP. PhotoCraft has its own
+  sandboxed WebAssembly plug-in ABI v1 (`docs/plugins.md`), but ABI v2 (panels, progress, other
+  layers) is missing (AUTO-219-6).
+- **Estimate:** 250–450 h. **Human:** .8BF needs native code under the isolated-`unsafe` rule.
+
+## UI / UX gaps
+
+### G3. User-reported bug backlog
+- **Missing:** fixes for 579 open issues (2026-10-10), filed mostly by 0.5/0.6 users. By title:
+  ~150 UI/panels/menus/shortcuts, 94 file/format, 87 layers, 54 painting, 54 selection,
+  46 type, 41 performance, 34 transform/move, 25 Linux/Wayland, 24 Windows, 20 macOS.
+- **Impact:** each is a place where a Photoshop user hits something different from what they
+  expect; together they are why ready-for-real-work is ~45% while breadth is ~76%.
+- **Estimate:** 250–400 h (many overlap with other gaps; count each fix once).
+
+### G4. Brush feel
+- **Missing:** lag with big brushes (#2619, #1994) and under the cursor (#2566, #1343); jumpy
+  pressure on Windows (#2046) and tap sensitivity (#1933); stair-stepped line weights (#1776),
+  interpolation and smoothing problems (#2138), edge falloff (#2149), pixelated cursor vs
+  Photoshop (#1433), brush rotation not driven by pen rotation (#1301); live preview only for
+  Brush and Eraser (TOOL-213-7); no image file as a brush tip (TOOL-213-12); per-control pressure
+  curves (TOOL-213-11); 4 of Photoshop's symmetry modes.
+- **Detail:** [`brush-parity.md`](brush-parity.md).
+- **Estimate:** 100–180 h. **Human:** checking feel on real tablets.
+
+### G5. Transform, handles and modifiers
+- **Missing:** right-click menu during Free Transform (TOOL-214-5); commit/cancel buttons in the
+  options bar (#2442); H/V skew fields and the full interpolation list; transform keeps layer
+  order and blend mode in its preview (#2439); Hand tool interrupting a transform (#2302);
+  blend mode/opacity edits during a transform (#1384); locked layers refused before the drag
+  (#2585); ⌘-drag temporary Move without box or snapping (#2681), Auto-Select box follows the
+  previous layer (#2666); ⌘-arrow layer nudge (#2474); path-point nudge; polygonal lasso Shift
+  angles (#2352); autoscroll at the edge while selecting (#2231); hold-a-key spring-loaded tools
+  (#2672); double-click Zoom tool to 100% (#2508); wheel and pinch behaviour (#2485, #2463,
+  #1989); toolbar press-and-hold flyouts (#2301); tool-key memory in a group (#2608).
+- **Detail:** [`ui-parity.md`](ui-parity.md).
+- **Estimate:** 60–110 h.
+
+### G9. Panels, preferences and workspace
+- **Missing:** 46 of 149 preferences do nothing (scorecard prefs audit); panel groups can't
+  float or tear off into OS windows (UI-217-5, #1828, #1213, #2305); Color panel lacks wheel and
+  slider modes (UI-217-17, #2569); UI scale presets (UI-217-9, #2492); more themes (UI-217-11,
+  #2661); Stamp Visible (UI-217-22); layer clipboard
+  (UI-217-23, #2670); tool-specific context menus beyond brushes, selections and Pen
+  ([context-menu-parity.md](context-menu-parity.md)); Libraries, Comments, Contextual Task Bar
+  panels.
+- **Estimate:** 150–250 h.
+
+### G2. Interactive performance
+- **Missing:** 22 of 25 budgeted scenarios are over budget (baseline 2026-10-05, a heavily
+  loaded machine): layer move with styles 321 ms p50 against 25 ms, opacity change 276 ms
+  against 20 ms, paste 12 MP 1.2 s against 100 ms, layer move on 15000×10000 16-bit 2.7 s,
+  Content-Aware Scale 24 MP 954 s against 3 s, Select Subject 559 ms against 100 ms; the
+  150-layer nudge crashes the GPU compositor (P12); a 50 MP PSB with feathered masks redraws in
+  ~10 s (#2377); menus slow with thousands of layers (#2460); old machines lag (#2687).
+- **Estimate:** 200–350 h. Serial in places (compositor architecture).
+
+## File-format gaps
+
+### G10. Missing and partial file formats
+- **Missing:** JPEG Options dialog (FILE-215-19), Photoshop PDF read and write and multi-page PDF
+  (FILE-215-10, #2541), AVIF read (#463) and default-build AVIF write, JPEG XL, JPEG 2000 (#1047),
+  DPX/Cineon (#1049), DICOM, EPS/DCS, PCX, Pixar, IFF, WBMP, MPO/JPS, PICT, Scitex CT, Photoshop
+  Raw; layered TIFF extra alpha channels (FILE-215-7); spot plates (FILE-215-17); .atn, .asl,
+  .shc, .csh presets (FILE-215-15); TGA/PNG alpha channels on open (#2225); C2PA (#1050).
+- **Detail:** [`file-format-parity.md`](file-format-parity.md).
+- **Estimate:** 120–220 h.
+
+## Hardware gaps
+
+### G12. Pen tablets and display hardware
+- **Missing:** tilt and rotation on Windows (pressure only, via WM_POINTER; no WinTab); native
+  Wayland tablet-v2 (#79; the pen does not click on GNOME 50, #2622, KDE #2297); eraser end on
+  flipped Wacom pens (#2285); HDR/EDR output (Photoshop has "Precise Color Management for HDR
+  Display"); monitor profile detection off macOS and live profile changes; Touch Bar (macOS
+  Intel only, low value); 3D mice.
+- **Detail:** [`hardware-parity.md`](hardware-parity.md).
+- **Estimate:** 80–150 h. **Human:** tablets and HDR displays to test on.
+
+## Localization gaps
+
+### G13. Hindi, Arabic, Vietnamese, RTL and engine messages
+- **Missing:** catalogs for Hindi, Arabic and Vietnamese (3 of the 12 key languages); right-to-
+  left UI layout; Devanagari/Thai/Khmer fallback fonts; engine error and status messages in
+  English in every language; browser-locale detection and CJK fonts on the web; Japanese IME on
+  Windows (#590); installed Japanese fonts missing from the font list (#2330); Arabic layer names
+  as tofu (#1408, #1693).
+- **Detail:** [`localization-parity.md`](localization-parity.md).
+- **Estimate:** 80–140 h. **Human:** native-speaker review of every catalog (none is recorded).
+
+## Platform gaps
+
+### G17. Platforms and distribution
+- **Missing:** Windows code signing (material not obtained), Windows ARM64 (DIST-220-2), in-app
+  update check (DIST-220-1, #2650, #2375), in-app bug report (DIST-220-3), Flathub (#173),
+  starting without a GPU adapter (#2519, #2412), Windows 7 (#2310, out of scope), Intel Macs with
+  OpenCore (#1859).
+- **Estimate:** 30–60 h. **Human:** signing certificates.
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | minor | Checked against the alpha gate (no alpha blockers); #2374 marked closed |
+| 2026-10-10 | major | First version, from the full re-measure against Photoshop 2026 27.11.0 and the open issues |
@@ -0,0 +1,64 @@
+# Hardware parity
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: GPU, pen tablets, displays, gestures per platform against Photoshop 2026) · **Target:** Adobe Photoshop 2026
+
+Each hardware feature Photoshop 2026 (27.11.0) uses, per platform, against PhotoCraft. Photoshop
+side: its bundle (Metal and OpenCL paths, "Precise Color Management for HDR Display" and EDR
+strings, Touch Bar resources, pen pressure/tilt strings) and Adobe's GPU FAQ. PhotoCraft side:
+`crates/tablet`, `crates/ui-egui/src/stylus.rs`, `apps/photocraft/src/gpu_startup.rs`,
+`engine/src/display_color.rs`, `apps/photocraft/src/monitor_profile.rs`.
+
+**Summary (estimated):** ~50% ready. GPU compositing and colour-managed display are solid; pen
+input is uneven by platform (Windows lacks tilt/rotation, Wayland is broken) and there is no HDR
+output. Remaining: 80–150 h (gap G12). Human: tablets (Wacom, XP-Pen, Huion) on each OS and an
+HDR display to verify.
+
+## GPU
+
+| Feature | Photoshop | PhotoCraft | Status |
+|---|---|---|---|
+| GPU canvas, zoom/pan, rotate view | Metal (macOS), DirectX/OpenCL (Windows) | wgpu 30: Metal, Vulkan, DX12, GL; backend selectable | done |
+| GPU compositing of layers, blend modes, effects | partial (Photoshop composites mostly on CPU) | wgpu compositor, ≤ 1/255 vs CPU oracle; Multichannel and > texture-limit regions fall back to CPU | done (ahead in places) |
+| GPU filters (Blur Gallery, Liquify, Smart Sharpen, Camera Raw) | yes | CPU (rayon) for filters; GPU for compose and effects | partial |
+| Documents larger than the GPU texture limit | yes (tiled) | CPU fallback; tiling in progress (#49) | partial |
+| CPU-only fallback when no GPU | yes, reduced | `cpu` canvas preference; fails to start without an adapter on some Linux systems (#2519, #2412) | partial |
+| Old / integrated GPUs | yes | black line on Intel HD (#2426), OpenCore Macs fail (#1859), Rgba32Float panic (#2020) | partial |
+
+## Pen tablets
+
+| Feature | Photoshop | macOS | Windows | Linux X11 | Linux Wayland | Web |
+|---|---|---|---|---|---|---|
+| Pressure | yes | yes (AppKit monitor) | yes (WM_POINTER), jumpy (#2046) | yes (XInput2) | via Xwayland only; GNOME 50 pen doesn't click (#2622), KDE (#2297) | yes |
+| Tilt | yes | yes | **no** | yes | no | yes |
+| Barrel rotation | yes | yes (not applied to tip angle, #1301) | no | partial | no | yes (twist) |
+| Eraser end | yes | yes | no (#2285) | partial | no | partial |
+| WinTab driver mode | yes | – | no | – | – | – |
+| Pressure curve | Windows Ink / driver | global curve in Preferences (#2195) | same | same | same | same |
+
+## Displays and colour
+
+| Feature | Photoshop | PhotoCraft | Status |
+|---|---|---|---|
+| Colour-managed canvas (document → monitor profile) | yes | yes, folded into the GPU 3D LUT | done |
+| Monitor profile detection | per display, live | macOS per display at launch; others sRGB or a chosen .icc | partial |
+| HDR / EDR output for 32-bit and HDR images | yes ("Precise Color Management for HDR Display") | no; 8-bit sRGB-encoded canvas texture | missing |
+| High-DPI, physical 100% | yes | yes (#1943) | done |
+| Multiple monitors: documents in separate OS windows | yes | yes (`extra_windows`) | done |
+| Panels in separate OS windows | yes | in-app floating only (UI-217-5) | missing |
+
+## Input devices
+
+| Feature | Photoshop | PhotoCraft | Status |
+|---|---|---|---|
+| Trackpad pinch zoom, two-finger pan | yes | yes; pinch scrolls instead in 0.6.0 (#2463) | partial |
+| Trackpad rotate (Rotate View) | yes | yes (`rotate_view.rs`) | done |
+| Touch Bar | yes (Intel Macs) | no | missing (low value) |
+| Touch screens (Windows) | yes | basic pointer only | partial |
+| 3D mouse, MIDI, Surface Dial | Surface Dial on Windows | no | missing (low value) |
+| Scanners (WIA / TWAIN) | WIA menu (Windows) | menu item live; no driver integration | partial |
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | major | First version |
@@ -0,0 +1,64 @@
+# Localization parity
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: per-language status measured from the catalogs, against Photoshop 2026's language packs) · **Target:** Adobe Photoshop 2026
+
+Per-language status of PhotoCraft's UI. How catalogs work and how to add one:
+[`localization.md`](localization.md). Language notes: [Korean](localization-ko.md),
+[Simplified Chinese](localization-zh-hans.md).
+
+**How measured.** `cargo xtask i18n-coverage` (`xtask/src/i18n_coverage.rs`) defines the key
+universe: `tl!("…")` literals and `trn(` plurals in `ui-egui` (tests excluded), menu-catalog
+labels and path segments, `menus.rs` UI commands, and engine `CommandSpec` labels and menu paths:
+**1,608 English keys** on 2026-10-10. Each catalog (`crates/ui-egui/src/i18n/<code>.tsv`) was
+checked against that universe by a source script replicating the xtask (no build). Catalogs hold
+more rows than the universe because they also translate dynamic `tl!(expr)` strings (~159 call
+sites the tool can't enumerate). **Not covered anywhere:** engine error and status messages and
+filter parameter help, which are English in every language; so no language is `full`.
+
+**Photoshop 2026** ships UI language packs for cs, da, de, en, es, fi, fr (and fr-CA), hu, it,
+ja, ko, nb, nl, pl, pt-BR, ru, sk, sv, tr, uk, zh-Hans, zh-Hant (24 `.lproj` markers; one pack is
+installed per Creative Cloud install, here `en_US`). It has no Hindi, Arabic, Indonesian or
+Vietnamese UI; Arabic and Hebrew users get English "Middle East" builds with the World-Ready
+composer, and its text engine handles RTL, Devanagari and Indic scripts.
+
+**Summary.** 7 of the 12 key languages at 100% of measured keys (status `partial` because of the
+engine messages), 3 absent (Hindi, Arabic, Vietnamese), plus 8 other languages. Weighted by
+speakers, ~55% ready. Remaining: 80–140 h, plus native-speaker review (none recorded for any
+catalog).
+
+## The twelve key languages
+
+| Language | Code | UI strings (measured keys) | Catalog rows | Dialogs / tooltips / help | Script support | Native review | Status | To `full` |
+|---|---|---|---:|---|---|---|---|---:|
+| English | `en` | 1,608 / 1,608 (100%) | source | all | Latin | – | full | 0 h |
+| Simplified Chinese | `zh-hans` | 1,608 / 1,608 (100%) | 2,366 | dialogs, tooltips yes; engine messages English | CJK fonts (lazy loader), IME preedit/commit, vertical type | no | partial | 8–12 h |
+| Spanish | `es` | 1,608 / 1,608 (100%) | 2,413 | as above | Latin | no | partial | 6–10 h |
+| Hindi | `hi` | 0 (0%) | – | – | Devanagari shaping in the text engine (parley/HarfRust); no Devanagari fallback font (TYPE-218-3) | no | none | 20–30 h |
+| Arabic | `ar` | 0 (0%) | – | – | Arabic shaping and bidi in the text engine, Arabic fallback font; no RTL UI layout (DIST-220-4), paragraph direction has no UI (TYPE-218-2), layer names as tofu (#1408, #1693) | no | none | 40–60 h |
+| French | `fr` | 1,608 / 1,608 (100%) | 2,766 | as above | Latin | no | partial | 6–10 h |
+| Portuguese (Brazil) | `pt-br` | 1,608 / 1,608 (100%) | 2,788 | as above | Latin; every `pt` locale uses it | no | partial | 6–10 h |
+| Indonesian | `id` | 1,608 / 1,608 (100%) | 2,764 | as above | Latin | no | partial | 6–10 h |
+| Japanese | `ja` | 1,608 / 1,608 (100%) | 3,356 | as above | CJK fonts, IME (Windows IME broken, #590), vertical type (tategaki); installed Japanese fonts missing from the list (#2330) | no | partial | 10–16 h |
+| German | `de` | 1,608 / 1,608 (100%) | 2,400 | as above | Latin | no | partial | 6–10 h |
+| Korean | `ko` | 1,608 / 1,608 (100%) | 2,936 | as above; terminology from Photoshop's Korean UI ([notes](localization-ko.md)) | Hangul fonts, IME | no | partial | 6–10 h |
+| Vietnamese | `vi` | 0 (0%) | – | – | Latin with stacked diacritics (shaping supported) | no | none | 12–18 h |
+
+## Other shipped languages
+
+8 more, each at 1,608 / 1,608 measured keys, status `partial` for the same reason: Traditional
+Chinese `zh-hant` (2,329 rows), Russian `ru` (2,544), Ukrainian `uk` (3,375), Czech `cs` (2,337),
+Polish `pl` (3,366), Greek `el` (3,481), Dutch `nl` (3,568), Italian `it` (2,411).
+
+## Cross-language gaps
+
+- Engine error, status and parameter-help messages are English in every language.
+- No right-to-left UI layout (needed for Arabic, Hebrew, Persian, Urdu).
+- Web build: no browser-locale detection, no CJK web fonts.
+- WebAssembly plug-ins disappear from Filter › Plug-ins when the UI isn't English (#2540).
+- A user couldn't find the language option (#2532): it is in Preferences › Interface.
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | major | First version, measured from the catalogs with the `i18n-coverage` rules |
@@ -1,92 +0,0 @@
-# Feature-parity estimate and remaining effort
-
-How close PhotoCraft is to Adobe Photoshop 2026, and how much Opus 5.5 work remains. Generated by
-assessment, not a tool; revisit when a major arc lands. Menu coverage is from `cargo xtask parity`;
-fidelity is from the PSD oracle (`cargo xtask corpus` / `oracle_diff`).
-
-## Headline
-
-> **Update 2026-10-03:** menu/command breadth reached **100% (625/625)** — every Photoshop
-> menu item now dispatches a live command (incl. Variables/Data Sets, Trap, the Timeline panel and
-> Layer › Video Layers). Remaining parity work is depth and fidelity, not breadth: PSD oracle
-> (~66%), Camera Raw/video-layer depth, and the AI/Neural tier. See the table below.
-
-
- **Menu / command breadth:** 625 / 625 Photoshop menu items live = **100%** (every menu item has a live command).
- **PSD rendering fidelity (oracle):** ~111 / 170 corpus files pixel-match = **~65%**.
- **Overall functional parity (weighted, mainstream workflows):** **~86%** (menu breadth complete; depth + fidelity are the remaining gap).
- **Overall including Adobe's AI / Firefly features:** **~75%** (these are mostly cloud/ML and the
-  hardest to reach clean-room).
-
-Menu coverage (96%) overstates true parity because many live commands are faithful-but-approximate
-(Camera Raw without masking/calibration, Vanishing Point without its UI, heuristic Select Subject/Sky
-vs Adobe's ML, Content-Aware Fill basic). The weighted figure accounts for depth and fidelity, not
-just presence.
-
-## Where we stand, by area
-
-| Area | Parity | Notes |
-|---|---:|---|
-| Layers, masks, groups, blend modes, smart objects, layer styles | ~92% | Broad + GPU-accelerated; bevel/satin shapes and a few blend edge cases remain |
-| Selections, paths, vector shapes, pen | ~90% | Select menu 100% of items; Select Subject/Sky are heuristic, not ML |
-| Paint / brush engine, retouch (clone, healing) | ~85% | Solid; no Mixer Brush depth, no Content-Aware healing |
-| Adjustments + adjustment layers | ~90% | All kinds present; Camera Raw depth and a few toe-curve models approximate |
-| Filters (incl. Filter Gallery ×47, Liquify, blurs, render) | ~88% | Gallery looks are approximations; no Neural Filters |
-| Type / text engine, character & paragraph styles, glyphs | ~90% | Vertical type laid out and rendered (#199; no mojikumi or upright-Roman option yet); OpenType alternates thin |
-| Color management (ICC v2/v4, intents, soft-proof) | ~90% | Own pure-Rust CMS; no Adobe CMYK profiles (synthetic) |
-| File I/O: PSD, native .pcraft, flats (png/jpg/tiff/webp/exr/…) | ~85% | PSD oracle ~65% pixel-exact; most structure round-trips |
-| Transform, warp, puppet, perspective, liquify | ~88% | No Face-Aware Liquify (needs a landmark model) |
-| Automation: CLI, MCP, headless server, control channel | ~95% | Ahead of Photoshop here (see below) |
-| Image modes (RGB/Gray/CMYK/Lab/Indexed/Duotone/Multichannel) | ~85% | Multichannel display-only; no ink-plate model |
-| Print / export / Save for Web / slices / Generator | ~85% | No Separations/bleed; droplets basic |
-| Computational photo (Photomerge, HDR, Lens, Camera Raw, Wide Angle) | ~70% | Classical implementations; generic lens profiles only |
-| Data-driven graphics (Variables, Data Sets) | ~0% | Not started (5 menu items) |
-| Video / Timeline / animation | ~0% | Not started (13 menu items) — a whole subsystem |
-| AI / Neural / Generative (Firefly, Sensei, Neural Filters) | ~5% | Mostly absent; needs licensed/trained models |
-| Prepress niche (Trap, measurement scale pro) | ~30% | Trap, ink spread not modelled |
-
-## The remaining 25 menu items, grouped
-
- **Video & Timeline (13):** Video Layers (new/insert/duplicate/delete/replace/interpret/restore/
-  reload/rasterize), Rasterize › Video, Timeline panel, Render Video, Import Video Frames to Layers.
-  One cohesive subsystem: a frame/time model, a video decoder/encoder (clean-room Rust, no ffmpeg),
-  onion-skinning, and a timeline UI.
- **Data-driven graphics (5):** Variables › Define / Data Sets, Apply Data Set, Export Data Sets as
-  Files, Import Variable Data Sets. A template-binding model over layers + a CSV/data-set engine.
- **Niche / platform (7):** Trap (prepress ink spread), Frame from Layers, Migrate Presets, WIA
-  Support (Windows scanners), and two already-adjacent items.
-
-## Effort to parity (Opus 5.5)
-
-Two clocks: **agent-hours** (total model work) and **wall-clock** (with this session's observed
-~6–8× parallelism across independent arcs). This session landed ~70 menu items + major subsystems
-in a handful of wall-clock hours using 8 parallel agents, so wall-clock ≈ agent-hours ÷ 6.
-
-| Milestone | Scope | Agent-hours | Wall-clock |
-|---|---|---:|---:|
-| **To 90% functional** | remaining non-AI/non-video menu items (Variables, Trap, Frame, Migrate, WIA), Camera Raw depth, Vanishing Point UI, selection quality, polish | 60–90 | 10–15 h |
-| **To 95% functional** | the above + PSD oracle 65% → ~90% (bevel/satin shapes, Lab blending, noise-gradient RNG, toe curves), broad UI-parity pass, perf pass | +80–130 | +15–22 h |
-| **Video / Timeline subsystem** | frame model, clean-room decode/encode, timeline, animation, Render Video | 50–80 | 10–14 h |
-| **Credible AI tier** | ML Select Subject/Sky, Content-Aware Fill (PatchMatch++), basic generative fill/expand — needs model integration; some features may be infeasible fully clean-room without licensed models | 120–250+ | 25–45 h |
-
-**Totals:**
- To **~95% functional parity** (everything except video and AI): **~140–220 agent-hours ≈ 25–40
-  wall-clock hours** of Opus 5.5 with parallel agents.
- To **~98%** incl. the video subsystem: **+50–80 agent-hours ≈ +10–14 wall-clock hours**.
- To **~99%** incl. a credible AI tier: **+120–250 agent-hours ≈ +25–45 wall-clock hours**, with the
-  caveat that pixel-for-pixel Firefly/Sensei parity is not achievable clean-room without trained
-  models; we target functional equivalents.
-
-**Best single number:** PhotoCraft is **~82% of the way** to full Photoshop parity, and reaching a
-**95%, ships-for-most-users** state is roughly **25–40 wall-clock hours** of parallel Opus 5.5 work.
-Closing the last 5% (video + AI) is the long, uncertain tail.
-
-## Assumptions & caveats
-
- "Parity" = functional + visual equivalence of what a user can do, not byte-identical output. Clean-
-  room means no Adobe code, profiles, shaders, or trained models; AI features target equivalents.
- Wall-clock assumes 6–8 independent arcs in parallel with their own target dirs, as this session ran.
-  Serial work is ~6× longer.
- Adobe removed 3D/Substance from Photoshop, so 3D is intentionally out of scope.
- Fidelity hours are the least certain: the remaining oracle failures are proprietary RNGs and
-  unpublished curve models fitted by observation.
@@ -1,14 +1,41 @@
 # Roadmap

