Files
2026-11-17 06:46:10 +00:00

26 KiB
Raw Permalink Blame History

TextureCraft roadmap

Stage: alpha (readiness assessment) · next: beta, ~75% readiness plus scoped compatibility and integration gates; remaining effort unestimated

Last reviewed: 2026-11-16 · Last updated: 2026-11-16 · Change: major (UI/functionality audit, possible format/package reworks, unfinished .mcmat integration and incoming ArtCraft UI/shared-library work) · Target: reference app 12.2.0

TextureCraft is a clean-room implementation of the professional texture-painting workflow. This assessment covers main at 62b2b13, version 0.6.1, compared with the installed reference app on 2026-11-16. It follows the PhotoCraft roadmap format. Detail: UI gaps and functional task audit, features, menu checklist, architecture, automation, localization, releasing.

Format direction: .tcraft remains the native editable project format. Read-only opening/import of .spp is a future compatibility goal: leave the source untouched and save converted work as .tcraft. Supported versions, recoverable layers/resources and handling of unsupported content still need investigation; no .spp reader is implemented. Writing .spp, round-tripping edits back into it and compatibility with other proprietary asset/preset formats are outside this goal. Pipeline exchange uses USD, glTF/GLB, MaterialX and texture maps.

The current .tcraft schema/container and output-package contracts may be reworked; their present implementations are not a frozen design. MaterialCraft .mcmat import remains a goal whose next format is unfinished upstream. ArtCraft UI refactoring and further shared-library adoption also add integration and regression work. These dependencies are part of the remaining roadmap, even where a current task already passes.

Headline numbers

Value Kind
Feature breadth ~80% (75–85%) Estimated across feature families; no exhaustive target inventory
Ready for real work ~60% (50–70%) Estimated readiness of the assessed revision; no completion credit for incoming reworks/integrations
Mainstream practitioner ~60% (50–70%) Estimated for regular asset texturing, baking, masking and pipeline delivery
Essentials user ~75% (65–85%) Estimated for a simple mesh, layered materials/paint and texture export
Desktop menu catalog 59/59 live Observed through the running app; this is TextureCraft's own catalog
Engine commands 178 registered Observed through the running app; registration does not prove every parameter path
Functional review 20 scoped check groups UI, mixed-surface and command cases recorded separately; not an exhaustive coverage score
Remaining to beta Unestimated Requires acceptance tasks, performance budgets and a decision on required format/package/UI migrations
Remaining to full scoped parity Unestimated Includes possible .tcraft/output reworks, upstream .mcmat, ArtCraft UI/shared-library adoption, .spp feasibility and workflow/fidelity work

The former ~94% figure mostly counted implementation paths. This review also asks whether an artist gets equivalent controls, predictable output, responsive interaction and dependable delivery. The change is a measurement rebaseline and an expanded import goal, not a measured loss of functionality. Percentages are rounded judgments; ranges express uncertainty, not statistical confidence intervals. The old “~6 h + baking TBD” remainder is withdrawn: this review provides no calibrated engineering or agent-hour forecast.

The percentages describe behavior at 62b2b13, not how much of a settled delivery plan is finished. Potential rework can revisit working code and require repeating the checks below. Its scope and cost are not yet measured, so remaining effort cannot be calculated as 100 minus the readiness score. Incoming upstream work is not counted as landed functionality.

Readiness by audience

Audience Ready % Hours to ~95% Work that dominates
Full scoped target ~60% Unestimated Workflow depth, format/output-package decisions, .mcmat, ArtCraft UI/shared libraries, bake/display fidelity, color management, read-only .spp and platform validation
Mainstream practitioner ~60% Unestimated Reliable masks/painting, MaterialCraft exchange once its contract is ready, project/package compatibility, larger projects and destination-app checks
Essentials user ~75% Unestimated Discoverable controls, predictable startup/GPU selection and save/reopen reliability retained through format/UI migrations

Why alpha. Mouse/keyboard tasks verified 3D/2D painting, undo/redo, material drag and drop, mask and generator editing, UV rectangle fill, UI baking, native saving and texture export. Separate command checks covered high-to-low transfer, UDIM sizes/packing and scene/material exports. The interface is recognizably close in its painting, bake and export workspaces, but still lacks mask-thumbnail effect menus, several fill controls and a MaterialX export option. Generator depth is limited, bake equivalence is open, and hardware/performance coverage is incomplete. There were 96 open issues on the review date; these include requests and older reports, not 96 reproduced defects. The published v0.6.1 release already exists; “first release v0.1.0” is no longer an upcoming milestone.

