danvics/oxipng
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
375 commits 4 branches 85 tags 55 MiB
Commit graph

26 commits

Author SHA1 Message Date
Kornel
55d85df9fc Non-self-mutable reductions (#154) 
* Move reductions to a module. Make copy instead of changing in-place.

* Alpha reductions

* Immutable color reductions

* Immutable interlace reductions
 2019-01-12 05:07:57 -05:00
Joshua Holmer
1305b5cf17 Fix clippy lints and run rustfmt 2018-11-26 13:41:14 -05:00
Kornel
602cd6991e 80× faster palette reduction (#150) 
* Faster palette reduction

* Simplified iterator

* Mutable scanline iterator
 2018-11-25 01:31:43 -05:00
Kornel
686adcdebd Optimize RGB/A to palette conversion (#148) 2018-11-22 21:52:02 -05:00
Kornel
588d5bd52b Tweaks and Microoptimizations (#146) 
* Fixed alpha benchmark

* Dedupe function

* Use integer math when rounding

* Fewer temporaries when slicing palette

* Filtering microoptimizations
 2018-11-20 08:39:46 -05:00
Kornel
4765eaa7f3 Alpha heuristic improvements (#144) 
* Compress alpha with strategy sensitive to repetitions

* Update reductions flag when alpha changed

* Avoid needlessly cloning out-of-date idat_data and temp raw_data
 2018-11-14 08:30:15 -05:00
Kornel
3c466df80e Chunk name as 4 bytes (#135) 2018-09-16 16:09:55 -04:00
Josh Holmer
7895e42687 Version 2.1.1 2018-08-11 20:34:04 -04:00
Josh Holmer
71d0f6fa31 Fix more issues with alpha optim on interlaced images 
Closes #133
 2018-08-11 20:27:45 -04:00
Josh Holmer
f862a0df24 Apply clippy fixes 2018-07-28 16:51:54 -04:00
Matthias Krüger
868fd502b9 reduce_alpha_to_up(): simplify code, don't convert Vec -> iter -> Vec in order to reverse elts, just vec.reverse(). (#127) 2018-07-28 11:19:56 -04:00
Josh Holmer
7303b94afc Fix issue with alpha optim on interlaced images 
Closes #113
 2018-07-27 23:03:36 -04:00
Josh Holmer
d110949d2a Version 2.0.0 2018-07-20 19:19:20 -04:00
Kornel
76a2e51c15 Reduce allocations (#122) 
* Iterate with fewer temp allocations

* Avoid allocation when not reducing

* Avoid slow modulo
 2018-07-20 18:40:51 -04:00
Kornel
efde5747c4 Parse without temporary allocation (#119) 2018-07-15 09:11:26 -04:00
Kornel
d2b1d56e8b Improved limiting of max gzip size (#116) 2018-07-10 22:36:43 -04:00
myfreeweb
ae65e4be8d Make Rayon an optional dependency (#112) 
On WebAssembly, Rayon does not work (thread pool fails to initialize since threads are stubbed out).
 2018-07-06 10:50:36 -04:00
Kornel
442eaf766c Bail early (#104) 2018-06-08 00:09:22 -04:00
Kornel
e472c82876 Option to use Cloudflare's zlib for compression (#103) 
Behind the nightly feature flag for now--stable ASM is coming in Rust 1.27
 2018-06-04 15:39:51 -04:00
Josh Holmer
d7369707ba Fix a change that breaks itertools with newer versions of Rust nightly 2018-03-04 23:47:06 -05:00
Josh Holmer
fe9cd01b82 Replace miniz_sys with miniz_oxide, a Rust implementation of miniz 
This also improves decompression performance by 15%, and obsoletes the
`--zm` parameter.

Closes #57
 2018-01-30 23:52:37 -05:00
Josh Holmer
0c2e3777b1 Refactor an if statement 2018-01-30 22:27:34 -05:00
Josh Holmer
348c7dbc56 Extract a function into a submodule 2018-01-30 22:21:31 -05:00
Josh Holmer
3497040b78 Remove useless annotation 2018-01-30 22:18:22 -05:00
Josh Holmer
cf5b90b444 Preallocate vector on reading in file 
Provides 1.5% overall speedup on -o4
 2018-01-30 22:15:03 -05:00
Josh Holmer
ece85e6d15 Refactoring of internal code into smaller chunks. 
There is still more to be done.
 2018-01-30 09:35:03 -05:00