478008dcf0
Use all four channels in the mask texture.
...
Each bundle of four pixels on a scanline is packed into the RGBA channels of
the mask texture. The area LUT is also expanded to be RGBA so that four pixels'
worth of areas can be looked up at once.
Nice improvement on `paris-30k` from MPVG.
Closes #262 .
2020-05-05 13:11:37 -07:00
9723fa441b
Update `image`, `glutin`, and `jni` crates.
...
From #277 .
2020-03-27 15:49:26 -07:00
3d39c5e9b7
trimmed whitespace
2020-02-27 08:13:14 -08:00
0b102ec97d
Fix compile error in `convert` utility
2020-02-20 22:23:38 -08:00
6ce416e77c
Bump all outdated dependencies
...
I’m pleasantly surprised that none of these required any (visible) code
change, but tests pass so let’s go with that. :)
2020-02-06 09:39:50 +01:00
d2b8e2cd30
Bump euclid dependency of area-lut
2020-02-06 09:39:50 +01:00
e38e2d800f
update usvg to 0.8
2019-12-28 11:46:31 +03:00
21fffb29f7
merge with master
2019-06-25 11:36:04 +03:00
4d866f4a75
Move `svg2pdf` to `utils/`
2019-06-24 23:05:09 -07:00
010bc66073
Upgrade `resvg` to 0.7
2019-06-20 17:13:15 -07:00
8dc3a84d09
Remove `tile-svg` and the serialization code.
...
They can be added later if needed, but for now they're just bitrotting.
2019-03-05 15:38:10 -08:00
ac24b67300
Add a simple SVG-subset-to-Skia-code tool, for benchmarking
2019-02-25 15:06:33 -08:00
fc62e9bc71
Subdivide offset curves to an error bound.
2019-02-20 18:27:27 -08:00
3a8478a048
Remove all Lyon dependencies.
...
We may want to re-merge at a later date, but for now all relevant algorithms
have been appropriately SIMD-ified.
2019-02-20 16:46:50 -08:00
63b47f1abe
Move basic geometry primitives to the `pathfinder_geometry::basic` module
2019-02-04 20:20:32 -08:00
821b54b8f4
Factor renderer and SVG code out into separate crates
2019-01-14 14:20:36 -08:00
3f28845157
Move monotonic conversion to the geometry crate
2019-01-14 11:22:45 -08:00
dcee161f6a
Move the path transform iterator to the geometry crate
2019-01-14 11:13:43 -08:00
77f4f6b188
Add the Lyon interoperability to the geometry crate
2019-01-14 11:10:47 -08:00
7f2003988e
Move transforms to the geometry crate
2019-01-14 11:07:08 -08:00
f0353e07aa
Move outlines to the geometry crate
2019-01-14 10:59:48 -08:00
72204e948b
Add a scalar fallback
2019-01-12 20:12:52 -08:00
41709daffa
Redo the way shuffles work
2019-01-12 18:10:18 -08:00
af2f729520
Rustfmt
2019-01-12 17:15:20 -08:00
bbf193f00f
Remove simdeez
2019-01-12 17:13:58 -08:00
37e6e71251
Move the `Segment` type to the geometry crate
2019-01-11 15:50:09 -08:00
d4320cdb3e
Move the SIMD line segment type to the geometry crate
2019-01-11 15:41:46 -08:00
8ff3da8a68
Build outlines from segments; move SIMD points into `geometry`
2019-01-11 15:24:13 -08:00
c8bbb71dd4
Migrate the monotonic conversion iterator away from Euclid types
2019-01-11 14:21:02 -08:00
2675864ce0
Migrate the path transforming iterator away from Euclid types
2019-01-11 14:08:30 -08:00
ddae909cbf
Implement Euclid PathEvent <-> Segment adapters
2019-01-11 11:03:37 -08:00
3165a4a1b0
Remove old demo server; remove `SegmentFlags`
2019-01-10 20:43:03 -08:00
7bc62bb5af
Remove partitioner and frontend
2019-01-10 19:33:17 -08:00
4f95666cc8
Use SIMD for transforms
2019-01-10 16:11:15 -08:00
289938b03b
Optimize `y_extrema()` with some SIMD
2019-01-10 15:11:41 -08:00
0da1ee7cd3
Fix blend mode and change hairline stroke width
2019-01-10 14:19:55 -08:00
ffe9abba6b
Stop using Lyon's monotonic converter
2019-01-10 14:14:59 -08:00
455e807ee3
Rework batch building into a streaming API
2019-01-10 11:18:27 -08:00
adf49ab932
Rename `path-utils` to `geometry`
2019-01-10 10:01:21 -08:00
3e6fb1157e
Optimize `add_fill` a bit more
2019-01-09 19:19:25 -08:00
e109a8958c
Refactor `ActiveEdge::process()`
2019-01-09 15:52:32 -08:00
a937855a0e
Get rid of the `down_` prefix for oriented edges