-Status legend: ✅ done · 🟡 in progress · ⬜ not started. Updated 2026-10-08.
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** minor (alpha gate added; earlier the same day: major, honest assessment moved to `target-app-parity.md`; milestones re-dated; new Current focus and What's next with estimates) · **Target:** Adobe Photoshop 2026

-**Parity metrics.** `cargo xtask parity` measures how much of Photoshop's menu tree is *wired to a
-command* and writes [`parity.md`](parity.md). It does **not** measure whether those commands behave
-like Photoshop, feel right, or survive real files. Read it next to the PSD composite oracle, the
-test count, and the [honest parity assessment](#honest-parity-assessment-2026-10-05) below, which
-is the reference answer to "how close are we really". The weighted feature estimate in
-[`parity-estimate.md`](parity-estimate.md) (~82%) counts feature surface and overstates user-facing
-readiness; treat it as an upper bound.
+Forward-looking plan: milestones, the **Current focus** list, and what's next with estimates.
+Where we stand today is in [`target-app-parity.md`](target-app-parity.md) (two numbers, by
+dimension and area) and the ranked work list in [`gaps.md`](gaps.md); the one-page summary is
+[`ROADMAP.md`](../ROADMAP.md). Measured numbers: [`scorecard.md`](scorecard.md)
+(`cargo xtask scorecard`: performance budgets, corpus floors, per-area checklists, dead
+preferences) and [`parity-checklist.md`](parity-checklist.md) (`cargo xtask parity`: menu wiring,
+628/628, which says nothing about behaviour).
+
+Status legend: ✅ done · 🟡 in progress · ⬜ not started. Hours are Opus 5.5 agent wall-clock hours
+for one agent working sequentially (calibration in
+[`target-app-parity.md`](target-app-parity.md#remaining-effort-and-calibration)).
+
+**Stage: alpha** (ready for real work ~45%). Beta needs ~75% and reliable PSD: ~1,100–1,800 h.
+
+## Alpha gate
+
+The core workflows a typical Photoshop professional runs every day, checked end to end on the main
+platform (macOS, 0.6.0), including saving and reopening the work. Any **no**, or a **partial**
+that blocks the workflow, would make PhotoCraft pre-alpha.
+
+| Core workflow | Works end to end? | Evidence | Hours to pass |
+|---|---|---|---:|
+| Photo retouch: open JPEG/raw, heal/clone/remove, adjustment layers, export JPEG, save and reopen | yes | Spot Healing, Healing, Clone, Patch, Remove tools; 16 adjustment layers; raw for DNG/CR2/NEF/ARW/RAF; JPEG export; MCP workflow tests (`automation/tests/agent_tasks.rs`). Quality complaints (#1092, #2104) and no CR3 don't stop the workflow | 0 |
+| Composite: layers, masks, selections, blend modes, smart objects; save as PSD and `.pcraft` and reopen | yes | PSD round trip ≥ 169/170, 308/309, 258/256 corpus files; `.pcraft` atomic save and crash recovery. Multi-layer Smart Object conversion (#2604) and mask targeting (#2458) are partial but have workarounds | 0 |
+| Cut-out: select a subject, refine the edge, mask, export PNG with transparency | partial (not blocking) | Object/Quick Selection, Select Subject (classical), Refine Edge dialog, layer masks, PNG export. Select and Mask is a dialog, not the workspace; no Refine Hair (#2671) | 0 (depth: 80–140 h, G7) |
+| Design: type, shapes, layer styles, artboards; export PNG/JPEG; save and reopen | partial (not blocking) | Point/paragraph type, shapes with stroke controls, all 10 layer effects, artboards, Export As. One macOS report of a 60 s freeze on choosing the Type tool (#2518, not reproduced); no type on a path | 0 (fix #2518: 2–6 h) |
+| Paint with a pen tablet: pressure brushes, layers, save and reopen | partial (not blocking) | Pressure, tilt, rotation on macOS (AppKit monitor); every Brush Settings section. Lag with big brushes (#2619) and feel gaps ([brush-parity.md](brush-parity.md)) | 0 (feel: 100–180 h, G4) |
+| Exchange PSDs with Photoshop users: open a client PSD, edit, send a PSD back | partial (not blocking for alpha; blocking for beta) | Photoshop-authored oracle 133/256 render as Photoshop does; most exports open in Photoshop (smart objects verified), but some don't (#1281, #2469) | 0 for alpha; 150–250 h for beta (G1) |
+
+**Result: passes.** Every core workflow completes end to end and the work saves and reopens in
+PhotoCraft; the partial rows are depth and fidelity gaps with workarounds, not blockers. Ready for
+real work is ~45%, inside the alpha band (~40–75%), but close to its floor.
+
+## Milestones

 | M | Status | Where we are |
 |---|---|---|
@@ -16,248 +43,44 @@ readiness; treat it as an upper bound.
 | M1 Foundation | ✅ | geom, color (27 blend modes), raster (COW tiles, any depth), doc, ops, cms (ICC) |
 | M2 PSD v1 | ✅ | photocraft-psd: 134/135 real files byte-exact round trip |
 | M3 Viewer app | ✅ | egui shell (Pro / Studio / Classic themes), 13+ codecs, native and web (trunk) builds |
-| M4 Native format + engine | ✅ | 500+ commands, `.pcraft` (incremental, autosave, crash recovery), CLI, persistent preferences |
+| M4 Native format + engine | ✅ | ~700 commands, `.pcraft` (incremental, autosave, crash recovery), CLI, persistent preferences |
 | M5 GPU compositor | 🟡 | wgpu compositor drives the canvas, layer effects, vector masks, artboards, pattern fills and every clip case included (≤1/255 vs CPU); Multichannel documents fall back to the CPU |
 | M6 Paint + select | 🟡 | brush engine, all selection tools, multi-layer selection, snapping + smart guides, free transform + warp; stylus pressure on Windows (WM_POINTER via winit), the web (Pointer Events, with tilt/twist), macOS (AppKit event monitor: pressure, tilt, rotation, eraser end) and Linux X11 (XInput2 raw valuators); on Linux Wayland a pen opens the window through Xwayland (native tablet-v2 still open, #79) |
 | M7 Adjust + filters | 🟡 | 16 adjustment layers + destructive-only adjustments, 70+ filters incl. Blur Gallery, Actions record/replay, Fade |
-| M8 PSD v2 | 🟡 | adjustments (incl. Selective Color, Color Lookup), fills, effects, patterns, text, shapes, smart objects, alpha channels; oracle 111/170 |
+| M8 PSD v2 | 🟡 | adjustments (incl. Selective Color, Color Lookup), fills, effects, patterns, text, shapes, smart objects, alpha channels; oracle floors 146/170 (io), 236/309 (psd-tools), 133/256 (Photoshop-authored) |
 | M9 Text, vector, styles | 🟡 | type engine + Warp Text, shapes / pen / paths, all 10 effects on CPU and GPU (parity ≤1/255, 30/31 corpus effect files on the GPU) |
 | M10 Smart features | 🟡 | classical Select Subject / Object, content-aware fill and scale, healing, auto-align / auto-blend; ML backend not started |
 | M11 Automation + formats | 🟡 | MCP (headless + live bridge), batch, Image Processor, prefs over MCP; DoD test passes (10 agent tasks over MCP, `automation/tests/agent_tasks.rs`); since 2026-10-07 the CLI rejects unknown flags and answers `<subcommand> --help`, batch runs report same-name outputs instead of overwriting them, MCP `command_batch` steps can start background jobs, and headless saves never flatten over the opened file; JP2 / DICOM / DPX / C2PA pending |
 | M12 Pro parity | 🟡 | CMYK / Lab / Indexed / Bitmap / Duotone, ICC + soft proofing, channels + Quick Mask, smart-object stack modes, artboards, layer comps; print, HDR, photomerge, timeline pending |

-**Menu parity: 532 / 625 (85.1%)** on 2026-10-01, up from 224 (35.8%) the day before. See [`parity.md`](parity.md).
+Releases: 0.1.0 (2026-10-02), 0.2.0 (2026-10-05, first signed and notarized), 0.3.0 (10-07),
+0.5.0 (10-08), 0.6.0 (10-10).