Native/output contracts and the future MaterialCraft, ArtCraft UI and shared-library integrations also remain unsettled. Passing a legacy reader or a standalone export check does not complete those future workflows.

A proposed beta gate is ~75% readiness plus repeatable artist tasks on representative assets: high/low baking and smart masks, precise 3D/2D painting, a multi-set UDIM project, save/reopen without lost work, and exports checked in destination tools. Startup failures and data-loss blockers on the supported test matrix must be resolved. This audit does not establish that gate. Beta scope must identify which reworks/integrations it requires and which remain later goals. Every adopted project/package version needs a compatibility policy and regression checks; MaterialCraft exchange must be validated against a named producer/consumer version pair before it is advertised as ready.

By dimension

All percentages below are estimates. The weights sum to 100; their additive result is 59.1%, rounded to ~60%. Hardware, performance and stability scores include confidence discounts for incomplete validation; they are not measured success rates or speed comparisons.

Dimension % ready Weight Hours to full Evidence / remaining work
Features (depth) ~65% 30 TBD Core editing tasks and high/low/UDIM outputs exercised; generators, properties and specialized workflows need depth. Mask anchors landed after this assessment; live UI validation remains
UI/UX fidelity ~65% 20 TBD Current editing/bake/export UI checked; gaps, ArtCraft UI adoption and revalidation after refactoring remain
File formats / interchange ~60% 10 TBD Current native save and basic exports exercised; possible .tcraft/output-package reworks, unfinished upstream .mcmat, .spp and consumer validation remain
Hardware (GPU, pen, displays) ~50% 8 TBD One Windows GPU exercised; pen tilt, pressure across platforms and display coverage remain incomplete
Localization ~35% 4 TBD Ten picker languages in source; optional shared catalogs/fonts and native-language review need current measurement
Performance ~40% 10 TBD Dirty-tile/budgeted architecture and benchmark code exist; no matched performance budgets or large-project results from this review
Stability ~55% 10 TBD Small save/reopen and undo/redo passed; recovery, large projects and long sessions need coverage, including after format/library migrations
Platforms ~65% 3 TBD Desktop/web build and packaging paths exist; this review ran Windows only
Automation / ecosystem ~75% 5 TBD 178 commands and CLI/control/MCP architecture; shared-library adoption, command/UI compatibility and the future MaterialCraft contract need integration checks

Features

These are approximate depth/readiness estimates, not percentages of matching pixels or completed menu entries. Feature rows overlap and are not an additional additive score.

