The scan/opening/status/error lines, offline mode and its retry, pinning a
collection offline, the sync and thumbnail-sweep state, and the callbacks
that route the grid to a rescan, a sync, another library, a panorama merge
or an export — the last of what AppWindow still carried under the
library- prefix — move onto the `Library` global started earlier on this
branch. Rust reaches them through window.global::<Library>() rather than
window.set_/get_/on_/invoke_ on the root.
library-visible is the one name that stays: it is computed from
active-page and active-view, the shell's own routing state, which a
global cannot read. AppWindow now declares no other library- property or
callback.
The date-range fields and their band on the capture-time axis, the
timeline's bars, labels, scrub/pinch/pan/zoom callbacks and its
sweep/current-bucket/anchored state, and the filter bar's people chips,
mode, eyes-open toggle and gesture reference move from AppWindow onto the
`Library` global. Rust reaches them through window.global::<Library>()
rather than window.set_/get_/on_ on the root, as the earlier commits on
this branch did for the grid's cells, selection, ratings and keywords.
The keywording sheet's rows and its open/assign/unassign callbacks, the
star and flag callbacks a cell click or a judgement key fires, the burst
toggle and representative-chosen callbacks, the trash-selection shortcut,
and the rating/unjudged/flag filter chips with their rating-counts model
move from AppWindow onto the `Library` global started in the previous
commit. Rust reaches them through window.global::<Library>() rather than
window.set_/get_/on_ on the root.
AppWindow carried the grid's loaded window of cells, the keyboard cursor,
drag and drop, the held-row long-press state, columns and cell size, the
scroll and viewport bookkeeping, the photo roll's pick and centre-request,
and the local-only/reorder/collection-filing gestures that act on a
selection, as properties and callbacks on the root component. That state now
lives in the `Library` global declared in library.slint, next to the structs
(LibraryCell, TimelineBar, KeywordRow, PersonChip) it and the grid's other
components already share; Rust reaches it through
window.global::<Library>() instead of window.set_/get_/on_/invoke_ on the
root, the same change collections.slint's `Collections` global made for the
sidebar.
library-visible stays on AppWindow: it is computed from active-page and
active-view, the shell's own routing state, which a global cannot read.
Everything else still prefixed library- — the timeline, the filter bar,
ratings and flags, keywording, and the routes and status lines — stays on
the window for now and moves in the commits that follow.
controller holds LibraryController and the window-sizing constants every
other module reads and writes through pub(super) fields, the same shape
collections_ui and develop already use. open is the launch-to-scan cycle and
the worker that checks the catalog file before either touches it. offline is
what of a collection is on this device and the prompt that offers to change
it. window fills the grid model from the catalog and drains the thumbnail
fetch, which is the piece the catalog-reads-are-proportional-to-what-changed
rule (docs/catalog.md §1) bears on most directly. sync is the background
passes that reach beyond the loaded window: the metadata sweep, the
whole-library thumbnail pass, and the exchange with the server.
ratings_keywords applies a judgement or a keyword to a selection and queues
the sidecar and XMP writes behind it. timeline is the capture-time sidebar
and the photographer's place together, kept in one file because a restored
place ends by moving the timeline marker and a scrub is a restore of one
instant, so most calls between the two would otherwise cross a module
boundary. grid wires the grid's own callbacks — the keyboard cursor,
cell-size zoom, the routes into and out of develop — and filter_bar wires
the rating, people, date and offline-scope filters, calling back into
whichever of the above owns the work a filter change triggers.
Extracted by item rather than by line range, so every doc comment and
TRACES/GESTURE annotation stayed attached to the code it describes; the
sorted set of TRACES/GESTURE lines in the new directory is identical to the
original file's. Tests moved with the code they exercise, including the
handful of fixtures — settle, model_of, with_catalog, zoom_cell, pinch_step
— that only one target module needed and so were not worth sharing through
a test_support module the way the other splits use one. Items that crossed
a new module boundary were widened from private to pub(super), narrower
than the whole-file access the original gave them; a few items already
pub(crate) for recovery_ui or presets stayed there rather than being
narrowed, since nothing needed them tightened further.
mod.rs re-exports the same surface library_ui:: callers used before, so
lib.rs and every other caller needed no change.
AppWindow carried the sidebar's tree, its row menu, renaming, drag and
drop between rows, the trash row, and the membership sheet as ~40
properties and callbacks on the root component, in the pattern CH-1
describes and the develop screen's globals (Adjustments, Framing, Steps,
...) already replaced. Collections.* in collections.slint now holds that
state, declared next to the MembershipRow struct it and the membership
sheet both use; Rust reaches it through window.global::<Collections>()
instead of window.set_/get_/on_/invoke_ on the root.
collection-selected, collection-select, and collection-offline-menu stay
on AppWindow: library_ui.rs invokes collection-select directly and
registers collection-offline-menu's handler, and lib.rs reads
collection-selected for back-navigation, so moving them would have meant
editing library_ui.rs, which another change on this branch is splitting
into a module directory. collections-visible stays too — it is
lib.rs's panel-layout state, seeded from the saved layout before the
sidebar exists, and collections_ui never touches it. Everything prefixed
library- (the grid's drag, selection and keyword state that the sidebar's
Rust also wires for cross-feature gestures like filing a selection into
a collection) stays on the window as well, since it belongs to the
library screen, not the sidebar.
app.slint carried show-launch, show-library, show-identity, show-settings,
show-import and show-merge as separate booleans, so the root component chose
what to draw with five- and six-term conjunctions and nothing stopped two of
them being true at once. Replaced with two enums: View { develop, library,
identity, launch } for which top-level screen is showing, and Page { none,
settings, import, merge } for which page, if any, is drawn over it.
Two values rather than one, because the two questions are genuinely
different. Settings, Import and Merge are reachable from more than one View
and are drawn outermost without touching it — closing one has to return to
whichever View was already current, and today that works because the
underlying property is left alone while the page sits over it. A single
View with five or more variants would need a second field remembering what
to return to; Page needs nothing to remember, since View was never
overwritten in the first place. Identity, by contrast, genuinely replaces
the window the way Launch and Library do (see the existing "like the launch
screen" comment on its `if`), so it is a View variant, not a Page.
Every `if` chain in app.slint that used to compare four, five or six
booleans now compares active-view and active-page to at most one variant
each. library-visible collapsed from a six-term conjunction to
`active-page == Page.none && active-view == View.library`.