-2026-10-08: Affinity `.af`, `.afdesign`, `.afphoto` and `.afpub` documents (container versions
-8–12) open **natively** ([format notes](../crates/affinity/README.md)): pages and artboards, layers
-and groups, curves and geometric shapes with fills, gradients and strokes, text as type layers,
-placed images as smart objects, pixel layers and masks, as an 8-bit RGB document without a source
-save path. Measured against the thumbnail Affinity embeds in each file, 21 pinned public documents
-(`cargo xtask corpus --affinity`) differ by 0–4.6 of 255 on average. Layer effects, adjustments,
-live filters, brush strokes, special shapes, master pages and CMYK/Lab/16-bit document colour are
-approximated or left out, each with a warning; damaged or unknown files fall back to the embedded
-preview. Affinity writing is not implemented: no Affinity installation was available to check
-written files, so `.af` export stays unsupported.
+## Current focus

-2026-10-10: current `.af` artboard properties (`phrp`/`aprp`) are recognized alongside the
-legacy flag, including converted curve boards. Synthetic two-board regressions check separate
-bounds, overflow clipping, `.pcraft`/PSD board and pixel round trips, and moving one board with its
-children through undo/redo. Nested boards remain masked groups; rotated/curved board outlines
-still import at their bounding rectangle.
+Ranked; each item links to its gap entry. Pick from the top unless an issue is assigned to you.

-2026-10-09: `corpus/affinity/` now has 39 pinned documents and 21 PNGs (20 rendered references and
-one bitmap-fill texture): the prior 21 public documents plus 18 CC0 samples for #1606. The new
-samples compare against their exported PNGs;
-mean differences range from 0.44/255 (conical gradients) to 86.05/255 (RGB/32 reduced to 8-bit).
-Clouds, hearts, cogs, callouts, arrows, double stars, tears, crescents, diamonds and circular segments
-now import as parametric shapes; the special-shapes sample measures 5.68/255 against its PNG. The
-remaining gaps in #1606 stay tracked with per-sample corpus ceilings.
+1. **PSD that Photoshop trusts** ([G1](gaps.md#g1-psd-fidelity-and-photoshop-acceptance)):
+   fix Photoshop refusing our files (#1281, #2469; text-layer case #2374 closed); add a
+   Photoshop re-open check (FILE-216-4); Photoshop-oracle floor 133/256 → 200+. 150–250 h.
+2. **Performance budgets** ([G2](gaps.md#g2-interactive-performance)): the 22 over-budget
+   scenarios, the P12 GPU crash at 150 layers, brush and cursor lag (#2619, #2566). 200–350 h.
+3. **User bug backlog** ([G3](gaps.md#g3-user-reported-bug-backlog)): 579 open issues, most from
+   0.5/0.6 users; triage weekly, fix regressions first. 250–400 h.
+4. **Brush and transform feel** ([G4](gaps.md#g4-brush-feel), [G5](gaps.md#g5-transform-handles-and-modifiers)):
+   [`brush-parity.md`](brush-parity.md) and [`ui-parity.md`](ui-parity.md) rows marked partial
+   or missing. 150–250 h.
+5. **Missing tools** ([G6](gaps.md#g6-missing-tools)): 15 of Photoshop's 68. 60–100 h.

-## Honest parity assessment (2026-10-05)
+## What's next, with estimates

-This is the reference answer to "how close are we to Photoshop parity, really". Agents: read it
-before picking work. Update it (with dated measurements) when the numbers move; don't restate the
-menu-parity number in its place.
-
-**Measured numbers live in the [scorecard](scorecard.md)** (`cargo xtask scorecard`, kept current
-by CI): performance scenarios against their budgets, corpus floors, per-area checklists with
-done / partial / missing counts, and the number of settings that do nothing. Where the scorecard
-and the estimates below disagree, the scorecard's numbers supersede them. As of 2026-10-05 it
-counts 67 of 129 preferences that nothing reads; 3 of the 25 budgeted performance scenarios meet
-their #209–#211 targets (first baseline, taken on a heavily loaded machine), and the 150-layer
-nudge scenario crashes the GPU compositor; the open area issues' checklists (#203–#220) are
-almost entirely missing or partial.
-
-2026-10-07: UI Font Size now applies to interface text, including CJK fallbacks, independently
-of display and canvas zoom (#532). The current preference audit drops from 59 to 58 unread
-settings out of 135; see the regenerated scorecard.
-
-2026-10-09: Output-preserving Radial Blur direct sampling measured on a local
-i7-9750H, 12 workers, 24 MP RGBA8, Good quality, amount 1: Spin 66.10 → 16.43 s
-(4.02×), Zoom 19.19 → 5.35 s (3.59×), one paired run each, exact output equality.
-These are local measurements at amount 1; see [method and limits](radial-blur-performance.md).
-
-2026-10-10: Iris Blur pin preparation, saturated-distance evaluation and trailing
-Gaussian-level halo trimming measure 1.18–1.76× faster across eight local 24 MP
-RGB U8/U16/F32 cases (Ryzen 7 9700X, Windows, eight workers). Default blur 15:
-626.8 → 426.9 ms; two pins: 1789.9 → 1016.2 ms. The default 1500×1000 filtered
-proxy for 24 MP, document blur 15/80: 28.7/51.8 → 24.9/42.6 ms. At 36 MP,
-two pins: 2595.8 → 1822.8 ms; small/medium U8 gains are inconclusive.
-Differential tests preserve exact finite pixel bits.
-These are computation timings, not GUI frame latency or new Photoshop parity;
-see [measurements, upstream SIMD findings and limits](iris-blur-performance.md).
-
-2026-10-10: native selection distance transforms measured 7.59–10.57x faster on six
-24–36 MP synthetic masks on an AWS c7i.4xlarge (16 workers, three paired release runs).
-The transform also corrects f32 envelope errors past coordinate 4096, including nonzero
-distances at selected pixels. The parallel path needs an additional four bytes per pixel;
-see [measurements, correctness and limits](selection-distance-performance.md).
-
-2026-10-10: Quick Selection object strokes through raster surfaces measured
-3.46-3.70x faster at 8-bit, 16-bit and 32f on 24-36 MP documents (AWS
-c7i.4xlarge, five paired release runs), with identical region bounds and mask bytes.
-Hybrid FIFO push/relabel and sparse edge enumeration preserve the fitted energy.
-See [stroke measurements and reproduction](quick-selection-performance.md).
-
-**Bottom line.** Two days after 0.2.0 we had merged ~96 PRs and closed ~48 issues, but **real
-Photoshop parity is still well below 50%**. The biggest gaps are AI, missing tools, professional
-workflow depth and the plug-in ecosystem. Most fixes since 0.2.0 have passed our tests but have
-**not yet been validated by users**, and this week proved our tests miss what users hit:
-on first public use they found broken basics (text selection offset, 214 shortcut failures, panels
-resizing themselves, an immovable crop frame, folders that wouldn't collapse, lag on layout PSDs)
-that all counted as "live" in `parity.md`.
-
-**Overall.** Feature surface: roughly **60–70%** of what a typical Photoshop user touches exists in
-some form. "A professional could switch for daily work": roughly **25–35%**. True 1:1 parity is
-**many months** of focused work, and some areas need product decisions rather than effort.
-Confidence: moderate — the next users of 0.2.x will move these numbers either way.
-
-### By dimension
-
-2026-10-10: Native CMYK solid and gradient fills share depth-quantized samples between
-rendering and PSD export. The uncached 16-bit CMYK gradient now round-trips with zero
-rendered error; psd-tools round trips rise from 307 to 308. Mixed and Photoshop round
-trips remain 169 and 258; rendering oracle counts remain 146, 237 and 133.
-Exact 8-bit CMYK row reuse reduces 1 MP conversion-read time by 77% for smooth ramps,
-33% with dithering and 10% for random colours (M4 Max, release optimization level 3).
-
-2026-10-10: [Shape stroke controls](shape-strokes.md) expose existing Solid/Dashed/Dotted,
-custom dash/gap, offset, caps, joins, alignment, miter-limit and opacity capabilities in
-Shape/Pen options and Properties. Properties stages a cached canvas preview before one
-undoable edit. Gradient fill controls from #1053 remain open; the Line tool still draws a
-filled bar. This extends UI reachability and does not claim Photoshop-authored stroke parity.
-
-| Dimension | Measured / evidence (2026-10-05) | Grade | Notes |
-|---|---|---|---|
-| Menu wiring | 626/626 menu items dispatch a command (`parity.md`) | high but shallow | Says nothing about behaviour. |
-| PSD fidelity (rendering) | Corpus oracle 115/170 (68%): 30 differ, 26 have no usable reference, 1 import error. 2026-10-07: io corpus 146/170, psd-tools corpus 236/309 (was 229: Advanced Blending knockouts), Photoshop oracles 132/258 | medium | Push to 170/170 under way (effects/strokes, multi-instance effects, 16/32-bit and colour modes, references for skipped files). |
-| PSD round trip | 169/169 re-import identically; every adjustment layer and blend mode round-trips | high (within corpus) | Floors in `crates/io/tests/corpus.rs`; raise, never lower. Corpora: `cargo xtask corpus --all` (ours: https://github.com/storytold/photocraft-corpus). |
-| OpenRaster read/write | 2026-10-09: 34 Krita 5.2.9-authored `.ora` files (27 blend modes, groups, pass-through, offsets, opacity, visibility, masks, 16-bit, gray) render within 3/255 of Krita's merged image (30) or differ for known reasons (4: Krita's lighter/darker-colour tie-break, Hard Mix at exactly 1, a linear-light 16-bit document); Krita re-opens our exports and renders them identically to its originals (33/34; Soft Light is written as `svg:soft-light`) | medium (one authoring app) | MyPaint and GIMP files untested. Masks are applied to pixels and layer styles dropped on save (reported). See [OpenRaster](ora.md). |
-| Paint.NET import | 2026-10-09: 14/14 local PDN3 textures preserve editable layers exactly through `.pcraft` and render within 2.142/255 of full-size previews; a supplied 5.x document (1280×720, 11 layers) passes editing/history/native saves and matches its thumbnail within 0.418/255 mean; all 14 blend modes exercised synthetically | medium (limited corpus) | Import only; metadata omitted. More current-version files and full-size exports needed. See [PDN support](pdn.md). |
-| Smart filters / text / effect shapes in PSDs | Measured on our Photoshop-authored set (https://github.com/storytold/photocraft-corpus, `corpus/photoshop`, 258 files): see the per-group floors in `crates/io/tests/corpus.rs` and `crates/engine/tests/photoshop_oracles.rs`. Smart objects and smart filters now survive PSD save and open (41 corpus files, 206 smart objects, round trip strict; Photoshop opens our exports with live filters); re-rendering Photoshop's smart filters with ours matches 5/30 (was 1/30) | low–medium | Remaining re-render gaps are filter maths (Gaussian/Motion Blur, Unsharp Mask, Emboss, Add Noise RNG) and bicubic placement. |
-| Core editing (layers, masks, selections, adjustments, filters, transforms) | Broad engine coverage; many interaction bugs fixed after 0.2.0 (adjustment dialogs, Curves, crop, Move/Transform modifiers, gesture origin) | medium | Fixes not yet user-validated. |
-| UI / UX polish | Shortcut audit 214 → 0 failures; dock, Layers rows and menus reworked; first visual-QA sweep found 14 defects (#147–#157). 2026-10-07: the keyboard-only shortcuts with no menu item (⌥[ ⌥] ⌥, ⌥. layer navigation, ⇧⌥[ ⇧⌥] to extend the selection, 1–0 for opacity and ⇧ for flow or fill, ⇧[ ⇧] hardness, ⌥⌘T to transform a copy and ⌥⇧⌘T to step and repeat), ⌥-click colour sampling with painting tools, double-click a Layers row for Layer Style, File › New from Clipboard, a centred main window and remembered Liquify settings (#352, #417, #350, #368, #419, #418). 2026-10-09: canvas zoom is physical at any display scale — 100% is one document pixel per physical display pixel, and every screen-space overlay, cursor, scrollbar, navigator and fit follows it (#1943) | low–medium | Needs recurring visual QA with realistic documents. |
-| Tools | 2026-10-09: Pencil, Mixer Brush, Patch, Content-Aware Move, Vertical Type, Pattern Stamp and Rotate View are toolbar tools. Remaining missing tools include Red Eye, Art History Brush, Freeform/Curvature Pen, Add/Delete Anchor Point tools, single row/column marquee, Color Sampler, Perspective Crop, type masks and Frame | low–medium | Patch has a live healing preview. Content-Aware Move is partial: Move/Extend, Structure/Color and Sample All Layers exist; Transform On Drop and a live result preview remain missing (`TOOL-213-4` in the scorecard). Magic/Background Eraser added; live gradients in progress (#180). Magnetic Lasso added 2026-10-08 (live-wire edge tracing, Width/Contrast/Frequency, `select.magneticLasso`). Pattern Stamp added 2026-10-08 (S flyout, `paint.patternStamp`, Aligned / Impressionist). Rotate View (2026-10-09) turns the canvas camera around its centre without rewriting pixels; Reset View and Match Rotation copy the angle. Crop rotation (2026-10-09, #1792): dragging outside the crop frame turns it (⇧ 15° steps, angle readout), and ↵ rotates the document and crops in one undo step (`image.crop` `angle`); the frame turns over the image as in Photoshop's Classic Mode, and the options-bar Straighten is still missing. |
-| Painting | Brush model and Brush Settings panel near Photoshop; .abr/.grd import; persistent presets; pen pressure/tilt on Windows, web, macOS and X11 | medium | Native Wayland pen input open (#79; a pen opens the window through Xwayland meanwhile); X11 pressure confirmed by a user, macOS not yet verified on tablet hardware. |
-| Text / typography | Engine works; caret placement and size editing fixed; OpenType features, text-on-path editing, composer parity partial | medium-low | Measure with the Photoshop-authored set. |
-| Colour management | Colour-managed canvas (document → monitor), embedded CMYK profiles, linear EXR/HDR, 16-bit float canvas | medium-high | Monitor profile follows only at launch. |
-| Performance | 14k+ px on the GPU at ~⅓ the memory; adjustment preview 285 ms → 4–9 ms; font-size edits 297 ms → 4.6 ms; 2026-10-07: 30 MP TIFF open (banded, parallel strip/tile decode) Deflate 345 → 32 ms, LZW 428 → 43 ms, BigTIFF and every IFD readable; 2026-10-09: 4.2 MP Indexed Color, 256 colours, full-resolution CPU preview p50 3018 → 1061 ms (Ryzen AI 7 350, three measured runs; GPU upload excluded) | medium-high on rasters | Complex layout documents still laggy (#125/#128); >16384 px GPU tiling in progress (#49). Native Indexed Color previews run on one background worker with stale-result rejection; large palettes can still take about a second to compute. Web previews remain synchronous. |
-| Stability | Never-crash lint series, crash guard, `panic_hunt` fuzzing in the gate | medium-high | No field crash data yet. |
-| Camera RAW | DNG, CR2, Sony ARW (lossless + compressed), Nikon NEF (lossless + lossy compressed), RW2, uncompressed ORF; 2026-10-09: uncompressed Fujifilm RAF, Bayer and X-Trans (edge-directed X-Trans demosaic; 26 MP X-Trans develop 306 ms on 12 cores) | medium | CR3, compressed RAF, Nikon "lossy after split" and calibrated colour for non-DNG cameras (no Fuji colour calibration: neutral fallback) still open (#50). |
-| AI / generative | none | ~0% | Deferred by decision (#41). |
-| Ecosystem | Sandboxed WebAssembly plug-ins instead of .8BF; no ExtendScript/UXP/.atn; no Adobe Fonts/Libraries/cloud docs | low | By design for 8BF; scripting compatibility open. |
-| Platforms | macOS (notarized), Windows, Linux (AppImage/deb/rpm/Flatpak bundle), web | medium-high | Flathub later (#173); Windows signing material pending. |
-| Localisation | 2026-10-07: 10 UI languages; menu, `tl!`, blend mode, preference and brush-section coverage enforced by tests; live switching and scoped Preferences previews | medium | Engine errors/status messages still partly English; CJK web fonts, browser-locale detection, and RTL remain open. |
-
-2026-10-08: Motion Blur adds adaptive FFT convolution for wide streaks while retaining the
-row kernel merged in [#902](https://github.com/storytold/photocraft/pull/902) for shorter
-streaks and unsupported cases. Native release medians of three paired runs on an Intel
-i7-9750H with 12 Rayon workers, synthetic RGBA8 at 30° / distance 1000: 24 MP
-22.594 → 9.785 s (2.31×), and 1.5 MP
-2.455 → 0.592 s (4.15×).
-Full-image comparisons differ by at most one U8 code level; distance 64/256 retain the
-byte-identical row output. These measure decode/halo/filter/write/prune, excluding UI proxy
-creation, compositing and upload; the UI remains synchronous. FFT work is scheduled within
-a conservative 512 MiB working-set estimate, excluding stored Surface tiles and row fallback
-memory. The crossover is a heuristic; images dominated by near-cutoff alpha can need extra
-scalar work. These measurements do not establish Photoshop filter parity.
-
-2026-10-10: Proximity Match search medians on an AWS c7i.4xlarge improved 1.54–3.82x
-for four synthetic stroke regions (50–300 px holes), with identical source displacements.
-Bounding the mask table to the hole and pruning losing SSD candidates reduced table storage
-by about 90–93 percent there. Full 24–36 MP kernel inputs improved 4.60–6.24x, but normal
-engine strokes already use a cropped region. See [method and limits](proximity-match-performance.md).
-
-2026-10-09: the Remove tool is in the J flyout (`paint.remove`, a background job). A stroke
-around an object also removes what it encloses. The fill is non-local patch completion with
-texture features (A. Newson et al., IPOL 2017: `crates/algo/src/nonlocal.rs`), finished by a
-best-patch copy and gradient-domain seam hiding. On 20 holes in public-domain photos
-(`remove_quality`, `docs/development.md` › Remove Tool quality) the fill keeps 0.89 of the
-original's texture (Wexler/PatchMatch completion as Content-Aware Fill uses it: 0.54). On a 24 MP
-document a 100 px ring takes 120 ms and a 1000 px ring 2.5 s (`remove_bench`). It is not
-generative: large objects over complex structure fill less convincingly than Photoshop's AI mode
-(#41).
-
-2026-10-08: the tools update above is checked against the current toolbar groups in
-[`panels.rs`](../crates/ui-egui/src/panels.rs), the Pencil and tool-cycle tests in
-[`pencil_tests.rs`](../crates/ui-egui/src/pencil_tests.rs), Patch's live-preview test in
-[`patch_preview.rs`](../crates/ui-egui/src/patch_preview.rs), and Vertical Type's point/paragraph
-creation and undo test in [`type_tool_tests.rs`](../crates/ui-egui/src/type_tool_tests.rs).
-These implementations remove the tools from the missing list; they do not establish Photoshop
-behavioural parity. The historical estimates and corpus measurements in this assessment are
-unchanged; current measured floors remain in the scorecard.
-
-2026-10-07: Camera Raw PSD mapping covers relative custom white balance, Light/Presence,
-parametric and four point curves, HSL, Color Grading, sharpening/noise detail, grain and numeric
-post-crop vignette controls. Two revisions of one supplied Photoshop ACR 18.4 PSD preserve their
-filter descriptors without edits. The updated revision stays opaque because of active manual
-Optics and unverified vignette style; a descriptor projection verifies its supported controls.
-Synthetic 8/16/32-bit PSD → edit → `.pcraft` → PSD tests preserve the filter stack, mask and
-undo/redo. Unmapped Camera Raw fields/versions remain opaque; Photoshop
-acceptance of generated exports and pixel parity are still unverified. Corpus floors above are
-unchanged.
-
-2026-10-10: Select and Mask's Shift Edge now computes the exact octagonal footprint
-with a square window extreme plus a dyadically built Manhattan ball. On a c7i.4xlarge
-(16 vCPUs), full tiled CPU refinement at radius 64 / Shift Edge +100% takes
-3.019 s -> 0.762 s at 24 MP (3.96x),
-3.706 s -> 0.921 s at 36 MP (4.02x).
-Three alternating paired release runs include guide sampling, guided filtering,
-boundary distances, quantization and assembly. Every output mask matches exactly.
-The dense fractional 24 MP radius-64 dilation kernel improves 36.00x;
-circular soft-mask kernels improve 20.76-172.16x at 24-36 MP, radii 4-64.
-Bounded halo tiles reduce the large-mask workspace allocation bound from 192/288 MB
-to about 115/163 MB, including output and 16 workers' scratch, excluding the source.
-See [method and results](octagonal-mask-performance.md).
-
-2026-10-10: Pixelate's Facet shares exact 3x3 quadrant statistics and Color Halftone
-shares dot radii within each rotated screen grid. Full tiled CPU application at 24 MP
-on a 16-vCPU m7i.4xlarge improves Facet from 0.684 s to 0.312 s in RGB8
-(2.20x), and from 7.888 s to 0.574 s in CMYK8 (13.74x).
-Color Halftone improves from 1.137 s to 0.604 s in RGB8 (1.88x)
-and from 3.589 s to 0.718 s in Lab8 (5.00x).
-Three alternating paired release runs also cover 36 MP RGB, 16/32-bit storage,
-and screen radii 2/8/64. All 36 complete output comparisons match exactly.
-See [method, raw timings and checks](pixelate-performance.md).
-
-### Where we're going (priority order)
-
-1. **Ship the fixes:** cut 0.2.1 once the current batch lands, so users validate them.
-2. **PSD fidelity to 170/170** with round trips, plus the Photoshop-authored reference set for smart
-   filters, the text engine and effect shapes (https://github.com/storytold/photocraft-corpus, fetched by
-   `cargo xtask corpus --all`); enforce floors in `corpus.rs`.
-3. **Workflow acceptance tests:** 30–50 real tasks (retouch a portrait, social post with text and
-   effects, composite with masks and adjustment layers, CMYK print prep…) scripted end to end and
-   checked against Photoshop's output on every build. Their pass rate becomes the headline parity
-   number.
-4. **Tool coverage and workflow depth:** Pen variants (Direct Selection landed, #790) and
-   Perspective Crop; finish Content-Aware Move's Transform On Drop and live result preview.
-   Pencil, Mixer Brush, Patch, Vertical Type, Pattern Stamp and Rotate View are already available on the toolbar.
-5. **Complex-document performance** (#125/#128) and GPU tiling beyond the texture limit (#49).
-6. **Recurring visual QA** (`cargo run -p photocraft-engine --example designer_psd`) and fast
-   turnaround on user reports (OS, document size, layer count, screenshot).
-7. Later / needs decisions: generative AI backend (#41), scripting compatibility (ExtendScript /
-   UXP / .atn), Flathub (#173), Wayland pen pressure (#79).
-
-## Current focus (infrastructure before the long tail)
-
-Landed on 2026-10-01:
- multi-layer selection, live smart objects + smart filters, alpha channels + Quick Mask;
- patterns, Warp, preferences, snapping, ~33 filters, the remaining core adjustments;
- Layer Comps and Artboards, GPU layer effects, Liquify / Puppet Warp / Perspective Warp;
- the PSD fidelity pass (oracle 102 → 111), the M11 MCP acceptance test, and the release pipeline
-  (`docs/releasing.md`).
-
-Next:
-1. **Follow "Where we're going" in the assessment above.** 0.2.0 shipped signed and notarized on
-   2026-10-05 (`docs/releasing.md`); Windows code-signing material still needs to be obtained.
-2. **Fidelity**: the PSD oracle (113/170). Modern Brightness/Contrast and grayscale Levels
-   curves; chisel-soft / stroke-emboss bevel shapes; Photoshop's 8-bit blend rounding; non-Normal
-   modes in Lab documents; smart-filter maths against Photoshop's (re-render oracle 5/30).
-3. **GPU**: only Multichannel documents and regions over the texture limit fall back to the CPU.
-4. **Vanishing Point, Camera Raw / Lens Correction, Face-Aware Liquify** (needs a landmark model).
-5. **Panels**: Patterns, Styles, Glyphs, Character/Paragraph Styles, Timeline; Custom Shape tool.
-6. Print, Photomerge, Merge to HDR, video layers.
+| Next milestone | Scope | Estimate |
+|---|---|---:|
+| 0.7 "files you can trust" | Current focus 1 plus JPEG Options, AVIF read, Photoshop PDF read | 200–320 h |
+| 0.8 "fast enough" | Current focus 2; GPU tiling past the texture limit (#49); complex layout documents | 200–350 h |
+| 0.9 "feels like Photoshop" | Current focus 4 and 5; transform context menu; floating panels; Color panel | 300–500 h |
+| Beta | All of the above, type depth, Select and Mask / Content-Aware Fill workspaces, Hindi/Arabic/Vietnamese + RTL, bug backlog under 100 | 1,100–1,800 h total |
+| Later, needs owner decisions | AI backend (#41), plug-in hosting (.8BF / UXP), scripting compatibility | 550–1,050 h |