Area % ready Hours What works / main gaps
Projects and files ~80% TBD Current .tcraft new/save/reopen passed with layers, effects and baked maps retained. Schema/container rework remains possible; versioning/migration policy and renewed persistence/recovery tests are needed if adopted. Large cases, Alembic HDF5 and animation remain open; PBR folder/ZIP/selected-file import landed after this assessment: file.importPbrFolder creates one fill layer from named maps, including packed ORM and OpenGL/DirectX normals, with a UI dialog and optional native library copy; it is not credited in this score
Texture sets and UDIM ~80% TBD Two tiles exported at independent 128/256 px sizes with expected packed-channel values; 21 channel kinds in source. Many-set/many-tile production coverage remains untested
Layers, masks and effects ~75% TBD UI fill/black-mask creation, generator editing/duplication and native persistence passed at the assessed revision. Mask-thumbnail context actions remain missing; effect renaming and mask-stack anchors landed afterward, with live UI validation still open
Painting and precise 2D work ~70% TBD Mouse painting in both views, keyboard undo/redo, view shortcuts, brush gesture and UV-island rectangle fill passed. Nine tool kinds exist; brush feel, pressure/tilt, other tool task suites and complete trim sheets need validation
Generators and smart masks ~45% TBD Seven generator kinds and 24 smart-mask presets; Position/global-balance editing worked in the UI. Shared adjustments, resolution and input slots exist; richer generators, micro-detail workflows and saving user smart masks remain incomplete
Materials and asset library ~60% TBD 72 procedural materials and 21 smart materials; shelf search/drop worked. Legacy MaterialCraft paths exist, but the planned .mcmat contract is unfinished upstream. Library variety, previews and cross-app integration remain work
MaterialCraft .mcmat integration Incomplete; upstream-dependent TBD after contract Current reader accepts legacy v1/v2. The proposed MaterialX-based v3 still needs upstream completion, shared-library support, TextureCraft mapping/UI and tests with real producer outputs; this is not a completed material exchange workflow
Baking ~60% TBD All ten maps completed through the UI on CPU at 256 px and through commands on GPU at 128 px. Separate high/low normal/height transfer produced varying maps; cross-app pixel equivalence, larger assets and resolution-stable wear remain unproven
Viewport, shading and rendering ~70% TBD PBR/IBL, extra shading lobes, 3D/2D views, displacement and CPU/GPU path-tracing paths. Matching material/display behavior, larger scenes and hardware ray tracing remain open
Color management ~35% TBD Linear/sRGB handling, fixed export spaces and profile metadata; display model has Linear/ACES-filmic tone mapping. Configurable OCIO project/display workflows are absent
Texture and scene export ~75% TBD Current PNG/UDIM/USD/GLB/MaterialX checks passed within scope. Output-package layout/content may be reworked; package contracts and consumers need separate validation. MaterialX UI output, complex materials and larger UDIM assets remain incomplete
Read-only .spp compatibility 0% TBD after feasibility study Planned; no reader, version matrix or import-fidelity corpus. .tcraft remains the save format
Workspace, panels and preferences ~65% TBD Current views/layouts checked; narrow labels truncate. Incoming ArtCraft UI refactoring/shared components can change these surfaces; task, shortcut, accessibility and layout/preferences persistence checks must be repeated
Automation ~80% TBD Command inspection and scripted editing/export work; UI, CLI and MCP surfaces share engine commands. This review exercised the control channel and CLI, not every MCP/tool path

Source checks matter here: generator kinds number seven even though their shared controls have grown; polygon fill already includes triangle, polygon, UV-island and mesh modes. Current MaterialCraft import accepts v2 and recognizes the newer source archive layout. That is legacy compatibility, not completion of the next .mcmat integration. The upstream v3 proposal describes MaterialX plus resources in equivalent folder/ZIP deliveries and excludes native source. Its proposed exposed controls differ from re-cooking an embedded authoring graph. The final supported subset, version policy and authoring/re-cook workflow must be agreed and tested; the legacy promise must not be carried forward automatically.

The UI gap inventory records reproduction routes and acceptance checks: mask context/save/export actions; independent U/V tiling, filtering/wrap, physical-size and warp/per-tile projections; MaterialX UI output; and configurable color-management settings. Generator depth and narrow-layout readability are tracked as partial areas. Direct hardness/angle controls and polygon/UV rectangle fill already exist; older reports must not turn them into fresh missing-feature claims.