The Rust side follows: every set_show_*/get_show_* call in library_ui.rs,
identity_ui.rs, settings_ui.rs, merge_ui.rs, import_ui.rs, launch_ui.rs and
lib.rs now reads or writes active-view or active-page instead, including
lib.rs's startup match (View.launch vs View.develop, since a Startup that
skips the launch screen used to leave both old booleans false and fall
through the chain to develop) and identity_ui's close handler, which now
writes View.library or View.develop in one call where it used to write
show-library then show-identity separately.
back_one_step needed one deliberate adjustment beyond the mechanical
rename. Identity was never represented in NavState: back had nothing to do
when Identity was opened from the library (show-library stayed true,
unread by IdentityScreen's own condition) and could only reach ToLibrary
when opened from develop, which likewise wrote a property IdentityScreen
never read — so escaping out of Identity was invisible in both cases before
this change. With a single active-view, falling into the general case
would instead overwrite the value IdentityScreen's `if` does read and close
it as an unintended side effect. back_one_step now swallows the gesture
while View.identity is current, reproducing the same "nothing visible
happens" outcome for both origins without threading identity_ui's private
came-from-library state through lib.rs for one screen.
Verified with tools/manual/drive.py against a private Xvfb and the debug
build: launch screen to library, Settings opened and closed, Identity
opened and closed (including Escape doing nothing while it is open),
develop opened from a cell and closed both by the back button and by
Escape. Screenshots under verify/.
collections_ui.rs had grown to 4,591 lines covering the sidebar controller,
the click/drag selection policy, tree refresh, the drag gesture, the trash
worker, twelve wiring functions, and the row's rename/create/context menu,
all in one file. Split into collections_ui/ with one module per area, the
way develop/ and library/ were already split on this branch:
- controller.rs: CollectionsController and the pure drop/delete/release
decisions (decide_drop, decide_delete, decide_release, menu_detail,
delete_warning) that a test can drive without a window.
- press.rs: PressUndo and the click-and-release selection policy
(apply_press, select_row, commit_press, cancel_press).
- tree_sync.rs: rebuilding the sidebar from the catalog and pushing
catalog-derived state into the grid (refresh_tree, offline_state,
sync_lifted/sync_selection/sync_reorderable/sync_badges,
refresh_membership, direct_holdings).
- drag.rs: the cursor bitmap (compose_drag_image, blit_scaled) and the
hold/spring timers (arm_hold, arm_spring, should_spring,
collapse_spring_opened) plus their delay constants.
- trash.rs: the soft delete (start_trash, start_restore, drain_trash,
stop_trash, refresh_trash, format_bytes).
- wiring_grid.rs / wiring_tree.rs: the wire() entry point and its twelve
wire_* functions, split in two because together they were the largest
single piece (grid-facing selection/drag/trash vs. sidebar-facing
navigation/create/rename/row-drag/menu/membership).
- rename_menu.rs: creating, naming and renaming collections, and the row's
context menu (apply_rename, create_child, unique_name, open_row_menu,
close_row_menu, close_rename).
mod.rs carries the module's own top-level doc comment, the `pub use`
re-exports for the eight items the rest of the crate reaches by
`collections_ui::` path (CollectionsController, wire, refresh_tree,
sync_badges, sync_selection, select_row, commit_press, cancel_press), and a
shared `test_support` for the one fixture (`ids`) more than one file's
tests needed. Every item that only crossed a boundary within this module,
not out of it, was narrowed to `pub(super)` rather than kept at the
crate-wide `pub` a single file gave it for free.
Extracted with a brace-aware pass that kept each item's own leading doc
comment and attributes attached to it, and tests moved with the code they
exercise; every TRACES/GESTURE comment lands on the same code it did
before. No file outside the new directory changed — lib.rs's `mod
collections_ui;` resolves to the directory automatically, and every
outside caller's `collections_ui::` path still resolves through mod.rs's
re-exports.
The 188-line wire() registered the import page's callbacks in four
comment-delimited sections. Lift each into its own fn: wire_opening_and_closing,
wire_choosing_a_source, wire_options, wire_running. context stays generic
over C on each of the three functions that use it, matching how survey()
and start() already take it (impl Fn, implicitly Sized) rather than coercing
it to a trait object, which would have needed ?Sized added to those two
unrelated functions for no benefit.
run() was 2,724 lines that built every controller, owned the develop
session and its render loop, and registered every develop-screen callback
inline — the state CH-1 in docs/dev/code-health.md describes. This is the
mechanical split CH-1 calls for, done in one pass rather than section by
section since develop_ui.rs only compiles once lib.rs stops registering
those callbacks itself.
develop_ui.rs is new: a DevelopWiring struct holding the session, the rows
model, the redraw/render closures and the other controllers' handles, and
one wire_* function per section run() used to contain — presets, export,
the adjustment panel, film, undo/redo, zoom/pan/crop, rotation/flips/
straightening, navigation and peaking — called in the order run()
registered them. window travels as its own parameter throughout rather
than living on the struct, because the generated AppWindow type is not
Clone; every other field is an owned clone so each function's body could
be pasted from run() unchanged.
lib.rs::run is now under 300 lines: construction and startup only, calling
named functions for the window and its diagnostics/inference wiring, the
launch/library/collections/identity/settings screens, import and merge,
the render loop (render_now/redraw/show), the remote open path, the
window chrome (resize, layout class, panel toggles, back gesture), and
develop_ui::wire for the rest. Every TRACES and GESTURE comment moved with
the code it annotates.
Two latent type errors surfaced while restructuring rather than being
introduced by it: activity and display were being passed around as bare
ActivityLog/DisplayWatch instead of the Rc<...> their constructors
actually return, which only worked before because nothing needed to name
the type explicitly.
The 225-line wire() registered the launch screen's callbacks in nine
comment-delimited sections. Lift them into fn's, merging a few adjacent
ones that were only a handful of lines each: wire_sign_in covers both the
browser flow and the app-password fallback (the same button in two forms),
wire_choose_folder_and_open covers opening the folder browser and the
final "Open library" press, since both are short and sit back to back.
wire_use_folder, wire_sign_out, wire_formats, wire_folder_picker_navigation
and wire_copy_url stay as they were sectioned. wire() calls each in the
original order and keeps the closing render() call, which every section
relies on having run once at startup.
The 312-line wire() registered the panorama page's callbacks in three
comment-delimited sections. Lift each into its own fn: wire_start (the
"Merge to panorama" press, its fetch, and the DARKROOM_START_MERGE dev
entry point — all three only ever used together), wire_decision (confirm,
projection and border chips, the fill knobs), and wire_stop_and_leave
(abandon, close). wire() itself now just calls the three in order; S, C
and F stay generic on wire_start since sources/context/on_done are used
nowhere else.
The 837-line wire() had almost no section comments, unlike its siblings, so
the seams had to be found by reading it rather than following markers. Lift
each into its own fn: wire_dials (the two grouping sliders), wire_navigation
(open/close/switch person — needs models() too, for the missing-model
banner), wire_rename_and_merge (a rename and the namesake offer it can
raise), wire_face_actions (pick/confirm/reject/split, the grid's own
actions), wire_grouping_preview, wire_recluster, wire_indexing (the shared
launcher behind Index/Re-index plus Stop), and wire_coverage_and_ignore.
wire() keeps the generic-to-trait-object coercions and the eyes_available
closure, since most of the above need it, and calls each function in the
original order. The reload! macro moved from inside wire() to module scope,
dedented, since macro_rules is scoped textually and every extracted function
uses it.