 ## Milestone definitions

@@ -278,3 +101,12 @@ Each milestone has a **definition of done (DoD)** and must leave `main` green on
 | **M10** | Smart features | `ml` (ort native / ort-web on the web), Select Subject/Object/Sky, Remove BG, Remove tool, content-aware fill, healing, AI denoise, RAW develop | Quality benchmarks on a public dataset; timing budgets |
 | **M11** | Automation + formats | MCP server, batch, scripting, remaining formats (JP2, DICOM, DPX…), C2PA | An agent completes 10 scripted edit tasks via MCP |
 | **M12** | Pro parity | CMYK/Lab UI, print, HDR display, photomerge/HDR merge, timeline, layer comps, artboards, symmetry, neural filters | `xtask parity` ≥ 90% of Photoshop menu checklist |
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | minor | Alpha gate table (core-workflow gate from the progress-docs standard): passes, stage stays alpha |
+| 2026-10-10 | major | Progress-docs standard: honest assessment and dated measurements moved to `target-app-parity.md`, Affinity notes to `file-format-parity.md`; new Current focus and estimates; `parity.md` renamed `parity-checklist.md` |
+| 2026-10-08 | minor | Tools, Affinity import and performance measurements added |
+| 2026-10-05 | major | Honest parity assessment added |
@@ -221,11 +221,11 @@ Unread: `general.color_picker`, `general.beep_when_done`, `general.resize_image_

 ## Automation

-Menu wiring (`docs/parity.md`): 628 of 628 menu items live (100.0%). This counts dispatch, not behaviour.
+Menu wiring (`docs/parity-checklist.md`): 628 of 628 menu items live (100.0%). This counts dispatch, not behaviour.