Evidence from this review

Check Result and limit
Baseline/build Fetched origin/main; HEAD matched 62b2b13. cargo build -p texturecraft -p texturecraft-cli --profile verify passed; binary reported 0.6.1 and that revision
Reference UI/functionality On an original UV cube, pointer painting, paint/fill/black-mask creation and an empty generator slot worked; seven-map self-bake at 1024 px completed. Inspected fill/projection/scale menus, mask menu and export dialog. The “Read only” title did not prevent these edits; reference save/export and output equivalence remain unchecked
TextureCraft UI tasks Mouse/keyboard new project, 3D/2D strokes, Ctrl+Z/Y, F1/F2/F3, brush gesture, shelf search/drop, black mask, Position generator/global balance and effect duplication worked. UV rectangle fill wrote 88 tiles per channel and undo removed it
Persistence Native UI save of the deeper four-layer case followed by command reopen preserved layers, effects, 21 painted tiles per channel and ten baked maps. Earlier command save/reopen also passed. UI-only open-dialog completion was not established
Bake GPU command check: ten maps at 128 px / 16 rays. UI check: ten maps at 256 px, logged CPU. Separate CPU high/low check: normal and height varied on a two-triangle low plane / 1,152-triangle analytic high surface. No matched reference quality or timing comparison
UI texture export Settings/template/list tabs inspected and Export clicked; eight 1024 px PNGs decoded successfully, and base color was inspected visually. All eight matched command export byte for byte from the same project/options. Creating/editing templates was not exercised
UDIM and channel packing Two tiles exported at 128 and 256 px. Both ORM maps sampled RGB (255,153,51), matching AO 1, roughness 0.6 and metallic 0.2; this is a small command-only case
Interchange Five 256 px OpenPBR PNGs plus parsed MaterialX XML; textured GLB and USDA reopened through TextureCraft's CLI with the original 12 triangles, 24 vertices and material. No external renderer/material equivalence checked
Layout Whole-window PNGs inspected at 1920 × 1032 and 1280 × 800. Narrow layout retained both views and toolbar scroll controls, with severe layer-label truncation; no full narrow-window task suite or pixel-by-pixel UI score
Rework/upstream scope Possible .tcraft/output-package reworks and incoming ArtCraft UI/shared-library work are planning inputs, not completed changes. The .mcmat v3 proposal and current legacy reader were reviewed; no future contract or current MaterialCraft producer/consumer pair was validated
Not run Full Rust tests, clippy, wasm, other operating systems, pen/display hardware matrix, large UDIM projects, long sessions, matched performance benchmarks and reference image-difference corpus

Languages

The picker contains ten languages. Shared catalogs are optional at build time; missing entries fall back to English, and builds without catalogs offer English only. The older coverage table predates catalog migration, so its “100%” rows are not a current measurement. This audit ran an English build without external catalogs.

Language Code UI strings Status
English en Source language Exercised in this audit
Simplified Chinese zh-Hans Not remeasured Catalog/font dependent; native review pending
Spanish es Not remeasured Catalog dependent; native review pending
Hindi hi No picker entry Not offered
Arabic ar Not remeasured Catalog/font dependent; shaping/layout needs current review
French fr Not remeasured Catalog dependent; native review pending
Portuguese (Brazil) pt-BR Not remeasured Catalog dependent; native review pending
Indonesian id No picker entry Not offered
Japanese ja Not remeasured Catalog/font dependent; native review pending
German de Not remeasured Catalog dependent; native review pending
Korean ko Not remeasured Catalog/font dependent; native review pending
Vietnamese vi No picker entry Not offered

Also in the picker: Italian (it), with catalog coverage and native review not remeasured. Engine error details remain English; UI-key coverage alone would not establish full localization.

Upcoming

These are acceptance-oriented work areas, not claims that their linked reports were reproduced here. Open pull requests are not credited until merged into the assessed main revision.

Rank Milestone Estimate
1 Reliable startup and project safety across the supported GPU/platform matrix; reproduce startup/GPU reports and exercise recovery/large saves (#541, #489) TBD after reproduction
2 Define .tcraft and output-package direction, scope any reworks, and establish versioning, migration and producer/consumer acceptance cases Unestimated; design scope open
3 Plan/adopt the incoming ArtCraft UI refactor and shared-library changes; preserve command reachability and repeat UI/persistence/platform checks after adoption TBD after upstream contracts and consumer scope
4 Bring in the completed MaterialCraft .mcmat contract: agree the supported MaterialX/resources/controls subset, implement the consumer through shared APIs and validate folder/archive outputs from named versions Upstream-dependent; format unfinished
5 Bake fidelity and dependable smart-material wear: matched assets/settings at multiple resolutions, all ten maps, high/low pairs and documented differences (#536, #538) TBD after map corpus
6 Generator, anchor and property depth; finish the artist-facing control inventory while preserving existing shared controls (#374, #389, #433) TBD after control audit
7 Brush/pen precision, feedback and discoverability; complete 3D/2D/trim-sheet tasks and set responsiveness budgets (#545, #485, #306) TBD after task tests
8 Color-managed pipeline and destination checks for USD, glTF/GLB, MaterialX and texture maps, including UDIM, complex materials and the adopted output packages TBD after consumer matrix
9 Read-only .spp discovery: clean-room feasibility, recoverable data and unsupported-content reporting; convert into the adopted native format without changing the source Unestimated; feasibility unknown
10 Release/platform regression checks, web polish, current translation coverage and native review; expand the original asset library TBD after validation

Dependencies and possible rework

These tracks add discovery, migration, integration and repeated validation work. A possible rework is not an approved replacement design, and an upstream proposal/merge is not proof of TextureCraft adoption.