The 1,457-line wire() registered every collections-sidebar and grid-drag
callback in one function, sectioned only by comment. Lift each section
into its own fn: wire_selection (split further into wire_selection and
wire_selection_filing, since the original section ran to 375 lines),
wire_drag, wire_trash, wire_trash_from_grid, wire_tree_navigation,
wire_create, wire_rename, wire_remove, wire_row_drag (the tree row's own
hold-drag, which the original "remove" comment's span covered but which
is really a separate feature), wire_row_menu, and wire_membership.
wire() itself now just coerces the shared closures to trait objects and
calls each in the original order. visible_ids is coerced to
Rc<dyn Fn() -> Vec<ImageId>> at the top, alongside on_scope_changed and
session, so the new functions take plain trait objects instead of
threading a generic parameter through every one of them.
library.rs was 7,729 lines wiring together everything "open a remote
library" touches: scanning, pulling other devices' judgements out of
sidecars found along the way, writing local edits back out to the
sidecar outbox, pushing/reloading XMP by hand, fetching and prefetching
thumbnails and originals, generating thumbnails locally, the metadata
and thumbnail background sweeps, on-disk paths for the catalog and
model files, and reading the grid's cells, spans and rating filter.
Same motivation as the develop.rs split (docs/dev/code-health.md CH-1):
a pure, no-behaviour-change move into one file per area, each under
about 1,500 lines.
Tracing actual call sites rather than trusting the file's physical
layout mattered here: `persist`, `load_folder_etags`, `pull_sidecars`,
`load_sidecar_etags`, `record_sidecar_read` and `apply_judgement` sit
textually beside the XMP push/reload functions but are called only
from `run_scan` (pulling a device's own past judgements out of the
sidecars a scan just walked), so they went to scan.rs and not xmp.rs.
`cells` came out at over 1,800 lines once its tests moved with it and
split further into cells.rs (windowed reads, trash, ordinals) and
spans.rs (collection scope, manual reordering, the capture-time
histogram) -- ten submodules rather than the nine first planned.
Previously-private items reached from a sibling module became
`pub(super)`, narrower than the whole-crate reachability one file gave
them. Tests moved with the code they test; the two test fixtures used
across more than one file (`scanned`, and develop.rs's
`session_with_a_left_half_subject` in the matching commit) joined the
shared `test_support` module alongside the existing `entry`/
`with_images`/`image_ids` helpers. `mod.rs` re-exports every module's
public items under `library::`, including the `pub(crate)`
`test_support` module `repairs.rs` reads its fixtures from, so no file
outside `library` needed a change.
The previous commit split develop.rs the same way; taken alone it left
dr-ui without library.rs, so that intermediate commit does not build on
its own. This one restores it.
develop.rs had grown to 9,327 lines covering everything the develop
session does: opening a photograph, the parameter-row and curve-widget
panel model, mask viewing and editing, mask creation and the rasteriser
that turns a mask stack into GPU arrays, spot repairs, scene
segmentation, framing and zoom, white-balance sampling, rendering and
film choice, and the undo/snapshot history. docs/dev/code-health.md
CH-1 names dr-ui's lack of a view layer as the reason every feature
kept landing in a handful of files; this is the first of the two pure
splits it recommends as easy, no-behaviour-change wins independent of
that larger rework.
The boundaries follow the file's own sections (several were already
marked off with comment headers) and the seams a full read turned up
underneath them -- mask storage/rasterisation turned out to be a
distinct concern from mask viewing and editing, and rows/tabs/curves
from each other, so those split further than the headers alone
suggested. Each module stays under about 1,500 lines. Struct fields
and the handful of helper methods now called from a sibling module
became `pub(super)`, which is strictly narrower than the whole-crate
reachability a single file gave them; nothing gained visibility outside
`develop`. Tests moved with the code they test, including the few
cases where a helper one file's tests needed was itself only defined
in another's -- those became shared fixtures in `mod.rs` alongside the
`headless`/`read_back`/`grey_session` helpers that already worked that
way. `mod.rs` re-exports every item `develop::` callers outside this
module used before, so lib.rs, masks_ui.rs and the rest needed no
changes.
wire() registered every grid callback in one 1,214-line function behind
four section comments, two of which were themselves far over 300 lines
with no further markers. Each fenced section becomes its own function,
called from wire() in the original order with the section's own comment
kept as its doc comment:
- "the keyboard cursor (FR-CULL-4)" (496 lines) splits at its own topic
breaks into wire_grid_cursor_and_zoom (cursor movement, cell zoom and
pinch), wire_grid_sync_and_load (explicit sync/thumbnail requests and
the reloads a changed viewport, column count or scroll position
trigger), wire_timeline (the capture-time sidebar) and wire_grid_routes
(grid/launch/develop navigation and a manual rescan).
- "ratings and flags" and "keywords" were already under 300 lines and
become one function each.
- "the filter bar" (447 lines) splits into wire_filter_ratings_and_people
(which keeps the section's own comment), wire_filter_dates and
wire_filter_scope_and_offline.
The three callbacks registered before the first section comment
(on_settings_xmp_reload, on_library_cell_clicked, on_library_roll_pick)
and the trailing crate::recovery_ui::wire call stay directly in wire(),
since neither is inside a fenced section.
wire() registered every mask-panel callback in one 778-line function
behind section comments. Each of the seven fenced sections (computing
the region map, refining a subject's mask, dragging a gradient,
selecting on the photograph, the stack, the edge treatment, adding
layers) becomes its own function, called from wire() in the original
order with the section's own comment kept as its doc comment. The
"adding layers" section was itself over 300 lines and had no further
section markers inside it, so it is split at its own natural seam
between painting/viewing a mask (wire_layers_paint) and working the
parts and add-mask buttons (wire_layers_parts); the second half gets an
introductory doc line since there was no comment of its own to reuse.
Locals declared just for one section's closures (running, refining,
dragging) move into that section's function instead of staying in
wire().
wire() registered every settings-page callback in one 416-line function,
fenced only by section comments. Each fenced section (opening and
closing, cache, faces, export, reset) is now its own private function
that wire() calls in the same order, with the section's own comment kept
as its doc comment. on_budget_changed and on_open are coerced to trait
objects at the top of wire() so the new functions take a plain
Rc<dyn Fn> rather than needing their own generic parameter, with no
change in the closures registered or the order they are registered in.
docs/ had 26 developer documents flat beside the manual, and the two
audiences are very differently sized: most readers want the manual and
the gesture reference, a few want the register, the designs and the
measurements. The manual and gestures.md stay at the top; everything for
someone changing the code moves to docs/dev/, and the two documents that
name their own successors — the v0.1 milestone and the UI-refinement plan
— go to docs/dev/archive/ rather than being deleted, since both are still
cited. docs/README.md is the index, users first.