 | Id | Target | Status | Issue | Note |
 |---|---|---|---|---|
-| AUTO-1 | Every menu item dispatches a command id (UI, CLI, control channel and MCP share them) | done |  | docs/parity.md: all menu items live; menus.rs is_live |
+| AUTO-1 | Every menu item dispatches a command id (UI, CLI, control channel and MCP share them) | done |  | docs/parity-checklist.md: all menu items live; menus.rs is_live |
 | AUTO-219-1 | Record actions in a document-independent form (implicit layer, merged type keystrokes) | missing | [#219](https://github.com/storytold/photocraft/issues/219) | actions.rs:47 copies raw journal entries with explicit layer ids |
 | AUTO-219-2 | Actions panel: step and dialog toggles, reorder, delete a step, Stop, sets, .atn import | missing | [#219](https://github.com/storytold/photocraft/issues/219) | Panel deletes whole actions only (actions.rs:146) |
 | AUTO-219-3 | actions.record/stop/play/list commands for agents and the CLI | done | [#219](https://github.com/storytold/photocraft/issues/219) | actions.list/get/record/stop/play/delete in actions_cmds.rs; the panel, CLI and MCP share them. Nested steps pass the session authorize hook. |
@@ -0,0 +1,535 @@
+# PhotoCraft vs Adobe Photoshop: target-app parity
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (full re-measure against Photoshop 2026 27.11.0; merged `parity-estimate.md` and the roadmap's honest assessment into this file) · **Target:** Adobe Photoshop 2026
+
+The authoritative answer to "how close is PhotoCraft to Photoshop, and how much work is left". The
+one-page summary is [`ROADMAP.md`](../ROADMAP.md); the ranked work list is [`gaps.md`](gaps.md);
+the measured checklists are [`scorecard.md`](scorecard.md) (`cargo xtask scorecard`) and
+[`parity-checklist.md`](parity-checklist.md) (`cargo xtask parity`). Detail by dimension:
+[UI](ui-parity.md), [brushes and painting](brush-parity.md), [context menus](context-menu-parity.md),
+[file formats](file-format-parity.md), [hardware](hardware-parity.md),
+[localization](localization-parity.md).
+
+## Headline (2026-10-10, PhotoCraft 0.6.0, `main` at 3e000bc1)
+
+| Number | Value | Kind |
+|---|---|---|
+| **Feature breadth** (does each Photoshop feature exist?) | **~76%** | partly measured (menus, tools, panels, formats), weights below |
+| **Ready for real work** (could a professional replace Photoshop with it?) | **~45%** (range 40–50%) | estimated, weights and evidence below |
+| **Mainstream practitioner** (typical pro, weekly areas only) | **~43%** (38–48%) | estimated, [method](#mainstream-practitioner-43) |
+| **Essentials user** (core features only) | **~61%** (55–67%) | estimated, [method](#essentials-user-61) |
+| **Stage** | **alpha** | ready-for-real-work is inside the ~40–75% band and the [alpha gate](roadmap.md#alpha-gate) passes (six core workflows end to end); PSD exchange with Photoshop is not yet reliable enough for beta |
+| Remaining to **beta** (~75% ready, PSD reliable) | **~1,100–1,800 Opus 5.5 agent-hours** | estimated, calibrated below |
+| Remaining to **full parity** (~95–100%) | **~2,300–4,100 Opus 5.5 agent-hours** | estimated; includes AI and ecosystem, which need owner decisions |
+
+These numbers replace the 2026-10-03 estimate (82% "of the way", 140–220 agent-hours to 95%,
+kept below for the record). That estimate counted feature surface and was far too optimistic:
+about 1,230 issues have been filed since the first release (579 open on 2026-10-10), most of them
+against features already counted as present. The 2026-10-05 honest assessment (feature surface 60–70%,
+ready for real work 25–35%) was closer; ~700 merged PRs since then moved both numbers up.
+
+### Readiness by audience
+
+| Audience | Ready % | Opus 5.5 agent-hours to ~95% | Work that dominates |
+|---|---:|---:|---|
+| Full target (ready for real work) | ~45% (additive weighted sum: 46%) | 2,300–4,100 | Everything below, plus AI tier, plug-in/scripting ecosystem, video, localization, hardware |
+| Mainstream practitioner | ~43% | 1,700–2,800 | Depth in the 11 weekly areas (1,180–2,000), performance budgets, stability on real machines, PSD exchange with Photoshop |
+| Essentials user | ~61% | 400–700 | Startup and GPU robustness, brush and transform responsiveness, Type tool, discoverability (flyouts, commit buttons), AVIF/PDF open |
+
+Hours calibrated as in [Remaining effort](#remaining-effort-and-calibration) (2–9 h per checklist
+item, 8–20 h per format or subsystem, from the repo's 1,004 merged PRs); each tier is a subset of
+the one above. About 80% parallelizes (independent crates and command modules); performance
+architecture and the PSD writer are the serial parts.
+
+## Target measured
+
+- **Adobe Photoshop 2026, version 27.11.0** (build 20260929.r.26, macOS universal), installed at
+  `/Applications/Adobe Photoshop 2026`, with Camera Raw 18.7. Inspected without launching it:
+  `Info.plist` document types, the file-format plug-ins' PiPL read/write flags, `.lproj` and
+  `Locales/` language packs, `Default Keyboard Shortcuts.kys` (67 toolbox tools, 107 menu commands
+  with default shortcuts), `Required/` plug-ins (85 filter entries, 12 format plug-ins), presets,
+  and strings in the main binary (81 "… Tool" titles; HDR/EDR, Metal, Touch Bar, pen features).
+- PhotoCraft's own Photoshop menu tree (`crates/ui-egui/src/menu_catalog.rs`, 628 items) was
+  written from Photoshop's menus; it is the denominator of the menu number. Photoshop's binary
+  `Default Menus.mnu` was not parsed, so items added in 27.x that the catalog lacks would not show
+  (Generative and cloud items are the likely ones).
+- User reports on GitHub (`storytold/photocraft`, 579 open issues on 2026-10-10) as evidence of
+  depth and feel. Adobe's documentation and release notes for behaviour not visible in the bundle.
+
+## Feature breadth: ~76% (weighted)
+
+| Component | Weight | Ours / Photoshop | % | Kind |
+|---|---:|---|---:|---|
+| Menu commands | 30% | 628 / 628 catalog items dispatch a live command | 100% | measured (`cargo xtask parity`) |
+| Toolbox tools | 15% | 53 / 68 (Photoshop's 67 `.kys` tools plus Remove, Triangle, Curvature Pen; minus legacy Rounded Rectangle and Targeted Adjustment) | 78% | measured (`Tool` enum in `ui-egui/src/state.rs`, `TOOL_SECTIONS` in `panels.rs`) |
+| Panels | 10% | 30 / 35 Window-menu panels | 86% | measured (`view_cmds.rs`, panel modules) |
+| Filters and adjustments | 10% | every Filter and Image › Adjustments item live; Filter Gallery 47 looks | 100% | measured (menu catalog) |
+| File formats (usage-weighted) | 15% | 13 of 31 Photoshop formats supported in Photoshop's directions, 4 partial; the common ones (PSD/PSB, TIFF, JPEG, PNG, WebP) all present | ~80% (48% by count) | measured list, estimated weights ([file-format-parity.md](file-format-parity.md)) |
+| AI / generative | 10% | classical Select Subject/Sky, Remove tool, Content-Aware; no Generative Fill/Expand, Neural Filters, Sky Replacement, Super Zoom | ~15% | estimated |
+| Ecosystem (plug-ins, scripting, libraries) | 10% | WebAssembly plug-ins, CLI, MCP, actions; no .8BF, UXP/ExtendScript, .atn import, Libraries | ~20% | estimated |
+| **Weighted** | 100% | | **~76%** | |
+
+## Ready for real work: ~45% (weighted)
+
+Weights reflect what a working Photoshop user depends on (photographers, retouchers, designers).
+
+| Dimension | Weight | % ready | Remaining (agent-h) | Evidence | Detail |
+|---|---:|---:|---:|---|---|
+| Features in depth (tools, layers, selections, type, filters, transforms) | 25% | 55% | 600–1,000 | Scorecard checklists: Tools 20 done / 15 partial / 6 missing, Type 0 / 3 / 6, Automation 3 / 2 / 6. 15 tools missing; Select and Mask, Vanishing Point and Content-Aware Fill are parameter dialogs, not workspaces; type on a path, manual kerning, RTL paragraphs missing | [gaps.md](gaps.md) |
+| UI / UX fidelity and feel | 15% | 40% | 350–600 | ~150 open UI issues; Workspace checklist 5 / 7 / 11; 46 of 149 preferences do nothing; no floating panel groups, thin Color panel, no transform context menu, brush lag and pressure jumps reported | [ui-parity.md](ui-parity.md), [brush-parity.md](brush-parity.md) |
+| File formats and fidelity | 15% | 60% | 300–500 | PSD oracle on Photoshop-authored files 133/256 (52%); io corpus 146/170 (86%); psd-tools 236/309 (76%); round trips ~100% of corpus; but users report Photoshop refusing our files (#1281, #2469; a text-layer case, #2374, was fixed 2026-10-10); no Photoshop PDF, EPS, JPEG 2000, JPEG XL, AVIF read, CR3 | [file-format-parity.md](file-format-parity.md) |
+| Performance | 10% | 25% | 200–350 | 3 of 25 budgeted scenarios meet budget (baseline 2026-10-05); 150-layer nudge crashes the GPU path (P12); layer move on 15000×10000 16-bit 2.7 s p50; users report brush and cursor lag (#2619, #2566, #1994) | [scorecard.md](scorecard.md#performance) |
+| Stability | 10% | 45% | 150–250 | Never-crash lints on 28/28 crates, `panic_hunt` in the gate, atomic saves; but field reports of freezes (#2518 Type tool 60 s on macOS), GPU panics (#2020, #1215), no-GPU start failures (#2519, #2412) | [gaps.md](gaps.md) |
+| Hardware | 5% | 50% | 80–150 | GPU compositor (wgpu), pen pressure on all desktops but tilt/rotation not on Windows, Wayland pen broken (#2622), no HDR/EDR output, monitor profile auto-detected on macOS only | [hardware-parity.md](hardware-parity.md) |
+| Localization | 5% | 55% | 80–140 | 7 of the 12 key languages at 100% of measured UI keys; Hindi, Arabic, Vietnamese absent; no RTL UI; engine messages English | [localization-parity.md](localization-parity.md) |
+| Platforms | 5% | 80% | 30–60 | macOS (notarized), Windows (unsigned), Linux (AppImage/deb/rpm/Flatpak), web, FreeBSD partial; no Windows ARM64, no in-app updates | [scorecard.md](scorecard.md#distribution) |
+| Ecosystem (plug-ins, scripting, presets, libraries) | 5% | 15% | 250–450 | No .8BF hosting (#2647), no UXP/ExtendScript, no .atn import; .abr/.grd/.pat/.aco/.ase/.cube supported | [gaps.md](gaps.md) |
+| AI features | 5% | 10% | 300–600 | No ML runtime at all (scorecard AUTO-219-7); classical substitutes only. Needs an owner decision and openly licensed models (#41) | [gaps.md](gaps.md) |
+| **Weighted** | 100% | **~46%** | **2,340–4,100** | | |
+
+## Mainstream practitioner: ~43%
+
+The same question for the typical professional Photoshop user (photographer, retoucher,
+designer), not the whole product. Method: craftrules `standards/progress-docs.md`. Areas a
+professional uses weekly (the [alpha-gate](roadmap.md#alpha-gate) workflows plus their everyday
+tools), depth from the [feature areas](#feature-areas) table. Left out: AI, plug-ins and
+scripting, Libraries/cloud, video, specialist hardware (tilt, Touch Bar), languages.
+
+| Area | Weight | Depth |
+|---|---:|---:|
+| Layers, masks, blend modes, smart objects, styles | 20% | 65% |
+| Adjustments and filters | 15% | 70% |
+| Selections and masking | 15% | 55% |
+| Retouching | 12% | 50% |
+| Transform and warp | 8% | 55% |
+| Type | 8% | 40% |
+| Painting and brushes | 7% | 55% |
+| Vector shapes and paths | 5% | 55% |
+| Camera Raw | 4% | 45% |
+| Colour management | 3% | 70% |
+| Export, Save for Web | 3% | 50% |
+| **Weighted depth** | 100% | **57%** |
+
+Discounts for what still stops real work (estimated, each with evidence):
+
+| Discount | Factor | Evidence |
+|---|---:|---|
+| Interaction fidelity | ×0.92 | ~150 open UI issues; transform context menu, spring-loaded tools, panel tear-off missing ([ui-parity.md](ui-parity.md)); brush lag and pressure reports ([brush-parity.md](brush-parity.md)) |
+| Stability and speed on real machines | ×0.88 | 3 of 25 performance budgets met; 150-layer GPU crash (P12); Type tool freeze (#2518); GPU start failures (#2519, #2412, #2020); 41 open performance issues |
+| Exchanging files with Photoshop users | ×0.92 | Photoshop-authored oracle 133/256; exports refused by Photoshop (#1281, #2469); smart-filter re-render 5/30 |
+
+**57% × 0.92 × 0.88 × 0.92 ≈ 43%** (range 38–48), estimated.
+
+It is not higher than ready-for-real-work (~45%) because the areas it leaves out carry little
+weight in the full number (AI, ecosystem, localization and hardware together 20%, averaging
+~33%), while the discounts apply to every workflow. The two numbers agree within their error;
+the full number stays where its written weights put it.
+
+**User evidence** (GitHub, 2026-10-10; search hits include maintainer replies, so they are upper
+bounds): 0 issues mention switching from or replacing Photoshop; praise words appear in ~19
+("love"), 10 ("amazing"), 6 ("awesome"), 4 ("great work") issues. Of 579 open issues, 136 are
+feature requests and 443 bugs or reports, of which ~244 are on core paths (layers, selection,
+brush, transform, type, save/open, performance) by title. Users are trying it for real work,
+and most of what they report is core-path friction, not niche requests.
+
+## Essentials user: ~61%
+
+An occasional user who touches only the essentials: open or create a file, the most used tools
+and commands, undo, save and export, default settings. Excluded: advanced options, pro
+workflows, PSD exchange edge cases, and everything excluded above.
+
+| Core feature | Weight | Depth | Notes |
+|---|---:|---:|---|
+| Save, Save As, Export JPEG/PNG | 12% | 85% | |
+| Layers (add, reorder, hide, opacity) | 12% | 80% | |
+| New / open a photo | 10% | 90% | |
+| Select (marquee, lasso, magic wand, object) | 10% | 75% | |
+| Move and Free Transform | 10% | 70% | no commit/cancel buttons (#2442) |
+| Brush and eraser | 10% | 70% | lag with big brushes (#2619) |
+| Undo, History | 8% | 90% | |
+| Basic adjustments (brightness, levels, curves, hue/saturation) | 8% | 80% | |
+| Crop and straighten | 6% | 85% | |
+| Text | 6% | 55% | freeze report (#2518) |
+| Spot healing, Remove | 5% | 65% | |
+| Blur and sharpen filters | 3% | 85% | |
+| **Weighted depth** | 100% | **78%** | |
+
+Discounts: launch and stability ×0.90 (start failures without a usable GPU, #2519, #2412,
+#1859; Type tool freeze, #2518); discoverability and UI clarity ×0.92 (language option not
+found, #2532; flyouts only on right-click, #2301; toolbar hover issue, #2603); opening files
+people send ×0.95 (JPEG, PNG, HEIC, WebP, PSD open; AVIF and PDF don't, #463, #2541).
+**78% × 0.90 × 0.92 × 0.95 ≈ 61%** (range 55–67), estimated.
+
+## Feature areas
+
+Percent is ready-for-real-work for that area (depth and fidelity, not presence). Hours are to full
+parity for the area and are already contained in the dimension rows above.
+
+| Area | % | Hours | Notes |
+|---|---:|---:|---|
+| Layers, groups, masks, blend modes, smart objects, layer styles | 65% | 150–250 | Broad and GPU-accelerated. Open: multi-layer Smart Object conversion (#2604, #2648), layer clipboard between documents (UI-217-23), Stamp Visible, contour editor, smart-object perspective (FILE-215-6), mask targeting by command (#2458) |
+| Selections, Select and Mask, channels | 55% | 120–200 | All selection tools except single row/column marquee. Select and Mask is a generic dialog (TOOL-214-1), no Refine Hair (#2671); Select Subject/Sky classical, 559 ms p50 on 24 MP vs a 100 ms budget |
+| Painting and brush engine | 55% | 150–250 | Every Brush Settings section exists; feel is the gap: lag on big brushes, pressure jumps, smoothing and edge quality reports ([brush-parity.md](brush-parity.md)) |
+| Retouching (healing, clone, patch, remove, content-aware) | 50% | 100–180 | Remove tool non-generative; Content-Aware Fill has no workspace (TOOL-214-15); live stroke preview only for Brush/Eraser (TOOL-213-7) |
+| Transform, warp, puppet, perspective, liquify | 55% | 100–180 | Free Transform modifiers right; no transform context menu (TOOL-214-5), no skew fields, puppet mesh is a block grid (TOOL-214-6), Liquify called "useless" by a user (#2606) |
+| Adjustments and filters | 70% | 120–200 | All present; smart-filter re-render matches Photoshop on 5 of 30 files; Filter Gallery looks approximate; no Neural Filters |
+| Type | 40% | 120–200 | Engine works; no type on a path, manual kerning, RTL paragraphs, OpenType panel UI, live Warp Text preview; Type tool freeze on macOS (#2518) |
+| Vector shapes and paths | 55% | 80–140 | Pen, Direct/Path Selection, shapes, shape strokes. Missing Freeform/Curvature Pen, anchor tools, on-canvas corner radius, path-point nudge; shape parity tracking issue #2496 |
+| Colour management and modes | 70% | 60–100 | Own ICC engine, soft proof, CMYK/Lab/Indexed/Duotone/Multichannel; no Adobe CMYK profiles, no spot plates (FILE-215-17), monitor profile only at launch |
+| Camera Raw and computational photo (Photomerge, HDR, lens) | 45% | 120–200 | Raw for DNG/CR2/NEF/ARW/RW2/ORF/RAF(uncompressed)/PEF; no CR3; ACR dialog partial (no masking, calibration); generic lens profiles only |
+| Automation, actions, batch, scripting | 45% | 120–200 | CLI/MCP/control channel ahead of Photoshop; Actions panel and recording not document-independent (AUTO-219-1/2), no .atn, no ExtendScript/UXP |
+| File I/O | 60% | 300–500 | See [file-format-parity.md](file-format-parity.md) |
+| Print, export, Save for Web | 50% | 60–100 | JPEG Options dialog missing (FILE-215-19), no Photoshop PDF, no separations, Save for Web settings not remembered |
+| Video / timeline, data-driven graphics | 30% | 100–200 | Menu items live; frame model persists in `.pcraft`; no PSD video-layer mapping |
+| Workspace, panels, preferences | 40% | 150–250 | 46 of 149 prefs do nothing; no floating/OS-window panel groups; Libraries, Comments, Contextual Task Bar missing |
+| AI / generative | 10% | 300–600 | See the dimension row |
+
+## Remaining effort and calibration
+
+**Calibration.** The repository's history from the first commit (2026-09-30) to 2026-10-10: 1,152
+commits on `main` and 1,004 merged PRs in 10.5 days. Median PR open-to-merge time is 3.4 h
+(p75 6.4 h, p90 14.3 h; that includes CI and review queueing). Comparable features:
+
+| Feature | PRs | Lines added | PR hours |
+|---|---|---:|---:|
+| Crop tool parity set (#1919: nudge/swap/readout, overlays, Straighten, shield) | #1958, #2026, #2127, #2128 | ~2,050 | 6.5 |
+| Crop rotation (#1792) | #1879 | 710 | 2.8 |
+| Remove tool with non-local patch completion | #2193 | 2,005 | 9.0 |
+| OpenRaster open and save | #2257 | 1,331 | 8.9 |
+| Paint.NET import | #1998 | 1,461 | 7.0 |
+| Shape stroke controls | #2218 | 1,506 | 9.0 |
+| Red Eye tool | #1084 | 1,232 | 17.9 |
+| Korean catalog and live switching | #582, #604 | ~3,800 | 13.0 |
+
+So one checklist item of tool or UI depth costs about 2–9 agent-hours including tests, and a new
+format or subsystem 8–20. With roughly 2–4 agent-hours per merged PR, the ~1,000 PRs so far
+represent about **2,000–4,000 agent-hours**, which took ready-for-real-work from 0 to ~45%. The
+remaining points are harder per point (fidelity against Photoshop's unpublished maths, feel,
+performance on large documents, user-reported long tail), so the remaining estimate is not lower
+than the work already done.
+
+**Totals.**
+
+| Milestone | Scope | Opus 5.5 agent-hours |
+|---|---|---:|
+| **Beta** (~75% ready, PSD reliable) | PSD fidelity and Photoshop re-open check, the performance budgets, the user bug backlog, the 15 missing tools, Select and Mask / Content-Aware Fill workspaces, type depth, UI feel (brush, transform, panels), JPEG/PDF/AVIF/JXL formats, Hindi/Arabic/Vietnamese and RTL | **1,100–1,800** |
+| **Full parity** | the above plus the AI tier, plug-in/scripting ecosystem, video depth, long-tail fidelity | **2,300–4,100** |
+
+**Parallelism.** About 80% parallelizes: areas are independent crates and command modules, and
+the repository has sustained 10–20 concurrent agents. At ~12 agents, beta is roughly 100–150
+wall-clock hours. Serial parts: GPU compositor and performance architecture, the PSD writer.
+
+**Needs a human.** Owner decisions on the AI backend (#41: which openly licensed models, local
+vs remote) and plug-in hosting (.8BF needs native code under the isolated-`unsafe` rule; UXP is a
+JavaScript runtime, which AGENTS.md forbids in the UI); Windows code-signing material; tablet
+hardware checks (Wacom/XP-Pen on Windows, Wayland, macOS); native-speaker review of every
+catalog; checking exports by opening them in Photoshop (the owner's machine has it installed).
+
+## How to re-measure
+
+1. `cargo xtask parity` and `cargo xtask scorecard` for the measured rows.
+2. Tools: diff `Tool` in `crates/ui-egui/src/state.rs` against the `<tool>` entries of
+   Photoshop's `Locales/en_US/Support Files/Shortcuts/Mac/Default Keyboard Shortcuts.kys`.
+3. Formats: Photoshop's `Info.plist` `CFBundleDocumentTypes` and the PiPL flags of
+   `Contents/PlugIns/Required/File Formats`, against `crates/codecs/src/format.rs` and
+   `crates/io/src/lib.rs`.
+4. Languages: `cargo xtask i18n-coverage`.
+5. Re-grade the estimated rows from the open issues by area and the corpus floors. Move a number
+   only with new evidence, and add a revision-history row.
+
+---
+
+# Earlier assessments (kept for the record)
+
+The two sections below are superseded by the tables above. They are kept because they hold dated
+evidence that later numbers build on.
+
+## Feature-parity estimate of 2026-10-03 (from the former `docs/parity-estimate.md`)
+
+How close PhotoCraft is to Adobe Photoshop 2026, and how much Opus 5.5 work remains. Generated by
+assessment, not a tool; revisit when a major arc lands. Menu coverage is from `cargo xtask parity`;
+fidelity is from the PSD oracle (`cargo xtask corpus` / `oracle_diff`).
+
+### Headline
+
+> **Update 2026-10-03:** menu/command breadth reached **100% (625/625)** — every Photoshop
+> menu item now dispatches a live command (incl. Variables/Data Sets, Trap, the Timeline panel and
+> Layer › Video Layers). Remaining parity work is depth and fidelity, not breadth: PSD oracle
+> (~66%), Camera Raw/video-layer depth, and the AI/Neural tier. See the table below.
+
+
+- **Menu / command breadth:** 625 / 625 Photoshop menu items live = **100%** (every menu item has a live command).
+- **PSD rendering fidelity (oracle):** ~111 / 170 corpus files pixel-match = **~65%**.
+- **Overall functional parity (weighted, mainstream workflows):** **~86%** (menu breadth complete; depth + fidelity are the remaining gap).
+- **Overall including Adobe's AI / Firefly features:** **~75%** (these are mostly cloud/ML and the
+  hardest to reach clean-room).
+
+Menu coverage (96%) overstates true parity because many live commands are faithful-but-approximate
+(Camera Raw without masking/calibration, Vanishing Point without its UI, heuristic Select Subject/Sky
+vs Adobe's ML, Content-Aware Fill basic). The weighted figure accounts for depth and fidelity, not
+just presence.
+
+### Where we stand, by area
+
+| Area | Parity | Notes |
+|---|---:|---|
+| Layers, masks, groups, blend modes, smart objects, layer styles | ~92% | Broad + GPU-accelerated; bevel/satin shapes and a few blend edge cases remain |
+| Selections, paths, vector shapes, pen | ~90% | Select menu 100% of items; Select Subject/Sky are heuristic, not ML |
+| Paint / brush engine, retouch (clone, healing) | ~85% | Solid; no Mixer Brush depth, no Content-Aware healing |
+| Adjustments + adjustment layers | ~90% | All kinds present; Camera Raw depth and a few toe-curve models approximate |
+| Filters (incl. Filter Gallery ×47, Liquify, blurs, render) | ~88% | Gallery looks are approximations; no Neural Filters |
+| Type / text engine, character & paragraph styles, glyphs | ~90% | Vertical type laid out and rendered (#199; no mojikumi or upright-Roman option yet); OpenType alternates thin |
+| Color management (ICC v2/v4, intents, soft-proof) | ~90% | Own pure-Rust CMS; no Adobe CMYK profiles (synthetic) |
+| File I/O: PSD, native .pcraft, flats (png/jpg/tiff/webp/exr/…) | ~85% | PSD oracle ~65% pixel-exact; most structure round-trips |
+| Transform, warp, puppet, perspective, liquify | ~88% | No Face-Aware Liquify (needs a landmark model) |
+| Automation: CLI, MCP, headless server, control channel | ~95% | Ahead of Photoshop here (see below) |
+| Image modes (RGB/Gray/CMYK/Lab/Indexed/Duotone/Multichannel) | ~85% | Multichannel display-only; no ink-plate model |
+| Print / export / Save for Web / slices / Generator | ~85% | No Separations/bleed; droplets basic |
+| Computational photo (Photomerge, HDR, Lens, Camera Raw, Wide Angle) | ~70% | Classical implementations; generic lens profiles only |
+| Data-driven graphics (Variables, Data Sets) | ~0% | Not started (5 menu items) |
+| Video / Timeline / animation | ~0% | Not started (13 menu items) — a whole subsystem |
+| AI / Neural / Generative (Firefly, Sensei, Neural Filters) | ~5% | Mostly absent; needs licensed/trained models |
+| Prepress niche (Trap, measurement scale pro) | ~30% | Trap, ink spread not modelled |
+
+### The remaining 25 menu items, grouped
+
+- **Video & Timeline (13):** Video Layers (new/insert/duplicate/delete/replace/interpret/restore/
+  reload/rasterize), Rasterize › Video, Timeline panel, Render Video, Import Video Frames to Layers.
+  One cohesive subsystem: a frame/time model, a video decoder/encoder (clean-room Rust, no ffmpeg),
+  onion-skinning, and a timeline UI.
+- **Data-driven graphics (5):** Variables › Define / Data Sets, Apply Data Set, Export Data Sets as
+  Files, Import Variable Data Sets. A template-binding model over layers + a CSV/data-set engine.
+- **Niche / platform (7):** Trap (prepress ink spread), Frame from Layers, Migrate Presets, WIA
+  Support (Windows scanners), and two already-adjacent items.
+
+### Effort to parity (Opus 5.5)
+
+Two clocks: **agent-hours** (total model work) and **wall-clock** (with this session's observed
+~6–8× parallelism across independent arcs). This session landed ~70 menu items + major subsystems
+in a handful of wall-clock hours using 8 parallel agents, so wall-clock ≈ agent-hours ÷ 6.
+
+| Milestone | Scope | Agent-hours | Wall-clock |
+|---|---|---:|---:|
+| **To 90% functional** | remaining non-AI/non-video menu items (Variables, Trap, Frame, Migrate, WIA), Camera Raw depth, Vanishing Point UI, selection quality, polish | 60–90 | 10–15 h |
+| **To 95% functional** | the above + PSD oracle 65% → ~90% (bevel/satin shapes, Lab blending, noise-gradient RNG, toe curves), broad UI-parity pass, perf pass | +80–130 | +15–22 h |
+| **Video / Timeline subsystem** | frame model, clean-room decode/encode, timeline, animation, Render Video | 50–80 | 10–14 h |
+| **Credible AI tier** | ML Select Subject/Sky, Content-Aware Fill (PatchMatch++), basic generative fill/expand — needs model integration; some features may be infeasible fully clean-room without licensed models | 120–250+ | 25–45 h |
+
+**Totals:**
+- To **~95% functional parity** (everything except video and AI): **~140–220 agent-hours ≈ 25–40
+  wall-clock hours** of Opus 5.5 with parallel agents.
+- To **~98%** incl. the video subsystem: **+50–80 agent-hours ≈ +10–14 wall-clock hours**.
+- To **~99%** incl. a credible AI tier: **+120–250 agent-hours ≈ +25–45 wall-clock hours**, with the
+  caveat that pixel-for-pixel Firefly/Sensei parity is not achievable clean-room without trained
+  models; we target functional equivalents.
+
+**Best single number:** PhotoCraft is **~82% of the way** to full Photoshop parity, and reaching a
+**95%, ships-for-most-users** state is roughly **25–40 wall-clock hours** of parallel Opus 5.5 work.
+Closing the last 5% (video + AI) is the long, uncertain tail.
+
+### Assumptions & caveats
+
+- "Parity" = functional + visual equivalence of what a user can do, not byte-identical output. Clean-
+  room means no Adobe code, profiles, shaders, or trained models; AI features target equivalents.
+- Wall-clock assumes 6–8 independent arcs in parallel with their own target dirs, as this session ran.
+  Serial work is ~6× longer.
+- Adobe removed 3D/Substance from Photoshop, so 3D is intentionally out of scope.
+- Fidelity hours are the least certain: the remaining oracle failures are proprietary RNGs and
+  unpublished curve models fitted by observation.
+
+## Honest parity assessment of 2026-10-05, with dated evidence to 2026-10-10 (from the former section of `docs/roadmap.md`)
+
+This is the reference answer to "how close are we to Photoshop parity, really". Agents: read it
+before picking work. Update it (with dated measurements) when the numbers move; don't restate the
+menu-parity number in its place.
+
+**Measured numbers live in the [scorecard](scorecard.md)** (`cargo xtask scorecard`, kept current
+by CI): performance scenarios against their budgets, corpus floors, per-area checklists with
+done / partial / missing counts, and the number of settings that do nothing. Where the scorecard
+and the estimates below disagree, the scorecard's numbers supersede them. As of 2026-10-05 it
+counts 67 of 129 preferences that nothing reads; 3 of the 25 budgeted performance scenarios meet
+their #209–#211 targets (first baseline, taken on a heavily loaded machine), and the 150-layer
+nudge scenario crashes the GPU compositor; the open area issues' checklists (#203–#220) are
+almost entirely missing or partial.
+
+2026-10-07: UI Font Size now applies to interface text, including CJK fallbacks, independently
+of display and canvas zoom (#532). The current preference audit drops from 59 to 58 unread
+settings out of 135; see the regenerated scorecard.
+
+2026-10-09: Output-preserving Radial Blur direct sampling measured on a local
+i7-9750H, 12 workers, 24 MP RGBA8, Good quality, amount 1: Spin 66.10 → 16.43 s
+(4.02×), Zoom 19.19 → 5.35 s (3.59×), one paired run each, exact output equality.
+These are local measurements at amount 1; see [method and limits](radial-blur-performance.md).
+
+2026-10-10: native selection distance transforms measured 7.59–10.57x faster on six
+24–36 MP synthetic masks on an AWS c7i.4xlarge (16 workers, three paired release runs).
+The transform also corrects f32 envelope errors past coordinate 4096, including nonzero
+distances at selected pixels. The parallel path needs an additional four bytes per pixel;
+see [measurements, correctness and limits](selection-distance-performance.md).
+
+**Bottom line.** Two days after 0.2.0 we had merged ~96 PRs and closed ~48 issues, but **real
+Photoshop parity is still well below 50%**. The biggest gaps are AI, missing tools, professional
+workflow depth and the plug-in ecosystem. Most fixes since 0.2.0 have passed our tests but have
+**not yet been validated by users**, and this week proved our tests miss what users hit:
+on first public use they found broken basics (text selection offset, 214 shortcut failures, panels
+resizing themselves, an immovable crop frame, folders that wouldn't collapse, lag on layout PSDs)