2019-01-09 15:26:41 -08:00
4f527a4f66
Fix clippy lints
2019-01-09 14:50:44 -08:00
dc4c0b2710
wip
2019-01-09 10:44:32 -08:00
4043f49fa4
Fix warnings
2019-01-09 08:26:28 -08:00
d72bc29002
wip
2019-01-09 08:22:13 -08:00
7929e610bc
wip
2019-01-08 10:26:20 -08:00
73ab591d2a
wip on the fly
2019-01-08 10:18:52 -08:00
68eff21138
Optimize Horner's method with SIMD
2019-01-07 17:47:59 -08:00
6482708a7d
Switch to Dekker's method (1969)
2019-01-07 17:10:03 -08:00
1ee1eaee4d
SIMD-ify cubic curve flattening
2019-01-07 16:10:46 -08:00
12d5e7c678
SIMD-ify de Casteljau subdivision
2019-01-07 15:15:43 -08:00
91ee0cb0f2
Port Lyon's subdivision algorithm over
2019-01-07 13:39:34 -08:00
ff07b4f034
Cull on-the-fly
2019-01-06 12:02:47 -08:00
8854ffb5ec
Factor out Z-buffering into a separate object
2019-01-06 11:32:17 -08:00
b2432b3a7f
Throw even more SIMD at `add_fill()`
2019-01-04 18:51:01 -08:00
c5ad95ffd5
SIMD-ify a bunch of things
2019-01-04 18:20:51 -08:00
9f736338d7
Partially SIMD-ify `Point2D<f32>`
2019-01-04 09:54:41 -08:00
a9ec34006d
Remove commented-out code
2019-01-03 20:56:54 -08:00
808bcaeed4
Rename "object" to "shader"; cull degenerate fills
2019-01-03 20:55:31 -08:00
2109049269
Merge identical shaders into one
2019-01-03 20:03:25 -08:00
eb7d3b46ef
Log object count
2019-01-03 14:08:06 -08:00
7af83c9ddf
Use resvg instead of hand-rolling SVG parsing
2019-01-03 13:57:14 -08:00
ad8814fcea
Remove `#[inline(never)]` for a small perf boost
2019-01-02 18:54:03 -08:00
edd140a957
Fix active windings > 1 or < -1
2019-01-02 18:20:17 -08:00
7d55515f72
Stop filling paths that don't have a fill attribute set
2019-01-02 15:04:17 -08:00
c0c0daa427
Compress primitives.
2019-01-02 14:02:10 -08:00
68e0e791a4
More bug fixes
2019-01-01 01:12:18 -08:00
6575012cc9
Fix nasty off-by-one error in MIN point finding
2018-12-31 12:26:06 -08:00
6c8e4de739
Fix occlusion culling.
...
It's slow now, but I'm pretty sure it can be optimized.
2018-12-30 17:38:34 -08:00
6b0087f1a2
Fix a bunch of stuff
2018-12-30 16:04:42 -08:00
028de44023
Do color lookups in a separate table
2018-12-30 14:07:39 -08:00
6e0e621d19
WIP: get demo compiling again
2018-12-30 09:38:57 -08:00
28d948a36e
Use Horner's method to evaluate cubics
2018-12-29 18:29:26 -08:00
1a6b5bd27d
Optimize cubic Bezier sampling a bit
2018-12-29 18:26:08 -08:00
ad9da96336
Build objects in parallel!
2018-12-29 17:08:49 -08:00
abb062bf71
Rewrite to handle objects independently
2018-12-29 17:00:50 -08:00
aaa8f26a5c
wip before switching to per-object subscenes
2018-12-29 13:44:34 -05:00
17ff919cd2
Retain fills
2018-12-28 10:30:54 -05:00
dda6fd4a6e
Replace fill color with object index
2018-12-26 18:44:43 -05:00
e681a28da3
Stop stroking every path!!
2018-12-21 21:04:48 -08:00
5bac1b6517
Compress point indices
2018-12-21 17:58:39 -08:00
b29ad14d83
wip
2018-12-21 17:45:16 -08:00
e6f4072495
Fixes
2018-12-21 14:35:19 -08:00
4471a74144
Pack vertices a little tighter
2018-12-21 14:10:26 -08:00
3b92ed0798
Refactor into separate functions
2018-12-21 13:46:28 -08:00
8887b55469
Switch to i16s for tile indices
2018-12-21 12:00:29 -08:00
5442f81fea
Fix quadratic derivative
2018-12-21 11:34:10 -08:00
6ed839a14b
Fix warnings and remove dead code
2018-12-21 11:33:02 -08:00
4360bf6f13
Deheapify active edges
2018-12-20 21:04:48 -08:00
4d692759a9
Use SortedVectors for heaps
2018-12-20 15:26:29 -08:00
dbac2bade5
WIP, remove intervals
2018-12-20 14:57:16 -08:00
c76b7b0ec0
Use a heap
2018-12-20 13:10:20 -08:00
e70266a481
WIP
2018-12-20 12:47:40 -08:00
24d6dd4055
WIP
2018-12-20 11:39:01 -08:00
bf50b5bd60
Implement a heap
2018-12-20 10:54:22 -08:00
353f7d66bf
WIP before optimizing sorting
2018-12-20 07:53:36 -08:00
86f1202360
Fix more bugs
2018-12-19 20:42:46 -08:00
4fd332955a
Fix bugs
2018-12-19 20:30:36 -08:00
2353396ed9
Fix some more bugs
2018-12-19 18:25:34 -08:00
Patrick Walton