Every reference follows: code comments, Cargo manifests, the workflows,
the pre-commit hook, the bench and traceability tools (which locate the
repo root by docs/dev/requirements.md now), packaging, the Docker READMEs,
CLAUDE.md, CONTRIBUTING.md and the README. The matrix links one level
deeper and is regenerated. Links out of the moved documents into the tree
gain a level; a link checker over every Markdown file finds none broken.
The recording sampled a red brick wall and moved the sliders by three
units, which at GIF size is a click that does nothing. The scene now
drags the frame cold first and picks a white air conditioner, so the
correction is visible and the picker's being absolute - set from the
photograph, not from where the sliders were - is what the picture
shows. The text says so, and says a blown highlight is refused.
The probe's comment said a 192px render "averages a small neighbourhood
into each of its pixels". It does not: the composed shader fetches the
source at one position per output pixel - nearest for an unrotated
frame, four photosites blended otherwise - so the probe was a point
sample of a noisy sensor, and two painted-white air conditioners on the
same wall answered +37 and -50.
The tap is now narrowed to the patch of the canvas around the click, a
couple of percent of its width and square on screen, and rendered at
64x64 with interpolation forced on, which puts a sample on every sensor
pixel under it at any ordinary zoom. The samples are averaged, with the
void and clipped ones left out rather than allowed to pull the mean, and
fewer than half surviving is refused. compose_camera_probe takes the
patch; the merge's compose_camera_linear keeps its nearest sampling. The
readback shrinks from six megabytes to sixty-four kilobytes.
A frame of alternating warm and cool columns, averaging neutral, moves
the controls by at most two units; a point sample swung them to sixty.
The scene clicked 40px to the right of "pick", on "reset", so the
recording showed a neutral group being reset and a click on the wall
that panned. Re-recorded with the picker fixed: the word lights, the
sample moves temperature and tint, and Before shows what it corrected.
Sampling the overcast sky on a Canon 6D frame set tint to -100 and
temperature to -15 for a patch the canvas showed as pure white. A clipped
photosite is sensor white, not a colour: every channel stopped counting,
so what the tap hands back is the as-shot multipliers themselves, which
are strongly magenta, and the solver dutifully drove green to its stop.
The display shader already fades such a pixel to a neutral of the same
brightness before any operation runs, so the picker was balancing against
something the photographer could not see.
The probe now refuses a sample with any channel at or above the onset the
shader fades from, the way the solver already refuses black. The threshold
is one constant, CLIP_ONSET, formatted into the shader and read by the
probe, so the two cannot drift apart.
Pressing "pick" and clicking a near-neutral wall on a Canon 6D frame set
tint to -77 and turned the whole photograph green. The white balance
operation runs first in the chain, on camera RGB, before the body's base
curve and colour matrix; the probe was read off a display render after
all three, and the solve treated that sRGB triple as if the gains
multiplied it directly. On a JPEG the two spaces coincide, which is why
the existing tests passed while the picker was broken on every raw file.
The probe now reads the camera-space tap a merge stitches from, composed
under the edit's own framing so a fraction of the canvas is a fraction of
the probe, and puts the as-shot balance on itself - exactly the value the
operation's gains are about to multiply. No operations run in the tap, so
nothing has to be stripped and restored, and the display target is left
alone, so a sample that found nothing usable no longer needs a redraw.
A raw-frame test with the 6D's matrix and a typical as-shot balance
samples a warm grey and asserts the rendered pixel comes back neutral; it
fails on the previous probe.
Every face the user has ruled on entered the pass as an anchor, and the
scan is exhaustive by design (`dr_face::neighbours`), so a person with
750 confirmed faces cost 750 comparisons against every other face in
the library — and the cost of a library grew with how well it was
named. Most of those comparisons said nothing new: thirty frames from
one afternoon are one point of view, not thirty, and a face that
matches one of them matches the rest.
Each person now enters through at most 100 of their anchored faces
(`dr_face::references`). Eligible are those whose raw embedding is at
least 15 long — one above the gallery floor, since a reference speaks
for someone rather than merely being admitted — with an unmeasured
length admitted as it is everywhere else. From those, the set spanning
the greatest volume is chosen greedily: the longest vector first, then
at each step the face with the largest component orthogonal to the
chosen so far. That is pivoted Gram–Schmidt, and the product of the
residuals it picks is the Gram determinant, so the greedy step is the
exact greedy on the objective. A near-duplicate of a chosen face has
no residual and is passed over; the one profile shot among two hundred
frontal frames is taken early; faces inside the span of the chosen add
no volume and are not taken to fill the cap.
The faces not chosen keep their confirmations and are not touched by
the pass — they stay in the anchor map, so it never releases them —
they are simply not compared. A person none of whose faces is long
enough is still stood for, by their longest, rather than losing their
anchor and having their next face filed as a stranger. Under the cap
nothing changes: every eligible face stands, and the short ones stay
in as the probes they were.
At the reference library's 3,851 confirmations the scan shrinks by
about a fifth; at 15,000 it is a fifth of what it was.
The markers the previous commit stops writing are already in the
catalogs — 2 on the desktop, 429 on the tablet — and in the shards
both have exchanged. Renaming them to the faces' own id with a fresh
time is what makes the export send each image again, under an entry
newer than the empty one `held_model` would otherwise pick. Where the
old write had inserted its marker beside the right one, the wrong one
goes and the right one is refreshed for the same reason: its entry in
the shards is older than the empty one.
Images V14 left with faces and no marker are not touched. That state is
the quality pass's cue, and the fixed write marks them correctly when
it reaches them.
Checked against copies of both real catalogs: the desktop renames 2,
the tablet deletes 429, both in under 200 ms.
`record_updates` — the write behind the quality, eye and crop passes —
re-marked the image as indexed under the pipeline the pass ran as, and
left the faces it had updated under the id of the detector that found
them. On a desktop set to Thorough that put `scrfd_10g+w600k_mbf` over
faces spelled `w600k_mbf`; on the tablet, `scrfd_10g_i8+w600k_mbf` over
faces it had adopted from the desktop's thorough pass.
Every reader takes the marker and the faces to agree. `marker_under`
reads the marker as the detector having examined the image, so the
upgrade repair never revisits it. The shard store keys each face by
its pipeline id, so `export_to_shards` selects an image's faces by the
marker's id, finds none, and sends an entry that says the thorough
detector looked and found nothing — over photographs with named faces
on them. The desktop's shard index holds 54 such entries beside real
faces; the tablet's eye pass over the faces it had adopted made 430
more, and both devices have exchanged them. `held_model` takes the
newest entry for an image, which is the empty one. Nothing has been
lost yet only because the two spellings of the thorough detector rank
equal and neither side adopts the other's; a third device, or either
one after a reinstall, would adopt "nothing here" for 484 images. And
the desktop's eye pass is 4,739 images from doing the same to every
face from before V14 — which are the ones that only exist on the
desktop, and would then never reach anywhere.