+that all counted as "live" in `parity-checklist.md`.
+
+**Overall.** Feature surface: roughly **60–70%** of what a typical Photoshop user touches exists in
+some form. "A professional could switch for daily work": roughly **25–35%**. True 1:1 parity is
+**many months** of focused work, and some areas need product decisions rather than effort.
+Confidence: moderate — the next users of 0.2.x will move these numbers either way.
+
+### By dimension
+
+2026-10-10: Iris Blur pin preparation, saturated-distance evaluation and trailing
+Gaussian-level halo trimming measure 1.18–1.76× faster across eight local 24 MP
+RGB U8/U16/F32 cases (Ryzen 7 9700X, Windows, eight workers). Default blur 15:
+626.8 → 426.9 ms; two pins: 1789.9 → 1016.2 ms. The default 1500×1000 filtered
+proxy for 24 MP, document blur 15/80: 28.7/51.8 → 24.9/42.6 ms. At 36 MP,
+two pins: 2595.8 → 1822.8 ms; small/medium U8 gains are inconclusive.
+Differential tests preserve exact finite pixel bits.
+These are computation timings, not GUI frame latency or new Photoshop parity;
+see [measurements, upstream SIMD findings and limits](iris-blur-performance.md).
+
+2026-10-10: Quick Selection object strokes through raster surfaces measured
+3.46-3.70x faster at 8-bit, 16-bit and 32f on 24-36 MP documents (AWS
+c7i.4xlarge, five paired release runs), with identical region bounds and mask bytes.
+Hybrid FIFO push/relabel and sparse edge enumeration preserve the fitted energy.
+See [stroke measurements and reproduction](quick-selection-performance.md).
+
+2026-10-10: Pixelate's Facet shares exact 3x3 quadrant statistics and Color Halftone
+shares dot radii within each rotated screen grid. Full tiled CPU application at 24 MP
+on a 16-vCPU m7i.4xlarge improves Facet from 0.684 s to 0.312 s in RGB8
+(2.20x), and from 7.888 s to 0.574 s in CMYK8 (13.74x).
+Color Halftone improves from 1.137 s to 0.604 s in RGB8 (1.88x)
+and from 3.589 s to 0.718 s in Lab8 (5.00x).
+Three alternating paired release runs also cover 36 MP RGB, 16/32-bit storage,
+and screen radii 2/8/64. All 36 complete output comparisons match exactly.
+See [method, raw timings and checks](pixelate-performance.md).
+
+2026-10-10: Native CMYK solid and gradient fills share depth-quantized samples between
+rendering and PSD export. The uncached 16-bit CMYK gradient now round-trips with zero
+rendered error; psd-tools round trips rise from 307 to 308. Mixed and Photoshop round
+trips remain 169 and 258; rendering oracle counts remain 146, 237 and 133.
+Exact 8-bit CMYK row reuse reduces 1 MP conversion-read time by 77% for smooth ramps,
+33% with dithering and 10% for random colours (M4 Max, release optimization level 3).
+
+2026-10-10: [Shape stroke controls](shape-strokes.md) expose existing Solid/Dashed/Dotted,
+custom dash/gap, offset, caps, joins, alignment, miter-limit and opacity capabilities in
+Shape/Pen options and Properties. Properties stages a cached canvas preview before one
+undoable edit. Gradient fill controls from #1053 remain open; the Line tool still draws a
+filled bar. This extends UI reachability and does not claim Photoshop-authored stroke parity.
+
+| Dimension | Measured / evidence (2026-10-05) | Grade | Notes |
+|---|---|---|---|
+| Menu wiring | 626/626 menu items dispatch a command (`parity-checklist.md`) | high but shallow | Says nothing about behaviour. |
+| PSD fidelity (rendering) | Corpus oracle 115/170 (68%): 30 differ, 26 have no usable reference, 1 import error. 2026-10-07: io corpus 146/170, psd-tools corpus 236/309 (was 229: Advanced Blending knockouts), Photoshop oracles 132/258 | medium | Push to 170/170 under way (effects/strokes, multi-instance effects, 16/32-bit and colour modes, references for skipped files). |
+| PSD round trip | 169/169 re-import identically; every adjustment layer and blend mode round-trips | high (within corpus) | Floors in `crates/io/tests/corpus.rs`; raise, never lower. Corpora: `cargo xtask corpus --all` (ours: https://github.com/storytold/photocraft-corpus). |
+| OpenRaster read/write | 2026-10-09: 34 Krita 5.2.9-authored `.ora` files (27 blend modes, groups, pass-through, offsets, opacity, visibility, masks, 16-bit, gray) render within 3/255 of Krita's merged image (30) or differ for known reasons (4: Krita's lighter/darker-colour tie-break, Hard Mix at exactly 1, a linear-light 16-bit document); Krita re-opens our exports and renders them identically to its originals (33/34; Soft Light is written as `svg:soft-light`) | medium (one authoring app) | MyPaint and GIMP files untested. Masks are applied to pixels and layer styles dropped on save (reported). See [OpenRaster](ora.md). |
+| Paint.NET import | 2026-10-09: 14/14 local PDN3 textures preserve editable layers exactly through `.pcraft` and render within 2.142/255 of full-size previews; a supplied 5.x document (1280×720, 11 layers) passes editing/history/native saves and matches its thumbnail within 0.418/255 mean; all 14 blend modes exercised synthetically | medium (limited corpus) | Import only; metadata omitted. More current-version files and full-size exports needed. See [PDN support](pdn.md). |
+| Smart filters / text / effect shapes in PSDs | Measured on our Photoshop-authored set (https://github.com/storytold/photocraft-corpus, `corpus/photoshop`, 258 files): see the per-group floors in `crates/io/tests/corpus.rs` and `crates/engine/tests/photoshop_oracles.rs`. Smart objects and smart filters now survive PSD save and open (41 corpus files, 206 smart objects, round trip strict; Photoshop opens our exports with live filters); re-rendering Photoshop's smart filters with ours matches 5/30 (was 1/30) | low–medium | Remaining re-render gaps are filter maths (Gaussian/Motion Blur, Unsharp Mask, Emboss, Add Noise RNG) and bicubic placement. |
+| Core editing (layers, masks, selections, adjustments, filters, transforms) | Broad engine coverage; many interaction bugs fixed after 0.2.0 (adjustment dialogs, Curves, crop, Move/Transform modifiers, gesture origin) | medium | Fixes not yet user-validated. |
+| UI / UX polish | Shortcut audit 214 → 0 failures; dock, Layers rows and menus reworked; first visual-QA sweep found 14 defects (#147–#157). 2026-10-07: the keyboard-only shortcuts with no menu item (⌥[ ⌥] ⌥, ⌥. layer navigation, ⇧⌥[ ⇧⌥] to extend the selection, 1–0 for opacity and ⇧ for flow or fill, ⇧[ ⇧] hardness, ⌥⌘T to transform a copy and ⌥⇧⌘T to step and repeat), ⌥-click colour sampling with painting tools, double-click a Layers row for Layer Style, File › New from Clipboard, a centred main window and remembered Liquify settings (#352, #417, #350, #368, #419, #418). 2026-10-09: canvas zoom is physical at any display scale — 100% is one document pixel per physical display pixel, and every screen-space overlay, cursor, scrollbar, navigator and fit follows it (#1943) | low–medium | Needs recurring visual QA with realistic documents. |
+| Tools | 2026-10-09: Pencil, Mixer Brush, Patch, Content-Aware Move, Vertical Type, Pattern Stamp and Rotate View are toolbar tools. Remaining missing tools include Red Eye, Art History Brush, Freeform/Curvature Pen, Add/Delete Anchor Point tools, single row/column marquee, Color Sampler, Perspective Crop, type masks and Frame | low–medium | Patch has a live healing preview. Content-Aware Move is partial: Move/Extend, Structure/Color and Sample All Layers exist; Transform On Drop and a live result preview remain missing (`TOOL-213-4` in the scorecard). Magic/Background Eraser added; live gradients in progress (#180). Magnetic Lasso added 2026-10-08 (live-wire edge tracing, Width/Contrast/Frequency, `select.magneticLasso`). Pattern Stamp added 2026-10-08 (S flyout, `paint.patternStamp`, Aligned / Impressionist). Rotate View (2026-10-09) turns the canvas camera around its centre without rewriting pixels; Reset View and Match Rotation copy the angle. Crop rotation (2026-10-09, #1792): dragging outside the crop frame turns it (⇧ 15° steps, angle readout), and ↵ rotates the document and crops in one undo step (`image.crop` `angle`); the frame turns over the image as in Photoshop's Classic Mode, and the options-bar Straighten is still missing. |
+| Painting | Brush model and Brush Settings panel near Photoshop; .abr/.grd import; persistent presets; pen pressure/tilt on Windows, web, macOS and X11 | medium | Native Wayland pen input open (#79; a pen opens the window through Xwayland meanwhile); X11 pressure confirmed by a user, macOS not yet verified on tablet hardware. |
+| Text / typography | Engine works; caret placement and size editing fixed; OpenType features, text-on-path editing, composer parity partial | medium-low | Measure with the Photoshop-authored set. |
+| Colour management | Colour-managed canvas (document → monitor), embedded CMYK profiles, linear EXR/HDR, 16-bit float canvas | medium-high | Monitor profile follows only at launch. |
+| Performance | 14k+ px on the GPU at ~⅓ the memory; adjustment preview 285 ms → 4–9 ms; font-size edits 297 ms → 4.6 ms; 2026-10-07: 30 MP TIFF open (banded, parallel strip/tile decode) Deflate 345 → 32 ms, LZW 428 → 43 ms, BigTIFF and every IFD readable; 2026-10-09: 4.2 MP Indexed Color, 256 colours, full-resolution CPU preview p50 3018 → 1061 ms (Ryzen AI 7 350, three measured runs; GPU upload excluded) | medium-high on rasters | Complex layout documents still laggy (#125/#128); >16384 px GPU tiling in progress (#49). Native Indexed Color previews run on one background worker with stale-result rejection; large palettes can still take about a second to compute. Web previews remain synchronous. |
+| Stability | Never-crash lint series, crash guard, `panic_hunt` fuzzing in the gate | medium-high | No field crash data yet. |
+| Camera RAW | DNG, CR2, Sony ARW (lossless + compressed), Nikon NEF (lossless + lossy compressed), RW2, uncompressed ORF; 2026-10-09: uncompressed Fujifilm RAF, Bayer and X-Trans (edge-directed X-Trans demosaic; 26 MP X-Trans develop 306 ms on 12 cores) | medium | CR3, compressed RAF, Nikon "lossy after split" and calibrated colour for non-DNG cameras (no Fuji colour calibration: neutral fallback) still open (#50). |
+| AI / generative | none | ~0% | Deferred by decision (#41). |
+| Ecosystem | Sandboxed WebAssembly plug-ins instead of .8BF; no ExtendScript/UXP/.atn; no Adobe Fonts/Libraries/cloud docs | low | By design for 8BF; scripting compatibility open. |
+| Platforms | macOS (notarized), Windows, Linux (AppImage/deb/rpm/Flatpak bundle), web | medium-high | Flathub later (#173); Windows signing material pending. |
+| Localisation | 2026-10-07: 10 UI languages; menu, `tl!`, blend mode, preference and brush-section coverage enforced by tests; live switching and scoped Preferences previews | medium | Engine errors/status messages still partly English; CJK web fonts, browser-locale detection, and RTL remain open. |
+
+2026-10-08: Motion Blur adds adaptive FFT convolution for wide streaks while retaining the
+row kernel merged in [#902](https://github.com/storytold/photocraft/pull/902) for shorter
+streaks and unsupported cases. Native release medians of three paired runs on an Intel
+i7-9750H with 12 Rayon workers, synthetic RGBA8 at 30° / distance 1000: 24 MP
+22.594 → 9.785 s (2.31×), and 1.5 MP
+2.455 → 0.592 s (4.15×).
+Full-image comparisons differ by at most one U8 code level; distance 64/256 retain the
+byte-identical row output. These measure decode/halo/filter/write/prune, excluding UI proxy
+creation, compositing and upload; the UI remains synchronous. FFT work is scheduled within
+a conservative 512 MiB working-set estimate, excluding stored Surface tiles and row fallback
+memory. The crossover is a heuristic; images dominated by near-cutoff alpha can need extra
+scalar work. These measurements do not establish Photoshop filter parity.
+
+2026-10-10: Proximity Match search medians on an AWS c7i.4xlarge improved 1.54–3.82x
+for four synthetic stroke regions (50–300 px holes), with identical source displacements.
+Bounding the mask table to the hole and pruning losing SSD candidates reduced table storage
+by about 90–93 percent there. Full 24–36 MP kernel inputs improved 4.60–6.24x, but normal
+engine strokes already use a cropped region. See [method and limits](proximity-match-performance.md).
+
+2026-10-09: the Remove tool is in the J flyout (`paint.remove`, a background job). A stroke
+around an object also removes what it encloses. The fill is non-local patch completion with
+texture features (A. Newson et al., IPOL 2017: `crates/algo/src/nonlocal.rs`), finished by a
+best-patch copy and gradient-domain seam hiding. On 20 holes in public-domain photos
+(`remove_quality`, `docs/development.md` › Remove Tool quality) the fill keeps 0.89 of the
+original's texture (Wexler/PatchMatch completion as Content-Aware Fill uses it: 0.54). On a 24 MP
+document a 100 px ring takes 120 ms and a 1000 px ring 2.5 s (`remove_bench`). It is not
+generative: large objects over complex structure fill less convincingly than Photoshop's AI mode
+(#41).
+
+2026-10-08: the tools update above is checked against the current toolbar groups in
+[`panels.rs`](../crates/ui-egui/src/panels.rs), the Pencil and tool-cycle tests in
+[`pencil_tests.rs`](../crates/ui-egui/src/pencil_tests.rs), Patch's live-preview test in
+[`patch_preview.rs`](../crates/ui-egui/src/patch_preview.rs), and Vertical Type's point/paragraph
+creation and undo test in [`type_tool_tests.rs`](../crates/ui-egui/src/type_tool_tests.rs).
+These implementations remove the tools from the missing list; they do not establish Photoshop
+behavioural parity. The historical estimates and corpus measurements in this assessment are
+unchanged; current measured floors remain in the scorecard.
+
+2026-10-07: Camera Raw PSD mapping covers relative custom white balance, Light/Presence,
+parametric and four point curves, HSL, Color Grading, sharpening/noise detail, grain and numeric
+post-crop vignette controls. Two revisions of one supplied Photoshop ACR 18.4 PSD preserve their
+filter descriptors without edits. The updated revision stays opaque because of active manual
+Optics and unverified vignette style; a descriptor projection verifies its supported controls.
+Synthetic 8/16/32-bit PSD → edit → `.pcraft` → PSD tests preserve the filter stack, mask and
+undo/redo. Unmapped Camera Raw fields/versions remain opaque; Photoshop
+acceptance of generated exports and pixel parity are still unverified. Corpus floors above are
+unchanged.
+
+2026-10-10: Select and Mask's Shift Edge now computes the exact octagonal footprint
+with a square window extreme plus a dyadically built Manhattan ball. On a c7i.4xlarge
+(16 vCPUs), full tiled CPU refinement at radius 64 / Shift Edge +100% takes
+3.019 s -> 0.762 s at 24 MP (3.96x),
+3.706 s -> 0.921 s at 36 MP (4.02x).
+Three alternating paired release runs include guide sampling, guided filtering,
+boundary distances, quantization and assembly. Every output mask matches exactly.
+The dense fractional 24 MP radius-64 dilation kernel improves 36.00x;
+circular soft-mask kernels improve 20.76-172.16x at 24-36 MP, radii 4-64.
+Bounded halo tiles reduce the large-mask workspace allocation bound from 192/288 MB
+to about 115/163 MB, including output and 16 workers' scratch, excluding the source.
+See [method and results](octagonal-mask-performance.md).
+
+### Where we're going (priority order)
+
+1. **Ship the fixes:** cut 0.2.1 once the current batch lands, so users validate them.
+2. **PSD fidelity to 170/170** with round trips, plus the Photoshop-authored reference set for smart
+   filters, the text engine and effect shapes (https://github.com/storytold/photocraft-corpus, fetched by
+   `cargo xtask corpus --all`); enforce floors in `corpus.rs`.
+3. **Workflow acceptance tests:** 30–50 real tasks (retouch a portrait, social post with text and
+   effects, composite with masks and adjustment layers, CMYK print prep…) scripted end to end and
+   checked against Photoshop's output on every build. Their pass rate becomes the headline parity
+   number.
+4. **Tool coverage and workflow depth:** Pen variants (Direct Selection landed, #790) and
+   Perspective Crop; finish Content-Aware Move's Transform On Drop and live result preview.
+   Pencil, Mixer Brush, Patch, Vertical Type, Pattern Stamp and Rotate View are already available on the toolbar.
+5. **Complex-document performance** (#125/#128) and GPU tiling beyond the texture limit (#49).
+6. **Recurring visual QA** (`cargo run -p photocraft-engine --example designer_psd`) and fast
+   turnaround on user reports (OS, document size, layer count, screenshot).
+7. Later / needs decisions: generative AI backend (#41), scripting compatibility (ExtendScript /
+   UXP / .atn), Flathub (#173), Wayland pen pressure (#79).
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | minor | Readiness-by-audience table: percent and hours for full, mainstream and essentials; full number confirmed as the additive weighted sum over dimensions (46%, reported ~45%) |
+| 2026-10-10 | minor | Added mainstream-practitioner (~43%) and essentials-user (~61%) numbers with written weights and discounts; full ready-for-real-work rechecked against its written weights (46%, reported ~45%), unchanged |
+| 2026-10-10 | minor | Stage checked against the core-workflow alpha gate in `roadmap.md`: passes |
+| 2026-10-10 | major | Full re-measure against Photoshop 2026 27.11.0 (installed bundle inspected); two numbers, weights, per-dimension and per-area hours calibrated from git history; merged `parity-estimate.md` and the roadmap's honest assessment into this file |
+| 2026-10-05 | major | Honest parity assessment in `docs/roadmap.md`: surface 60–70%, ready for real work 25–35% |
+| 2026-10-03 | major | `docs/parity-estimate.md`: menu breadth 625/625, weighted ~82–86% |
@@ -0,0 +1,206 @@
+# UI parity: tools, handles, modifiers and feel
+
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** major (first version: tools, transform, nudging, snapping, modifiers, navigation, shortcuts, panels audited from source and open issues) · **Target:** Adobe Photoshop 2026
+
+How PhotoCraft's direct manipulation compares with Photoshop 2026 (27.11.0): the toolbox, the
+handles and modifier keys of each gesture, nudging, snapping, numeric precision, navigation,
+shortcuts and panels. Painting feel has its own checklist in [`brush-parity.md`](brush-parity.md)
+and right-click menus in [`context-menu-parity.md`](context-menu-parity.md). The design system
+(themes, tokens, widgets) is [`ui-design.md`](ui-design.md).
+
+Status: **done** (behaves as in Photoshop, with a test), **partial** (exists, differs or untested),
+**missing**. Evidence is a file under `crates/ui-egui/src/` unless another path is given, a
+scorecard id, or an issue. Modifier names are macOS first (⌘ ⌥ ⇧); Windows/Linux use Ctrl and Alt.
+
+**Summary (estimated, weights = how often the gesture is used):** ~40% ready. Tools 53/68
+present (78%, measured). Free Transform modifiers and Crop are close to Photoshop; the gaps are
+in the long tail of modifiers, in-transform UI, spring-loaded tools, panels and preferences.
+Remaining: 60–110 h for transforms and modifiers (gap G5), 60–100 h for tools (G6), 150–250 h
+for panels and workspace (G9).
+
+## Tools
+
+Measured: the `Tool` enum (`state.rs:64`) and `TOOL_SECTIONS` (`panels.rs:17`, 22 flyout slots)
+against the 67 `<tool>` entries of Photoshop's `Default Keyboard Shortcuts.kys` plus Remove,
+Triangle and Curvature Pen (minus legacy Rounded Rectangle and Targeted Adjustment) = 68.
+
+| Group | Present | Missing |
+|---|---|---|
+| Move, artboard | Move | Artboard (commands `layer.new.artboard*` exist) |
+| Marquee | Rectangular, Elliptical | Single Row, Single Column |
+| Lasso | Lasso, Polygonal, Magnetic | |
+| Selection | Object Selection, Quick Selection, Magic Wand | |
+| Crop and slice | Crop, Slice, Slice Select | Perspective Crop |
+| Frame | | Frame (`layer.new.frameFromLayers` exists) |
+| Eyedropper | Eyedropper, Ruler, Note, Count | Color Sampler (#1046) |
+| Retouch | Remove, Spot Healing, Healing, Patch, Content-Aware Move, Red Eye | |
+| Brush | Brush, Pencil, Mixer Brush | Color Replacement (`paint.colorReplacement` exists) |
+| Stamp | Clone Stamp, Pattern Stamp | |
+| History | History Brush | Art History Brush |
+| Eraser | Eraser, Background Eraser, Magic Eraser | |
+| Fill | Gradient, Paint Bucket | |
+| Focus, toning | Blur, Sharpen, Smudge, Dodge, Burn, Sponge | |
+| Pen | Pen | Freeform Pen, Curvature Pen (#1054), Add Anchor Point, Delete Anchor Point, Convert Point (⌥ with Pen only) |
+| Type | Horizontal, Vertical | Horizontal Type Mask, Vertical Type Mask |
+| Path selection | Path Selection, Direct Selection | |
+| Shapes | Rectangle, Ellipse, Triangle, Polygon, Line, Custom Shape | (Star is a Polygon option in Photoshop) |
+| View | Hand, Rotate View, Zoom | |
+| Toolbar controls | Edit Toolbar, colour chips, Quick Mask, Screen Mode | |
+
+| Toolbar behaviour | Status | Evidence |
+|---|---|---|
+| Letter shortcuts, ⇧+letter cycles the group, "Use Shift Key for Tool Switch" | done | `shortcuts.rs:413` |
+| Group remembers the last tool used | partial | L always resets to Lasso (#2608) |
+| Press-and-hold or click-drag opens the flyout | missing | right-click only (#2301) |
+| Edit Toolbar customization | done | `edit.toolbar` |
+
+## Free Transform (⌘T) and transform handles
+
+`transform_tool.rs` (3,158 lines), behaviour summary in [`ui-design.md`](ui-design.md#interaction-models-match-photoshop-cc).
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Eight handles, drag inside moves, outside rotates with rotation cursor and live angle | done | #1948 |
+| Corners and edges proportional by default; ⇧ frees them (edge stretches one axis) | done | #1873 |
+| ⌥ scales about the reference point; ⌥-click places it; draggable reference point | done | |
+| ⌘-corner distort, ⌘-edge skew, ⌘⌥⇧ perspective, ⇧ rotate in 15° steps | done | |
+| Warp: presets, bend, split crosswise/vertical/horizontal, 3×3–5×5 and custom grids | done | |
+| Options bar X/Y, W/H % with link, angle | done | |
+| Options bar H/V skew fields | missing | |
+| Interpolation in the options bar | partial | Bicubic/Bilinear/Nearest; no Bicubic Smoother/Sharper/Automatic, Preserve Details |
+| Commit/cancel buttons in the options bar | missing | #2442 |
+| Right-click menu during transform (Scale, Rotate, Skew, Distort, Perspective, Warp, flips, rotates) | missing | TOOL-214-5 |
+| Arrow keys nudge the box 1 px, ⇧ 10 px | done | `move_mods.rs` |
+| Preview keeps layer order, blend mode and mask | partial | mask yes (BUG-205-2); order/blend not (#2439) |
+| Preview clipped to the canvas, follows edits while open | done | #2201, #2099 |
+| Hand (space) mid-transform doesn't interrupt it | partial | #2302 |
+| Blend mode / opacity edits while transforming | missing | #1384 |
+| Locked layer refused before the drag | missing | fails only at commit (#2585) |
+| Transform a copy (⌥⌘T), step and repeat (⌥⇧⌘T) | done | 2026-10-07 |
+| Transform a targeted alpha channel | missing | BUG-207-4 |
+| Smart object perspective (non-affine) | missing | FILE-215-6, #579 |
+| Type-layer transform while editing (⌘ held), no ⌘-corner distort on type | done | `ui-design.md` |
+| Puppet Warp mesh follows the outline | missing | block grid (TOOL-214-6) |
+
+## Move tool and nudging
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Arrow 1 px, ⇧-arrow 10 px for layers | done | `move_mods.rs` |
+| Arrow nudge of a selection outline (selection tools) | done | #1502 |
+| Arrow nudge of the crop box, X swaps orientation | done | TOOL-214-19 |
+| Arrow nudge of path points with Direct Selection | missing | no arrow handling in `direct_select.rs` |
+| ⌘-arrow nudges the layer from any tool | missing | #2474 |
+| ⌥-drag duplicates, ⇧ constrains to 45° | done | `move_mods.rs` |
+| Auto-Select with hover outline, box-select layers, double-click fit | missing | TOOL-214-4 |
+| Auto-Select bounding box and snapping follow the picked layer | partial | follow the previous layer (#2666) |
+| ⌘-drag temporary Move shows the box and snaps | partial | #2681 |
+| ⌘-right-click layer list under the pointer | done | `layer_pick_ui.rs` |
+| Moving a selection on a mask moves the mask, not the pixels | partial | #2468 |
+
+## Selection modifiers
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| ⇧ square/circle, ⌥ from centre, live W × H readout (marquee) | done | `marquee_tests.rs` |
+| ⇧ add, ⌥ subtract, ⇧⌥ intersect from modifiers at press | partial | works; only a badge test (BUG-208-4) |
+| Space repositions the marquee while dragging | done | `hold_keys.rs` |
+| Polygonal Lasso ⇧ snaps to 45° | missing | #2352 |
+| Autoscroll at the window edge while selecting | missing | #2231 |
+| Single-pixel rows and columns kept on release | partial | #2391 |
+| Esc deselects / ⌘Z undoes the last selection | partial | #2674, #1432 |
+| Select › Modify live preview | missing | #1752 |
+
+## Pen, paths and shapes
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Click corner, drag smooth, ⌥ breaks a handle, ⌘ direct-selects, ⇧ 45° | partial | TOOL-214-10 |
+| Handle on the last point, close-drag handle | partial | #1482 |
+| Path operations in the options bar (combine, subtract, intersect, exclude) | partial | `layer.combineShapes.*`; Pen options #1397 |
+| On-canvas corner-radius handles on live shapes | missing | TOOL-214-12 |
+| Shape stroke controls (dash, caps, joins, align) | done | TOOL-1053-1, [`shape-strokes.md`](shape-strokes.md) |
+| Rotated shape selection hints correct | partial | #2512 |
+| Curved shape bounds exact | partial | one pixel too wide (#2537) |
+
+## Crop
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Drag outside rotates (⇧ 15°), angle readout; ↵ rotates and crops in one undo step | done | TOOL-214-18 |
+| Straighten button and ⌘-drag line | done | TOOL-214-21 |
+| Overlays (thirds, grid, diagonal, triangle, golden ratio/spiral), O / ⇧O | done | TOOL-214-20 |
+| Crop shield, Show Cropped Area (H) | done | TOOL-214-22 |
+| Default mode: the image turns and moves behind an upright box; Auto Center Preview; Use Classic Mode | done | TOOL-214-23 |
+| W × H × Resolution with units, built-in size presets, Front Image | done | TOOL-214-9 (`crop_size.rs`) |
+| Saved custom crop presets | missing | TOOL-214-9 |
+| Content-Aware crop fill, Delete Cropped Pixels off | partial | |
+
+## Snapping, guides, rulers, precision
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Snap to guides, grid, layers, document bounds and centre, selection, slices; ⌘ disables while dragging | done | `engine/src/snap.rs`, `snap_ui.rs` |
+| Snap vector tools and transforms to the pixel grid | done | `snap_ui.rs:196` |
+| Smart guides (magenta alignment lines) | done | |
+| Smart guide distance and spacing labels | missing | |
+| Ruler units px/in/cm/mm/pt/pica/% | done | |
+| Status bar follows the ruler unit | partial | #2602 |
+| Ruler corner and gaps drawn as Photoshop | partial | #2600, #2601, #2551 |
+| Guides stored as f64 | missing | `Vec<f32>` (UI-217-20) |
+| Geometry precision | partial | `geom::Point` f64, paths f64; paint internals mostly f32 |
+| Numeric fields: arithmetic, scrubby labels | partial | units in expressions and double-click reset missing (UI-217-13) |
+
+## Navigation
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Space = Hand, ⌘Space / ⌘⌥Space = zoom in/out, spring-loaded while held | done | `hold_keys.rs` |
+| Hold any tool letter for a temporary tool | missing | #2672 |
+| Scrubby zoom | done | `zoom_tool.rs` |
+| Flick panning | missing | `tools.enable_flick_panning` unread |
+| Animated zoom | missing | unread pref; #566 |
+| Double-click Zoom tool = 100%, double-click Hand = fit | partial | Zoom missing (#2508) |
+| ⌥-wheel zoom in gentle steps | missing | UI-217-18 |
+| Wheel, smooth scrolling and pinch as in Photoshop | partial | #2485, #2463, #1989 |
+| Trackpad rotate drives Rotate View | done | `rotate_view.rs:136` |
+| Zoom levels to 3200% and beyond | partial | #1813 |
+| Physical 100% at any display scale | done | #1943 |
+| ⌃Tab / ⌃⇧Tab switch document tabs | missing | #2340 |
+
+## Keyboard shortcuts
+
+| Measure | Value | Kind |
+|---|---|---|
+| Photoshop default shortcuts (`.kys`) | 58 tool keys, 107 menu commands with 115 shortcuts, 25 tool-area commands | measured |
+| PhotoCraft | ~199 shortcut assignments, 113 distinct modifier/F-key combos, 20 tool letters, 3 temporary-tool bindings | measured (source) |
+| Shortcut audit | 214 → 0 failures (2026-10-07) | measured then; not re-run here |
+| `.kys` import | done (`engine/src/kys.rs`) | |
+| Shortcut glyphs per platform | done (UI-217-19) | |
+
+## Panels and workspace
+
+30 of 35 Window-menu panels exist (measured). Missing: Libraries, Comments, Contextual Task Bar,
+Discover, Generative variations. Dock tabs: Layers, Channels, Paths, History, Actions, Layer
+Comps, Properties, Adjustments, Navigator, Histogram, Info, Color, Swatches, Character,
+Paragraph; the rest open as in-app floating windows.
+
+| Behaviour | Status | Evidence |
+|---|---|---|
+| Panel groups float and tear off into OS windows | missing | UI-217-5, #1828, #1213 |
+| Drag panels to reorder and regroup | partial | #2272, #2305 |
+| Docked resize pushes the neighbour | partial | #2573 |
+| Color panel wheel and RGB/HSB/CMYK/Lab sliders | missing | UI-217-17, #2569 |
+| Window size and dock width remembered | done | UI-217-2 |
+| Command palette with recents | done | UI-217-7 |
+| Every preference takes effect | partial | 46 of 149 do nothing (scorecard) |
+| UI scale presets | missing | UI-217-9, #2492 |
+| Themes: Photoshop's four greys | partial | five themes; #2661 |
+| Document tabs: overflow, context menu, reorder | done | UI-217-6; middle-click close #2570 |
+| Layers panel: wheel steps blend modes, drop on New/Trash | partial | UI-217-14 |
+
+## Revision history
+
+| Date | Change | Summary |
+|---|---|---|
+| 2026-10-10 | major | First version, from a source audit against Photoshop 2026 27.11.0's shortcuts file and the open issues |
@@ -1,5 +1,5 @@
 # Scorecard checklist: automation (#219). The command count and the menu wiring number come from
-# docs/parity.md.
+# docs/parity-checklist.md.
 # Statuses must be true on main. Verified against 4b53f62 on 2026-10-05.
 area = "automation"
 title = "Automation"
@@ -9,7 +9,7 @@ order = 6
 id = "AUTO-1"
 target = "Every menu item dispatches a command id (UI, CLI, control channel and MCP share them)"
 status = "done"
-note = "docs/parity.md: all menu items live; menus.rs is_live"
+note = "docs/parity-checklist.md: all menu items live; menus.rs is_live"

