Raster renders, ink renders, thumbnails and node renders now lend the
document's composite fonts to the font database while they draw, as the
PDF, SVG and EPS writers already do. A document that is not in front, a
document read from a file (library thumbnails, save previews) and a
placed document's art drew such type in the fallback font or in another
definition of that name.
render_node_with and render_node_thumbnail_with ignored RenderOptions::supersample
and RenderOptions::precise, so the Asset Export tiles, which ask for both, were drawn
with the canvas's shape-by-shape edges and the fast render mode instead of the export's
look. A node render now supersamples Art Optimized edges as a page render does, and
picks the quality render mode when precise is set, for the art's offscreen groups too.
a_coarser_glow_kept_whole_draws_after_a_pan_as_made_again compared only
whole-pixel pans with a fresh render. At a fractional pan it now compares
the pixels right of the rectangle, where only the glow draws, with a fresh
render at the whole pixel nearest the pan (either neighbor on a half pixel).
RenderOptions::background and render_node_thumbnail's background are
straight-alpha colors, as over_background reads them; their docs said
premultiplied.
Flatten Transparency composited each region over white, so a blend mode
on the bare page read that white. Raster renders now draw their
background under the finished art, and the flattened color no longer
matched them: a Color Burn object on the bare page flattened to white
and rendered in its own color. The region now composites the art alone
and puts the white under it.
- Tests that relied on the white under the art now draw art under the
style (`every_builtin_style_paints_visibly_on_a_rectangle`) or compare
only whole-pixel pans with a fresh render
(`a_coarser_glow_kept_whole_draws_after_a_pan_as_made_again`).
- New tests cover every blend mode on white and blue backgrounds, a
thumbnail and the ink planes with a Screen glow, and
`over_background` at background and pixel alpha 0, 128 and 255.
A raster export filled its background into the page before the art, so
blend modes read it as art underneath: an Outer Glow in Screen mode, the
default, turned white on a white background and was missing from PNG and
JPEG files while the canvas showed it. Blending against nothing also
differed from the canvas for Difference and the other modes.
The background now goes under the finished art, in frames and in object
thumbnails. A blend mode reads only the art below it. A white object
under a Difference object still inverts.
Page Isolated Blending no longer changes a raster export on a bare page.
Docs: docs/mcp.md, docs/target-app-parity.md.
Review of #6290.
- An edit is measured as the renderer draws: an object it leaves out
(hidden, a template layer or guide the view hides, drawn hidden, the
instance a symbol is edited from) is measured as nothing, by one
predicate the renderer and the measure share (node_checks::left_out,
before culling).
- An Image Trace object's members are compared one by one again, its
hidden source image, which some views draw, counted as shown: moving
the source redraws where it was and is, and recoloring a traced shape
redraws only the shape.
- The renderer checks the brush library between frames only while it
keeps bounds measured with brushed strokes in 3D art.
- Tests: the sweep of 3D art's bounds counted while the art is held,
and Fit to Artwork Bounds' menu answer kept for a revision.
When an edit changes an object, the canvas draws again only the area
the object painted before and after it, measured by its cull bounds.
Type and groups with Extrude & Bevel, Inflate or Rotate were measured by
their letters or members, not by the solid the effect draws, which
reaches well past them. Making the solid shallower or turning it left
pieces of the old solid on the canvas outside that area until the view
was drawn again.
An object with a 3D effect is now measured by the effect's art, as the
renderer draws it: with its geometry effects, each object in it with
the raster effects of its own fills and strokes and the art its brushed
strokes draw, grown by the reach of the object's own raster effects.
One function measures that art for every bounds the renderer and
exports take, where three copies of it disagreed.
- The renderer measures the art it draws. What it makes to measure an
object it holds until the frame ends, with what the art read from
outside the object (fonts, symbols, color settings), and keeps it
with the art it draws if it draws the object while that is
unchanged. Art measured but not drawn (culled, off the view) isn't
kept. It holds at most as much as it keeps for the art it draws
(64 MB); past that, an object is made again to draw it. An object
whose 3D effect builds nothing (Revolve on a group, a solid too
complex) is measured as it is drawn otherwise, as without a renderer.
FrameStats::three_d_built counts the 3D art the renderer makes.