The marker now takes the id the faces carry; the pass's own id is used
only when it dropped the last of them and there is no detector left to
name. A stale marker under another spelling of the same embedder is
removed in the same transaction, so one embedder has one marker.
The tablet showed a fraction of each person: 681 of the desktop's 3,851
confirmations, and none of Ian's 746, Catherine's 626 or my own 480.
Every face that existed on both devices agreed on who it was, and the
people rows were identical — the merge was fine. The missing 3,170
confirmations were on faces the tablet did not hold at all: the
desktop's 16,080 faces from the original detector, on 4,310 images,
detected before schema V14 kept the quality reading.
Those faces were in shards the tablet had already downloaded, in
August's export. `import_from_shards` looked at them on every sync pass
and declined each one, because a face without a quality reading was
"work this device cannot finish": adopting it would write the run
marker, and the marker was what stopped an image being looked at again.
That was true when it was written and has not been since the quality
repair existed — that pass lists its work by `f.quality IS NULL`, not by
the marker, exactly as the eye pass does, and faces without an eye
reading were already adopted on that reasoning.
The refusal had no exit. V14 had deleted the markers of every image
holding such faces so the quality pass would find them, and
`export_to_shards` walks the markers, so the desktop never re-exported
them either; the unmeasured August copies were the only ones there
would ever be. The tablet's answer was to queue all 17,727 images for a
re-detection of its own, a fetch of the whole library, while holding
the faces on disk.
Adopt them. The receiving device's quality pass measures them when it
reaches them, and the desktop's confirmations match onto them by box
overlap on the next catalog merge. The test that asserted the refusal
now asserts the adoption and that the image is still owed to the pass.
Eighteen declared operations, not fifteen; JPEG XL is an export format;
the grid does not filter by keyword, only the catalog's query can; and a
panorama's provenance is a sidecar beside the composite, not a history
step in it.
It said 0.9.0 against a 0.13.1 tree, listed focus peaking and burst
grouping as unbuilt when both have shipped, and opened with a page of
prose about the display path before saying what the application does.
Lead with what it is and a picture of it, say how to get it on each
platform and what state each channel is in, keep the honest account of
what is missing, and put the manual first in the documentation table.
A CLAUDE.md at the root, for anyone changing this code: the ways a redraw
and a catalog read came to cost half a second per click, what each fix
looked like, and how to measure the next one against a copy of a real
catalog.
"How many images still owe a quality reading" was a correlated EXISTS per
image over `faces`, and the face row is 8 KB of embedding and crop before
the column it looks at, so each count opened every row. Six such counts
run on every open of the Identity screen and at the end of every sweep:
160 ms on the reference library.
V19 adds three partial indexes holding only the faces still owing each
pass, keyed on the image and carrying the model id the predicate reads,
and replaces `faces_image` with `(image_id, model_id)` so "does this image
hold this embedder's faces" is answered from the index too. The planner
takes a partial index when the count is driven from `faces` and ignores it
inside the EXISTS, so `Needs::Face` carries the per-face fragment and
`repairs::count` spells the query from the faces' side; the list and the
per-image check keep the EXISTS. A test holds the two spellings to the
same answer for every repair.
Every `move_to` guaranteed its destination's parent with a `MKCOL` for
each ancestor down from the account root, and a trash folder under a
library root several levels deep meant three round trips answering
`405 Method Not Allowed` before the one `MOVE` that did anything — for
every image of a delete, on a connection built for that job.
The backend now records the collections it has confirmed exist and asks
about each once. It lives for one job, so a folder another client removes
mid-batch is the one case this misses, and the `MOVE` then reports the
`409` rather than hiding it.
`faces::people` grouped `face_person` after a LEFT JOIN over every person
and sorted the lot by name; the rail then discarded the empty, unnamed
groups a regrouping pass leaves behind — 17,000 of 19,000 rows on the
reference library. `people_in_use` filters them in the WHERE and joins
`people` to face counts aggregated first (2,000 groups), so the sort sees
only the rows that will be drawn. `count_unassigned` replaces fetching
2,400 ids to take their length. `load_people` 22 ms → 10 ms.
`confirm_all` called `faces::confirm` per face, and `split_off` called
`reject` then `confirm` per face: each opens and commits its own
transaction, so a click on a group of several hundred was several hundred
commits. `faces::confirm_all` is two statements — clear the rejections the
confirmations override, then flip the rows — and `faces::reassign` does a
split's reject-and-confirm for every face under one commit. 16 ms → 2 ms
and 22 ms → 4 ms on the largest group.
A confirm or a reject changes one row and redraws the whole grid, and the
redraw re-read every crop blob of the selected person (4 MB for the
largest) and decoded every one — 316 ms per click on the reference
library's 754-face person, to arrive at the pixels already on screen.
`load_faces` now takes the crops the previous load decoded, keyed by face,
and moves each into its new cell; the blob read is skipped when every face
is already in hand. `refresh` drains the old cells into it rather than
cloning them. The redraw is 2.6 ms.
Every click on the Identity screen's face grid — confirm, reject, split,
rename, merge — redrew the whole screen, and the redraw recomputed the
coverage line. That line lists every repair's outstanding images to count
them: six scans of the images table with a correlated EXISTS over the
8 KB face rows, an ORDER BY the job's visiting order, a Target with its
path per row, and a thumbnail-index query per image with faces. On the
reference library (24k images, 19k faces) that was ~200 ms of the
~540 ms each click cost, spent computing a figure a confirm cannot change.
`refresh` now takes what changed: `Changed::Identities` re-reads the rail
and the grid and leaves the coverage line alone; `Changed::Library` — an
open, a sweep ending or stopped, the face data deleted — re-reads it too.
For the times it does run, `repairs::counts` counts instead of building
and dropping the lists, and the thumbnail store is read once
(`ThumbStore::held`) rather than probed once per image in the audit, the
outstanding list and the proxy repair.
`identity_bench` is the measurement: the reads a click performs and the
batch writes, timed against a copy of a real catalog.
docs/manual/README.md is a tour for a photographer opening DarkRoom for
the first time — one picture per thing, moving where movement is the
point. tools/manual/ is how the pictures are made: drive.py puppeteers the
desktop build on a private Xvfb (launch, click, drag, type, screenshot,
record), scenes.py is each picture as a script, and record.sh runs them
all over a folder and writes the results into docs/manual/media/.
The media is in LFS, with the CI pulls excluding it as they exclude the
fixtures; a screenshot changes wholesale when the interface does.
