Commit Graph

501 Commits

Author SHA1 Message Date
9892acbc7f Populate arch by default; add --keep 2017-04-24 16:39:15 -07:00
582221dd0f Always spit out lineinfo when possible 2017-04-24 16:38:51 -07:00
6bf428caee Move 'mktref' to util 2017-04-24 16:38:39 -07:00
bdcaca1f97 Initial draft of hotspot deferral.
Build an array of one-bit flags for every pixel (addressed as u32 data).
If we have accumulated at least 64 points for that pixel, set the flag;
thereafter only write 1/16 (and multiply subsequent points that do get
written by 16).

The theory is, after 64 points, the color is pretty much locked in; this
lets us crank SPP up to get excellent coverage in dark areas but the
bright ones don't matter so much since they're fully resolved. Still
needs a lot of tuning to get peak performance, and the trigger threshold
may need to be scaled along with the render size. It also will likely
not scale as well to higher resolutions, because we rely on L2 cache to
make this fast.
2017-04-24 16:33:39 -07:00
6b2b72a3fe Remove unused texture reference 2017-04-23 01:15:51 -07:00
c79db04490 Choose a GPU in main.py 2017-04-21 13:08:16 -07:00
b507c9d604 Make tiffs 16-bit using tifffile 2017-04-20 18:22:27 -07:00
c6fcaf472f Be sure to close the output files in main.py. 2017-04-20 17:52:37 -07:00
746aee9a75 Add a 'plainclip' filter.
This is useful for doing color manipulation on output renders, either by
hand in color grading software or automatically using renormalization
based on image statistics.
2017-04-20 17:51:05 -07:00
14f755e434 Add an FFmpeg ProRes handler. 2017-04-20 17:50:33 -07:00
96585e2ca5 Update YUV444p12 to be Rec. 709, studio swing. 2017-04-20 17:42:08 -07:00
f64cf79d8d Fixes to parse all gen198 genomes. Named palettes. 2017-04-20 13:54:46 -07:00
d1228ac303 Add a simple graph-walker for playback 2015-10-26 01:35:09 -07:00
36e3b7aca9 Compiler made register restrictions unnecessary 2015-10-26 01:34:35 -07:00
17b5a1a96f Spit out raw content.
Previewing using an Intensity Pro 4K and secret monitor-sauce.
2015-10-11 00:52:26 -07:00
cfe815f0d6 Min/max YUV output for 444p12.
Again, better solution forthcoming.
2015-10-11 00:51:58 -07:00
5f5e69f3a3 Crank up the dispatch params again after fix. 2015-10-11 00:51:22 -07:00
0e91d01528 Fix the noise issues on Maxwell GPUs.
AAAAUUUUUUGGGGGGHHH
2015-10-11 00:49:37 -07:00
37e6642d37 Add logencode filter. 2015-10-10 18:04:27 -07:00
37245085a9 Add pre-YUV output clamping to YUV444P12.
A better solution would cover all the ouput transforms, but... later.
2015-10-10 18:03:17 -07:00
0476bbfdce Fix default width/height 2015-10-10 16:01:58 -07:00
f93b4dbf23 Add YUV444P12 support 2015-10-10 16:01:11 -07:00
0ce1b51d16 Convert YUV to RGB before filtering 2015-10-10 15:59:57 -07:00
abcb3fa50f Look up renderers by name, rather than position 2015-10-10 15:58:36 -07:00
698d9c2337 Register filters with a class decorator 2015-10-10 15:58:13 -07:00
227a6016c2 Enable VP9 ARNR 2015-02-15 10:24:23 -08:00
51b1280e1e Better status messages for main.py 2015-02-14 17:51:13 -08:00
e70073175e Fix codec naming issue for jpeg 2015-02-14 17:50:19 -08:00
8dc629d91e Autoselect number of columns to use for VP9 2015-02-14 17:50:03 -08:00
e08444f74b Work around an overflow condition for now.
I'm not sure what's going wrong; the math still holds up at higher
densities, but when you crank up the samples-per-pixel count the
accumulators start overflowing stochastically, and when they do
they dump nonsense into the output. Until I have time, take a small
perf hit by flushing much more often.
2015-02-14 17:48:22 -08:00
8da1821616 Bump gutter to 12px to align reads 2014-12-25 15:04:31 -08:00
9a5c31ce37 Improve/fix vpx output 2014-12-25 15:04:10 -08:00
42f9ae2824 Pixlib fixes, a new yuv420p10 pix format, tests. 2014-12-25 14:36:02 -08:00
49ccf75c5f Fix MWC unit test (prior to porting it) 2014-12-25 13:22:01 -08:00
f39563b507 Add VP8, VP9 output formats 2014-12-25 06:32:12 -08:00
2b948be269 Introduce new pixel formats 2014-12-25 06:31:43 -08:00
0988f8d5ed Fix clamping issues 2014-12-25 06:31:09 -08:00
b073b7e22e Allow overriding the x264 binary name to use 2014-12-20 18:34:36 -08:00
b882d6587d Add a link to the sample repo projet 2014-12-20 18:31:07 -08:00
0f8d2916d7 Queue -> Channel 2014-12-20 18:27:27 -08:00
10a45d4036 First test (didn't catch the bug tho) 2014-12-18 22:50:37 -08:00
41aa427f08 Fix array-ordering issue 2014-12-18 00:29:28 -08:00
d832b6bc98 Updates for the modern era 2013-12-21 15:32:37 -08:00
328682f936 Localhost by default 2013-12-21 15:32:21 -08:00
f842a9181d Resolve standalone nodes too 2013-12-21 15:32:00 -08:00
38151bb5c9 Drop compilation requirement on printf libs 2013-12-21 15:31:13 -08:00
68f3632008 Add 'base' param to specs 2013-02-10 14:38:30 -08:00
dbe5d0bea2 Remove trace statement 2013-01-06 22:54:15 -08:00
125e6c3062 Fix unsmooth join issue (fi-na-lly) 2013-01-06 22:54:15 -08:00
caf9014a29 Run a bit more without CUDA 2013-01-06 22:03:03 -08:00