 [[item]]
 id = "AUTO-219-1"
@@ -35,7 +35,7 @@ commands:
                  --pixls (or -p raw) also fetches corpus/pixls and runs the raw corpus tests;
                  --local takes corpus/photoshop from the photocraft-corpus authoring clone
  stats [--exact] count tests and lines per crate (--exact: ask the test harness via `-- --list`)
-  parity          Photoshop menu parity; rewrites docs/parity.md
+  parity          Photoshop menu parity; rewrites docs/parity-checklist.md
  i18n-coverage   report stable UI translation coverage for registered languages
  perf [--quick] [--update-baseline] [--threshold PCT] [--bench NAME]... [--skip-build] [--reuse]
                  run the release benches, merge them by scenario id into target/perf/results.json,
@@ -182,7 +182,7 @@ fn cmd_wasm() -> Result<(), String> {

 fn cmd_parity() -> Result<(), String> {
    let mut c = cargo();
-    c.args(["run", "-q", "-p", "photocraft-ui-egui", "--example", "parity", "--", "--write", "docs/parity.md"]);
+    c.args(["run", "-q", "-p", "photocraft-ui-egui", "--example", "parity", "--", "--write", "docs/parity-checklist.md"]);
    run(c, "cargo run -p photocraft-ui-egui --example parity")
 }