Nothing in the pictures shows a person, by design: the demo library is
seventy urban and alpine frames, chosen from the catalog's rows that face
detection found nobody in.
The traceability matrix is regenerated here after the rebase that
brought this branch up to master.
Without focus the keys typed into the keywords sheet went to the grid
behind the scrim, and the Return meant for the keyword opened a
photograph. The naming sheet already takes focus on open; do the same.
An empty trash said "No images found — check the library folder and which
formats are ticked", which sends someone off to fix a library that is
fine.
An empty device destination read "Ask each time" on the settings page,
and nothing asks: an export made with the field blank is refused with
"no export folder is set". Say what will happen.
wgpu reports a device out of memory by panicking, and a twelve-frame
merge on a GPU another process is using is where that happens. The panic
unwound the worker, the sender went with it, and the page sat on "Stop"
with every control disabled and nothing to say why — the crash record on
disk was the only sign. Catch the panic and send it as a failure, and
treat a closed channel with no final event as a dead worker too.
The eyes are per layer and outlive the mode, so a photographer coming
back finds the layers they were looking at still lit. But the tint is a
way of looking at a mask, and outside Local there is no mask being looked
at: the sky stayed red through Repair and back in Photo, a mode that had
been left leaving its overlay behind — the fault ui-navigation.md D-N1
exists to prevent.
Five chips in one row declare 440px, and the develop column takes the
widest panel's request — so selecting a category mask levered the sidebar
past the window's edge, clipping the histogram, the group strip and the
subject list. The same trap ChipGrid's comment records for film formats.
A heading was drawn only on a cell that both began a month and began a
row, so at seven columns most months were never named, and the one
heading on screen — always on the window's first cell — was wrong about
every row below it. Worse, two headings drawn on the same cell overprinted
each other. Now a row carries a heading whenever its first cell's month is
not the one last announced: a month starting mid-row is named on the next
row it opens, one row late and right about everything under it.
Every shard is in WAL mode and every put opens its own connection, so
while thumbnails are being generated on several threads — which is when
the first sync pass runs — the log is never checkpointed and the main
file holds whatever the last quiet moment left in it. For a shard created
seconds earlier that is nothing: a zero-byte file with the schema still
in the log. The sync read that file and uploaded it, and every other
device merging it failed with "no such table: thumbs" on every pass.
Copy the shard through SQLite's backup API into scratch first, which
serialises against writers and carries the log, and upload that.
Confirming "/" in the folder picker set an empty root, which the launch
model read as no root at all: "Open library" stayed disabled after the
question had plainly been answered, and a folder library — whose folder
is the whole library — could never be opened without first descending
into a subfolder of it. The empty string was carrying two meanings.
Record the choice as its own fact on the account (`root_chosen`, defaulted
so existing configuration loads unchanged), treat a folder endpoint as
chosen by definition, and let the launch screen say so: a folder is shown
as a LIBRARY rather than an ACCOUNT, the second question becomes an
optional "scan only a subfolder", and the library header names the folder
instead of calling it "· whole account".
The scan reads the first 256 KB of a file for its metadata. A camera
writes its IFDs at the front, so that is the whole structure; the linear
DNG a merge writes puts its first IFD after the pixels, and rawler,
given the head alone, finds no decoder in it. The composite was
catalogued without a date and sorted to the very end of the grid, after
every dated photograph — which is where a panorama merged on the tablet
went unfound.
dr-decode's own TIFF reader now reads through a head and a tail at a
known offset; trailing_ifd says where the tail starts and
metadata_split reads the two together. The scan, when the head fails
and points beyond itself, fetches from the IFD to the end — kilobytes —
and dates the file from both. Tested against the writer's own output.
The Malvar "R at green in R row" kernel weights the two greens two
sites away along the row at -1 and the pair up and down the column at
+1/2. The shader had the two swapped, in the comment as well as the
code, so the transcription checked against itself. Both sum to zero
and reconstruct a flat patch exactly, which is all the tests fed it.
On an edge the correction at green sites is half strength and the
false colour doubles: 0.375 against 0.19 on a grey step, and a
blue/yellow zipper around every clipped highlight at 1:1. The other
three kernels and the CFA tables were right.
A grey vertical step now runs through the pass; the transposed kernel
fails it at 0.375.
The reference desktop's only system ONNX Runtime is Arch's
onnxruntime-opt-cuda: 1.29, built without TensorRT and against cuDNN 8
on a cuDNN 9 machine. The probe rejects both providers correctly and
the app runs on the CPU provider, which is right and not what anyone
wants. runtime/ beside the models is now searched ahead of /usr/lib,
tools/fetch-desktop-runtime.sh fills it with the four libraries from
the current onnxruntime-gpu wheel (cuDNN 9, TensorRT 10), and the
About caption lists every rung that lost and why, not only the first.
Verified: the app selects TensorRT from that directory with no
environment variable set.
The unpack list gained migan-512.onnx without its length following;
nothing on the desktop compiles that crate, and the first Android build
of 0.13.0 stopped there.
A fill that went wrong took a seven-minute merge to look at again. Now
DR_FILL_DUMP=dir makes the merge write what the filler was given, and
the fill example runs fill_border on that, or a crop of it, on the engine
and writes coarse, each band and the feathered result as PPMs — seconds
per attempt on TensorRT. Both examples take DARKROOM_ORT_DIR as the app
does, and --wait-engines lets a compiling rung finish before timing.
A Border choice beside the projection — crop to the picture, or fill it
— that redraws the preview filled so the invented pixels are seen before
they are confirmed (FR-MRG-1), greyed with the reason when the model is
not there. The job fills at half the composite's resolution in a
display-ish space (white balance, matrix, gamma; invertible) and samples
the result back into the linear DNG wherever no frame reached; the
sidecar's merge line says border filled and with which knobs.
Experimental because the fill is right in thin borders and wrong in deep
corners, where the model's Places2 prior puts clouds in sky and water
under grass; so its six knobs — working scale, edge erosion, coarse pass,
band width, mirror depth, seam feather — are sliders under the choice,
each committing a redraw, until the defaults are right.
dr_pano::fill owns everything the model does not — which tiles, what
context, how to blend — behind an Inpainter trait, and dr_pano::migan is
that trait over the shipped generator on the inference engine.
The known content is mirrored across the coverage edge into the hole and
a 256-px ring, the nearest 48 px folded, so the model interpolates between
real and mirrored sky rather than extrapolating into nothing. A coarse
pass at a quarter decides the structure with the whole border in a few
tiles; fine passes in 96-px bands from the edge outward texture it; the
seam is blended over a feather inside the real edge. Every knob is a
Params field, and an Observer hears each stage for whoever is looking at
why a fill went wrong.
A border fill acquires the filler once a tile, and each acquire hashed
the 28 MB model twice — 60 ms a tile, a third of the tile's run on a
throttled TensorRT. The Model keeps its hash from open.