- The canvas measures what an edit changed through the renderer that
drew the frame shown (Renderer::changed). When it draws the part on
its own thread, it draws it with the art it made to measure it; when
the render worker draws the frame, or nothing in view changed, that
art is let go. Under Overprint Preview or Separations Preview, which
draw copies of the objects, the edit is measured afresh. An object
the frame doesn't draw (hidden, in a hidden group or layer, or
hidden by the view) is measured as nothing, before and after, so an
edit to it builds no solid: changed() without a renderer counts it as
nothing too, where it measured it. The canvas measures the edit
before it knows where the frame is drawn: when the worker draws it,
the solids it built to measure it are built again there.
- Fit to Artwork Bounds, the Artboards panel's check for it and
per-artboard SVG export measure every artboard through one renderer
(Renderer::art_bounds_on, Renderer::effect_bounds), which keeps each
object's bounds for the pass, a group holding 3D art or a hidden
member included, so each solid is built once: 30 extruded type
objects over 30 artboards take about 2 ms, where measuring their
letters took under 1 ms and building each solid for each artboard
would take 70 ms. Objects with brushed strokes, which culling leaves
room for and exports don't, are measured on each artboard. The
Artboard tool's canvas menu keeps whether Fit to Artwork Bounds has
art to fit for the document's revision, where it measured it on
every frame it was open.
- Other exports, Object › Rasterize and changed() without a renderer
build the art on each call, as 3D paths and groups already did.
Exports and Rasterize now take the solid of 3D type, where they
cropped it to the letters, and the art its brushed strokes draw, not
the room culling leaves for it. Rasterize on drawn 3D art measures it
once while the art is kept.
- A shadow after a 3D effect, on the object or on a fill after the
fill's own 3D effect, reaches past the solid once; it was counted
twice.
- With Crop Marks or Vector Halftone, which draw other art, the object
is measured by that art, as before.
- Culling leaves room for a 3D object's brushed strokes as for any
object's, and the renderer measures a 3D object again when the brush
library changes: a brush widened past that room would leave the
object culled.
A document's first frame builds the solid of every 3D object, those off
the view too, to know whether it reaches into the view: 200 extruded
words with 150 off the view take about 12 ms, where measuring their
letters took 6 ms.
Type whose object effects leave it as it is (Revolve builds no art from
type) is measured by its own box grown to hold its glyphs, as plain
type is; it was measured by its box alone, which needn't hold them.
The reference check now applies wherever a painted effect's geometry is
used: a result whose swatch or pattern is refused keeps its input instead
of giving its geometry. Pathfinder art of type, reshaped groups, compound
shapes, image fills, the renderer's bounds and Object > Expand look the
references up in the document, and the renderer drops its kept bounds and
geometry when the names of the swatches or patterns change.
An envelope mapped type from its laid-out box: ascent to descent for
point type, the frame for area type. The distorted glyphs then filled
less of the envelope than other art does, with room above the capitals
and below the baseline.
The envelope now maps type from the bounds of its glyph outlines,
underlines and strikethroughs when the type is on its own, in groups
and layers, or used as a clipping path. Inline graphics count by their
boxes. Rotated type and type in perspective map from their outlines as
they are drawn. Make with Mesh,
Make with Top Object and Reset with Mesh without Maintain Shape use
those bounds. Type that draws nothing keeps its laid-out box, and so
does type inside any other object, such as a blend, a repeat or a
symbol instance. Shrink Text to Fit type is measured as it is laid out
on the page.
vectorcraft-text's envelope_ink measures the type, and
register_envelope_ink registers it with the document crate. The engine
session, Renderer::new, live evaluation in vectorcraft-effects and the
.ai reader register it. The document crate keeps the measured boxes per
thread, by object and by a hash of the content, and measures again when
the type, the fonts, the hyphenation exceptions or the transform's
scale or rotation change.
This is VectorCraft's own rule. Envelopes of type made in VectorCraft
and in saved files map from the glyph outline bounds instead of the
type's laid-out box, so they redraw differently.
Opening a document laid out every type object for its bounds and
baselines and threw the layouts away, and the first frame laid the same
objects out again. The open now keeps them, up to 128 MiB, and a frame
takes each one once.
A frame takes a layout only for a text object equal to the one laid out,
and only while the fonts, the system fallback setting and the hyphenation
exceptions are unchanged. When the last frame being drawn ends, the rest
are dropped if a frame took one, and all of them are dropped after eight
frames that take none. A layout with bitmap or color glyphs and a text
object with a gradient are not kept.