@@ -7,7 +7,7 @@
 //! - `crates/io/tests/corpus.rs`: the corpus floors (`Source { .. }` constants).
 //! - `scorecard/*.toml`: per-area checklists (id, target, status, issue, note).
 //! - the prefs audit: `Preferences` fields that no code outside `prefs.rs` / `prefs_ui.rs` reads.
-//! - counted from the tree: never-crash attribute coverage, `docs/parity.md`'s live count.
+//! - counted from the tree: never-crash attribute coverage, `docs/parity-checklist.md`'s live count.
 //!
 //! `--check` fails when `docs/scorecard.md` differs from what would be generated.
 #![deny(clippy::unwrap_used, clippy::expect_used, clippy::panic, clippy::unimplemented, clippy::todo, clippy::unreachable)]
@@ -432,9 +432,9 @@ fn never_crash(root: &Path) -> (usize, Vec<String>) {
    (roots.len(), missing)
 }

-/// `**626 of 626 menu items live (100.0%).**` from `docs/parity.md`.
+/// `**626 of 626 menu items live (100.0%).**` from `docs/parity-checklist.md`.
 fn parity_line(root: &Path) -> Option<String> {
-    let s = std::fs::read_to_string(root.join("docs/parity.md")).ok()?;
+    let s = std::fs::read_to_string(root.join("docs/parity-checklist.md")).ok()?;
    s.lines().find(|l| l.contains("menu items live")).map(|l| l.trim().trim_matches('*').trim_end_matches('.').to_string())
 }

@@ -610,7 +610,7 @@ It is a heuristic lower bound: a generic name can look read when it isn't.\n\n"
            }
            "automation" => {
                if let Some(p) = &inp.parity {
-                    s.push_str(&format!("Menu wiring (`docs/parity.md`): {p}. This counts dispatch, not behaviour.\n\n"));
+                    s.push_str(&format!("Menu wiring (`docs/parity-checklist.md`): {p}. This counts dispatch, not behaviour.\n\n"));
                }
            }
            "reliability" => {