Sargsyan et al., ICCV 2023; MIT code and weights (models/LICENCE.md),
exported by tools/export-migan.sh at a fixed 1×4×512×512 from the
authors' checkpoint — six operator types, 28 MB, in LFS like the rest.
The package installs it beside the scene model and the APK unpacks it
with the others.
The fused shader stored black with alpha 1 for a pixel whose source
coordinate left the frame, and the merge's warp averaged it in like any
other: a dark, badly interpolated fringe along every frame's edge, visible
as a seam wherever a frame ended and, later, as the edge the border fill
continued. The display keeps its opaque black; CameraLinear stores alpha 0
and the warp weights each sample by the alpha it interpolated, dropping a
sample that has none.
MI-GAN is plain convolutions, so every rung serves it and none needs a
special form; the role exists so resolve_model and the probe's fingerprint
know the model, and so the merge job can open it through the engine rather
than tract, which takes 7.4 s a tile for it.
A library's records are never all complete at once. A face found before
its quality was kept has no quality; one found before the eye models
existed has no reading; one adopted from a peer's shard has no crop; an
image the fast detector examined on a 1024 px proxy has boxes the current
detector would not have drawn; an image the scan stat'ed has no capture
date. On the reference library that is 17,762 faces under the bare
w600k_mbf id with no quality, no reading and no dense landmarks, 4,144 of
them without a crop, beside 12,217 images the fast detector examined and
found nothing in. Every one of those gaps was its own pass — V14's
measuring pass, §17.5's eye pass, the sweep's proxy repair, the sweep's
detector upgrade — with its own work list, its own count and its own idea
of done, and adding a per-face field meant adding a pass. There was no
pass at all for the case the library is actually in: boxes and landmarks
drawn by a weaker detector on a proxy, which every later per-face pass
would have read from.
dr_ui::repairs replaces them with one job over a registry. A Repair names
one thing a record can lack — the predicate that says which images still
owe it, the input its handler needs (a header, the original, or a native
render), the handler, and what to record for an image that can never be
done. The job unions the predicates into one work list, fetches each
image once at the most any claimant asks for, renders it at most once,
and runs every handler whose predicate that image still matches, checked
again before each because a detection writes every field a per-face
handler would fill. The registry today: face-proxy, face-quality,
face-eyes, face-crop, face-detection, face-upgrade, metadata — the last
there to say that this is not a face job. Adding a field is one entry.
A repair's predicate is the only definition of its work: the count the
settings page shows, the list the job fetches and the check before its
handler run are one predicate, so the job converges. That is why the
registry is cut to what the device can do rather than listing what it
skips — an entry is a count and a set of originals to fetch — and why an
eye reading that cannot be cut is not a criterion.
The catalog side is generic to match: record_updates writes whichever
fields a FaceUpdate carries and re-marks the image so the shards export
it; faces_needing and count_needing answer a predicate the caller
supplies, replacing the measuring pass's three special cases.
Two buttons on the settings page run the job and differ in one
predicate. "Index faces" converges on coverage: has anything examined
this image. "Re-index every face" converges on provenance: face-detection
claims every image with no marker under the chosen detector, in either
of its forms (FaceDetector::model_ids, so a desktop in f32 and a tablet
on the Hexagon do not re-index each other's work), and a marker saying a
weaker one looked is not that. An original over the fetch budget is left
exactly as it was under the re-index, where the sweep marks it examined:
a re-detection with nothing found would delete the faces, and "cannot
fetch" is not "no faces".
record_detections replaces an image's faces and carried only the user's
confirmations onto the new ones, by box overlap above 0.5 IoU. Everything
else on the old faces was dropped: the suggestions the last grouping pass
made, and the people the user had said a face was not. On the reference
library that is 13,011 suggestions and 77 rejections beside 3,778
confirmations — a re-detection of it would have been correct by
FR-CULL-12's letter, since suggestions are derived data, and would have
handed back a People screen of strangers.
Now every old face is read before the delete — box, vector, assignment,
rejections — and matched to the new faces one-to-one, best pair first. A
pair qualifies when the boxes overlap at all and either the overlap alone
says so (IoU above 0.5, the old rule) or the embeddings do (cosine above
SAME_FACE_COSINE, 0.45, the reference library's P≈0.95 line). The
embedding route claims the box a low-resolution pass drew badly enough
that overlap alone would not; the vector is also what breaks the tie in a
group photograph, where two neighbouring faces overlap both new boxes.
Overlap is required on both routes, because the same vector elsewhere in
the frame — a mirror, a print on the wall — is not the same face and must
not take its name. Onto the matched face go the assignment as it was,
confirmed or suggested with its probability, and every rejection.
The merge's match_faces still matches by overlap alone across devices; it
is the same question and is not changed here.
ONNX Runtime's default is already its fullest level. ort-tract maps any
level but disabled to tract's optimiser, whose slice pass divides by
zero inside the segmenter's graph — a panic across the C API and so an
abort, which is what stopped dr-ui's develop test. The app never asked
tract for that and does not start now.
On the tablet the engine compiled arcface for the NPU: the routing
compared the form a rung wants with the form on offer, and for the
embedder both are f32, so nothing said no. A rung now says which roles
it serves at all, and the Hexagon does not serve the embedder (§7 —
its vectors must compare across devices). Tested at the routing seam.
dr-segment's onnx_probe example still named ort-tract, which is what
stopped the workspace test build.
The strict flag refused the Hexagon over the ten quantise/dequantise
nodes at the graph's edges that QNN declines by policy, which cost
microseconds. A provider that hands real work to the CPU is slower than
the CPU floor and the timing already rejects it; the tablet measured
2.3 ms on the NPU against a 29.7 ms floor.
ONNX Runtime's errors open with a source path and a template signature;
the first 160 characters of a CUDA failure were all signature. The
reason now starts at the first word a person can act on.
XFeat's two exports are a Keypoints role now; the crate no longer names
tract, and the app compiles TensorRT engines for both ahead of the
first merge. The probe picks the smallest *detector* rather than the
smallest file: the tablet's first run chose the 112 KB eye classifier,
which has no int8 form, and reported the Hexagon as failed for want of
one.
The first int8 files found no faces at all, and for two reasons the
tool now guards against. The calibration set was landscape photographs
with no faces in them, so the score head's ranges had never seen the
face regime; the set is now proxies from the library itself. And ONNX
Runtime's strided and moving-average calibration modes both degrade
these graphs measurably (a quarter of the faces at eight images, none
at ninety-six), while driving the calibrator in chunks by hand gives
ranges identical to a single pass — so the tool does that, four images
at a time, and feeds quantize_static through its range cache.