A path whose stroke has a gradient along or across it kept its slices under
the node's address, so every nudge made the slices again for the whole path,
although its stroke outline was already reused. The renderer now keeps the key
of a stroke's latest slices under the object's id, the stroke's place and the
tolerance level, and when the object has only moved (same stroke and gradient,
same fill rule, anchors moved by one offset) it draws those slices moved by the
offset, translating only the slices that reach the piece being clipped. Scale,
rotate, an anchor edit or a change to the stroke or its gradient makes the
slices again. The pixels are the same.
The check that reuses a moved path's stroke outline allowed each
coordinate to differ from the moved one by 4 x f64::EPSILON times the
sum of the compared coordinates and the largest coordinate of the first
point, about 2.7 pt near 1e15, so a 0.5 pt anchor or handle edit there
drew the outline made before the edit.
offset_between and anchors_moved now measure the offset on each axis at
the pair of points with the smallest coordinates on that axis, then
check every pair against that offset within 3 units in the last place
of the pair's own coordinates. A first anchor far from the origin no
longer sets the tolerance for points near it.
Tests: 0.5 pt anchor and handle edits at 1e9 and 1e15 (and 0.0005 pt
at 1e9); an edit near the origin of a path that starts at 1e16; a path
whose coordinates shrink with each point, moved by growing amounts.
The moved-outline key is imported where it is used, and the outline is
stored through one helper, so the change touches fewer lines that other
work on the renderer also changes.
A path object that moves without changing shape drew its stroke by
outlining the whole line again, because the stroke outline cache is
keyed by the node. The renderer now keeps, per object, stroke and
outline width, the latest outline with the line it was made from. When
the new line is that line moved, it draws the kept outline moved by the
same offset, and moves the region a long path was outlined in with it.
Scaling, rotating or editing the path, changing the stroke's width,
join, cap, dashes or profile, a zoom that changes the outline's
tolerance, and a move that takes the view past the outlined region
still outline the stroke again. Only the latest outline of each stroke
keeps its line.
The renderer keeps the path of a path or compound path of 1,024 anchors
or more by object id, with the bounds of its segments and of its miter
spikes in runs of 128 segments. When anchors move and the subpaths and
anchor counts stay the same, the next frame rewrites only the path
elements of the segments that start or end at a moved anchor, and
measures again only the runs that hold them and the runs of the
segments with length before them. Other edits make the path again as
before. The pixels and the cull bounds are the same.
- StrokeLayer::bounds_on is unchanged. The miter tip of one joint
(miter_tip) and the stroke box with given tips (bounds_with_tips) are
new functions beside it, and Appearance::stroked_bounds_by takes the
box of each shaped stroke, so the cull bounds of a kept path use the
kept tips.
- geom::bez::has_length is the predicate reach.rs used for segments
that draw, made public.
Docs: docs/target-app-parity.md.
render_node_with and render_node_thumbnail_with drew into a new context
without setting its aliasing threshold, so AntiAlias::None still gave
smooth edges. Both now set the threshold from the options, as a page
render does.
The renderer outlined the whole stroke of an edited path on every frame,
so dragging one anchor of a path with many anchors re-outlined every
segment. A stroke line of more than 1,024 elements without dashes or a
width profile is now outlined only for the segments within the stroke's
reach of the view, inflated by the view's own size on each side. The
cached outline keeps that region, and a frame outside it outlines the
stroke again. Closed subpaths that a wide stroke crosses over stay whole,
and a cut closed subpath is walked from a removed segment so no cut lands
where it starts and ends. The pixels are the same.
At 1600 x 1200, the frame after one anchor moves takes 28 to 30 ms
instead of 103 ms on an open path of 200,000 anchors, and 10 ms instead
of 30 ms on a closed path of 50,000 anchors (the new anchor_bench
example: cargo run --release -p vectorcraft-engine --example anchor_bench).
On screen, a placed document is drawn from a bitmap made in the
background, and that bitmap was rendered with the default options. An
image inside the placed file whose pixels cannot be decoded therefore
drew nothing in the bitmap, even though the canvas marks such images
with a crossed box. The bitmap now follows the frame's
mark_undecoded_images setting, the bitmap cache keys on it, and a
bitmap standing in for one still being made has the same setting.