Track Current state Completion / revalidation gate
Native .tcraft Current save/load works in the audit; schema/container may change Decide the native contract and compatibility policy. For any adopted rework, check existing projects and conversion explicitly: layers, masks, effects, resources, painted tiles, UDIM layouts and baked data; retain bounded reads, atomic saves and recovery behavior
Output packages Current individual exports work within the recorded scope; future package format/content/layout is not settled Define affected artifacts, manifests, naming, resource references and versioning. Test package relocation, folder/archive delivery where applicable, color/normal/displacement metadata and real consumers; repeat install/update checks if distribution artifacts change
MaterialCraft .mcmat Legacy v1/v2 support exists; v3 is unfinished upstream Agree the published contract and supported MaterialX subset, folder/ZIP equivalence, resources and exposed inputs. Adopt it through shared libraries and test actual MaterialCraft output in TextureCraft, including unsupported versions/features. Do not assume native-source recovery or the legacy re-cook behavior
ArtCraft UI refactor Incoming suite/UI work; shared palette/theme proposals such as craft-libs #102 and #397 are examples, not the whole refactor or an adoption claim Scope TextureCraft's adapters/components and design changes. Keep document behavior command-driven; repeat the UI task audit, input/accessibility checks, screenshots and layout/preferences persistence tests against the adopted design
Shared libraries Several shared crates are already pinned in Cargo.toml; format migration work continues Use reviewed public APIs and keep app-specific mapping/policy local. Each adoption needs consumer integration plus compatibility, security-limit, layering, wasm and regression checks; a library change alone does not complete the app's work

Current focus

Keep the current functional baseline while establishing project-safety and bake-fidelity checks. Scope .tcraft/output-package changes and coordinate ArtCraft UI/shared-library adoption alongside that work, so feature work budgets include possible migration and revalidation. Bring .mcmat forward against the finished upstream contract rather than treating the legacy reader as completion. Run .spp feasibility separately. Hardware ray tracing is a later enhancement; beta depends on the scoped acceptance/integration gates above. Release validation continues for existing releases; there is no outstanding “first release” gate.

Progress log

Date Entry
2026-11-16 Added possible native/output-package reworks, unfinished MaterialCraft .mcmat integration and incoming ArtCraft UI/shared-library work as explicit roadmap tracks. Current audit scores stay tied to the tested revision; migration scope and effort remain unestimated
2026-11-16 Deeper UI/functionality audit added in docs/ui-parity.md: twenty check groups, specific missing controls, reference editing/self-bake, UI export/save, analytic high/low transfer and UDIM packing. Overall percentages remain provisional estimates
2026-11-16 Approximate live review against main 62b2b13 and reference 12.2.0. Basic edit/save/reopen, GPU bake and interchange smoke checks passed. Rebaselined reporting to ~80% breadth / ~60% readiness and recorded validation limits
2026-11-16 Scoped reference-project compatibility to future read-only .spp; retained .tcraft for editing/saving and open pipeline exports. No importer or other implementation change in this review
2026-11-12 v0.6.1 published; supersedes the former first-release v0.1.0 focus
2026-11-04 Previous roadmap reported ~94% weighted implementation/workflow completion; retained here as historical context, not a comparable readiness measurement

Revision history

Date Change Summary
2026-11-16 major Added dependency/rework scope and adoption gates; distinguished legacy .mcmat support from the unfinished MaterialX-based proposal; expanded upcoming work without crediting unlanded integrations
2026-11-16 major Expanded live UI tasks and output validation; added a separate UI gap inventory with acceptance checks; distinguished UI, mixed and command evidence and CPU/GPU bake paths
2026-11-16 major Adopted headline/audience/dimension/feature/language/upcoming tables; added main-revision evidence and uncertainty, read-only .spp scope, current release status and acceptance gates; removed the unsupported six-hour remainder