Measured against f32 over 400 proxies (docs/inference.md §10.1): the
10g form finds every face above 32 px the f32 form finds; 500m and
2.5g find 96%, and what they lose sits at a median confidence of 0.52
against the 0.50 threshold. Shipped with the number on record.
The Android unpack list gains the three int8 files; without that the
tablet never saw them. D13's runtime half records the reopening.
tools/quantise-models.sh writes the QDQ form QNN's HTP backend takes
whole: opset 17, per-channel int8 weights, uint8 activations, ranges
from running the f32 graph over photographs fed exactly as the app
feeds them. The calibration is strided, four images at a time, because
every ONNX Runtime calibrator holds each image's whole set of
activations until it folds them — a gigabyte an image on the 10g
detector, and an OOM kill with no message when folded once at the end.
Release-time, never on the device (docs/inference.md §5): it needs
real photographs and a person reading the recall measurement that
gates whether each file is offered.
The desktop names where a package may have put libonnxruntime — an
override variable, beside the executable, the package's own library
directory, the Flatpak prefix, the system library directory — and
Android points at the APK's native library directory, which is also
what Qualcomm's DSP loader must be told for the Hexagon skel. Android
starts the engine at the end of the model unpack rather than at launch,
because the probe fingerprints the model files and a first launch has
none until then.
The About panel gains an Inference row beside Graphics, re-read every
two seconds while the probe runs and engines land, and faces.model_id
carries the detector's form: an int8 detector finds a different set of
faces and is a different population (docs/inference.md §7). A
low-memory signal drops every idle session with the GPU caches.
The APK assembly bundles ONNX Runtime and the Qualcomm HTP libraries
from Maven, fetched by tools/fetch-android-runtime.sh with their
published checksums; RUNTIME_DIR=none builds the tract-only APK, which
is a slower app and not a broken one. The desktop packages carry no
runtime yet.
Two probe fixes from the first desktop run: the floor must not be
built with CPU fallback disabled, and a versioned libonnxruntime.so is
a runtime too. On the reference desktop the probe now loads ONNX
Runtime 1.30, measures 30 ms on the CPU provider, and selects TensorRT
at 1.5 ms.
One crate names the runtime, the providers and the devices; dr-face and
dr-segment ask it for a session by role. It hands ort an API table once
per process — from a libonnxruntime it dlopens when the app names a
directory holding one, otherwise from tract — so the Rust build stays
free of C on every target and a package can install the runtime as a
file (docs/inference.md §3).
Sessions live in a registry behind a Model handle that holds the bytes,
not the session: every use refreshes a timestamp and a reaper unloads
whatever sat idle past the decay. A scan that runs the detector on each
image never lets it go idle; a click in the develop view lets the
segmenter go after thirty seconds; a handle used after that reloads,
and reloads on a higher rung if a compiled engine has landed meanwhile.
The probe walks the platform's ladder by building strict sessions and
timing them against the CPU provider, caches the choice against a
fingerprint of the runtime, driver, hardware and models, and compiles
engines for the selected rung in the background, smallest model first.
Nothing in this commit turns the native path on: the apps still run on
tract until they call init with a runtime directory.
tract runs every model on one core on every platform. Measured against
ONNX Runtime's providers on the MagicPad 2 and the reference desktop:
ORT CPU alone is 3-10x, the Hexagon at int8 runs the detectors in
1-3 ms, TensorRT is ~2x the CUDA provider. NNAPI, XNNPACK, WebGPU and
CUDA int8 were tried and excluded with the numbers that excluded them.
The spec keeps the build C-free: ort::set_api takes a table from a
dlopened runtime or from ort-tract, chosen once per process. Rungs
are chosen by building a real session, cached until an input changes,
and compiled engines are built in the background after the first
frame. The embedder stays f32 everywhere; int8 detectors are a
distinct model_id and are gated on a recall measurement.
Read and measured, not built. The bare 512 generator exports at a fixed
shape and loads under tract with nothing unsupported; at f32 on the
desktop CPU it takes 7.4 s per 512×512 tile, which puts a full-resolution
fill of the fixture's border at ten minutes. The three routes that would
make it viable are recorded, with the quarter-resolution fill the cheapest
and Hexagon int8 the one the model was designed for.
Picking a chip stored the choice for the merge and changed nothing on
screen — the chip did not even highlight, since the selected property
was never written back. Now the pick is reflected, and the job, waiting
for its decision, takes a Preview request, draws the alignment on the
chosen surface at proxy cost and reports again; the drain puts the new
picture and its size up. Auto is the surface the field of view suggests.
Also:
The largest rectangle inside the frames' coverage is found a row at a
time — a histogram of consecutive covered rows and a stack pass per row —
so the composite is never held to be measured (FR-MRG-11). It is written
as DefaultCropOrigin/DefaultCropSize (FR-MRG-4): the file opens on the
picture, the border is still in it, and resetting the crop shows it.
rawler reports the crop as the picture, which the test checks.
FR-MRG-4 records the question raised the same day — fill the border
rather than crop it — as open: a non-generative fill through the heal,
or a generative inpainter with its licence and weights. Neither decided.
A sweep whose frames overlap by more than half reads as one photograph,
and the page's job is to show frames. Each footprint is walked along its
border and drawn in amber where it lands, so twelve frames look like
twelve and a misplaced one is visible as such. The preview may take most
of the page's height rather than 320 px.
derived_from and merge are top-level sidecar fields (FR-MRG-6): one line
per source in order, and how the composite was made. A build that
predates them keeps the lines as unknown and writes them back. The job
writes the sidecar beside the composite and stages it with its own
record when the composite goes through the outbox.
DARKROOM_START_MERGE=a.CR2,b.CR2 lands on the merge page at startup with
the job running on local files, on the model of DARKROOM_START_IDENTITY,
for looking at the page where synthetic clicks do not reach it. The fetch
and the start are shared with the grid's button.
panorama.md §11 records what exists, the fixture's figures, and the six
things still open, auto-crop first.
dr_ui::merge is the orchestration with no interface in it: decode each
frame to sensor data and build its graph as a session would (orientation,
lens profile); render each through the camera-space tap at proxy size and
detect keypoints there, so the alignment is measured in the undistorted
frame the tiles are rendered in; align; solve one gain per frame from the
proxies' overlaps; draw the aligned set in colour for the page; then wait.
Nothing is written until a Decision arrives (FR-MRG-1). The merge writes
a linear DNG through the outbox with a destination record, so the drain
puts it beside its sources on a folder library and a server alike, and
the library rescans (FR-MRG-3).
merge.slint is the page, on the import page's model: the alignment
table with a failed frame named on its row and the button held off
(FR-MRG-5), the preview, the projection choice, Stop and Back. A
"Merge to panorama" button joins the grid's selection bar at two frames.
Headless, the example produces the fixture's 22 993 x 5 980 DNG in 45 s
on the reference desktop, exposures balanced across the stop of drift.
dtourolle