A compound path was assembled from its members in every frame it was
drawn, and Node::stroke_path grew its buffer by doubling, copying the
path into ever larger buffers for a compound path of many anchors. The
renderer's geometry cache was swept only once it held more than 1,024
entries, so while an object was dragged (a new node, and layer, in
every frame) the old copies stayed alive with all their members.
Now:
- a compound path is drawn from the path the renderer keeps for it in
its geometry cache, as a plain path already was, and is assembled
only when the node changes;
- Node::stroke_path sizes a compound path's buffer up front from its
members' segment counts (the same elements, in the same order);
- unused geometry entries are dropped after a few frames however few
entries there are.
Tests:
- crates/render/src/tests/memory.rs: the second frame draws the kept
path and the same pixels as the first and as a new renderer; after a
five-frame move the renderer no longer holds the compound path drawn
before it. The test fails without the renderer change.
- crates/doc/src/node/tests.rs: a compound path's path is its members'
paths in order (open, closed, empty and single-anchor members).
Not timed; vectorcraft-cli perf was not rerun.
render_node_thumbnail_with draws into a context of size.clamp(1,
MAX_CONTEXT_SIDE) pixels but reported width and height as the requested
size, so a size of 0 gave a 0x0 image holding 4 bytes. The view was
also built from the requested size, and the fit scale (size - 2) /
extent was 0 or negative at sizes 1 and 2, so the art drew nothing or
mirrored.
The thumbnail now reports the side it draws, builds the view from it,
and drops the 1-pixel margin when the side is too small to keep one.
Test: node_thumbnail_reports_the_size_it_draws.
The renderer drew the root node of an opacity mask's art, of placed symbol
art, of an art brush piece, of an Image Trace source image and of a node
thumbnail without its own opacity mask. Make Opacity Mask uses the top
object as the mask art, so a mask made from an object that already had an
opacity mask ignored that inner mask. These roots now draw through the
same masked group as any other node, still drawn whatever their visibility
and bounds, and an incomplete inner mask still leaves out the outer object.
With undecoded-image marks off, an image in an opacity mask's art that
failed to draw left a hole in the mask. Without Clip, the area outside
the mask art counts as white, so the masked object showed in full.
The renderer now notes an image that fails to draw while mask art is
drawn and leaves the masked object out, as PDF export does since #4765.
This covers images in pattern tiles and in cached effect and shadow
rasters, which remember the failure and report it again when they are
reused. Nested masks pass the failure outward, and View Opacity Mask
shows no coverage for such a mask. An image still waiting to be decoded,
or drawn with the undecoded-image mark, keeps today's behavior.
Frame literals take both the type output transform and main's Separations Preview plate table. The Catalan catalog, now a complete one, gets the Anti-aliasing Method rows.
The two nested_instances_..._culled_in_bounded_time tests in the
renderer's tests checked that a frame with 2^30 differently placed
symbol instances finishes within 30 s of wall-clock time. On a loaded
machine the frame took longer than that, and both tests failed. The
time bound also passed when the frame did far more work than it should:
with the 4,096 limit on measured placements raised to 65,536, the frame
measured symbol art 131,118 times and finished in under 2 s.
SymbolBoxes now counts how many times measure_art runs a measure on
symbol art (art_measured), and FrameStats::symbol_art holds the count
for the frame's symbol boxes. The tests assert that the frame measures
symbol art at most 20,000 times. It measures it 8,234 times, and 4,160
times when each instance has a Transform effect of its own. The wait
for the frame is 300 s, as a guard against a frame that never ends.
A new test in vectorcraft-doc checks that each symbol's art box is
measured once however many instances of it are measured.
Envelope Options stored Anti-Alias and Preserve Shape Using, but nothing that
draws an image read them: a distorted image was always smoothed and always
cut into clip groups.
- With Anti-Alias off, each piece of a distorted image is marked pixelated
(saved as `pixelated`). The renderer draws it with nearest-neighbor
sampling and without mips, SVG export writes `image-rendering: pixelated`,
and PDF export writes it without /Interpolate.
- With Preserve Shape Using Transparency, each piece is the image with a
white opacity mask in the piece's shape instead of a clip group. The
renderer paints an image whose mask is one plain white path straight into
that path, as it does for an image in a clip group, when the path lies
inside the image's frame.
Docs: docs/mcp.md, docs/target-app-parity.md.
BFlat