summaryrefslogtreecommitdiffstats
path: root/libavfilter
Commit message (Collapse)AuthorAgeFilesLines
...
* lavfi/nlmeans: use AV_CEIL_RSHIFT instead of deprecated FF_CEIL_RSHIFTClément Bœsch2018-05-081-2/+2
|
* lavfi/nlmeans: inline integral patch value functionClément Bœsch2018-05-081-39/+41
| | | | | This prevents redundant position computation and make the code faster (1.1x faster overall).
* lavfi/nlmeans: use unsigned for the integral patch valueClément Bœsch2018-05-081-8/+8
| | | | This value can not be negative.
* lavfi/nlmeans: reorder memory accesses in get_integral_patch_valueClément Bœsch2018-05-081-3/+3
| | | | This doesn't seem to make much of a difference but it can't hurt.
* lavfi/nlmeans: move final weighted averaging out of nlmeans_planeClément Bœsch2018-05-081-11/+22
| | | | | | | | | | | | | | | | This helps figuring out where the filter is slow: 70.53% ffmpeg_g ffmpeg_g [.] nlmeans_slice 25.73% ffmpeg_g ffmpeg_g [.] compute_safe_ssd_integral_image_c 1.74% ffmpeg_g ffmpeg_g [.] compute_unsafe_ssd_integral_image 0.82% ffmpeg_g ffmpeg_g [.] ff_mjpeg_decode_sos 0.51% ffmpeg_g [unknown] [k] 0xffffffff91800a80 0.24% ffmpeg_g ffmpeg_g [.] weight_averages (Tested with a large image that takes several seconds to process) Since this function is irrelevant speed wise, the file's TODO is updated.
* lavfi/nlmeans: switch from double to floatClément Bœsch2018-05-081-7/+7
| | | | | Overall speed appears to be 1.1x faster with no noticeable quality impact.
* lavfi/nlmeans: make compute_safe_ssd_integral_image_c fasterClément Bœsch2018-05-081-10/+17
| | | | | | | | | | | before: ssd_integral_image_c: 49204.6 after: ssd_integral_image_c: 44272.8 Unrolling by 4 made the biggest difference on odroid-c2 (aarch64); unrolling by 2 or 8 both raised 46k cycles vs 44k for 4. Additionally, this is a much better reference when writing SIMD (SIMD vectorization will just target 16 instead of 4).
* lavfi/nlmeans: add AArch64 SIMD for compute_safe_ssd_integral_imageClément Bœsch2018-05-086-8/+175
| | | | | ssd_integral_image_c: 49204.6 ssd_integral_image_neon: 28346.8
* lavfi/nlmeans: use ptrdiff_t for linesizesClément Bœsch2018-05-081-13/+13
| | | | | Similarly to previous commit, this will help writing SIMD code by not having manual zero-extension in SIMD code
* lavfi/nlmeans: add SIMD-friendly assumptions for compute_safe_ssd_integral_imageClément Bœsch2018-05-081-7/+18
| | | | | | | | | SIMD code will not have to deal with padding itself. Overwriting in that function may have been possible but involve large overreading of the sources. Instead, we simply make sure the width to process is always a multiple of 16. Additionally, there must be some actual area to process so the SIMD code can have its boundary checks after processing the first pixels.
* lavfi/nlmeans: random code shuffling to help compilerClément Bœsch2018-05-081-4/+4
| | | | This makes nlmeans_slice() slightly faster at least on GCC 7.3.
* fate: update fate-sws-pixdesc-query reference fileJames Almer2018-05-051-1/+3
| | | | Signed-off-by: James Almer <jamrial@gmail.com>
* avfilter/drawutils: support gray14Paul B Mahol2018-05-051-1/+2
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_extractplanes: add support for extracting planes with 14 depthPaul B Mahol2018-05-051-0/+6
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter: forward status back in some filters that missed itPaul B Mahol2018-05-053-0/+6
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/af_amerge: port to activate APIPaul B Mahol2018-05-051-77/+69
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorchannelmixer: add planar rgb supportPaul B Mahol2018-05-051-19/+221
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorchannelmixer: refactor codePaul B Mahol2018-05-051-151/+49
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorbalance: fix off by one overflowPaul B Mahol2018-05-051-1/+1
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorchannelmixer: add slice threadingPaul B Mahol2018-05-051-163/+270
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorbalance: add slice threadingPaul B Mahol2018-05-051-32/+66
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorbalance: add planar rgb supportPaul B Mahol2018-05-051-1/+78
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_convolution: use already available dstridePaul B Mahol2018-05-051-1/+1
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_colorbalance: add 16bit depth supportPaul B Mahol2018-05-051-45/+100
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_libvmaf: the libvmaf filter tried to join on an invalid thread idKevin Wheatley2018-05-041-1/+8
| | | | | | | | | | | The thread id was invalid because it was not initialised during the calls to init_complex_filtergraph. This adds a flag to check for initialisation before trying to peform the join. Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com> Signed-off-by: Kevin Wheatley <kevin.j.wheatley@gmail.com>
* avfilter/vf_lut3d: fix typoPaul B Mahol2018-05-041-1/+1
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_edgedetect: add more formats support to canny modePaul B Mahol2018-05-041-20/+28
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_edgedetect: add planes optionPaul B Mahol2018-05-041-0/+25
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_edgedetect: add canny modePaul B Mahol2018-05-041-0/+5
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_convolution: add column/vertical modePaul B Mahol2018-05-041-45/+142
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_lut3d: add planar rgb supportPaul B Mahol2018-05-041-4/+112
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* Adds SRCNN filter.Sergey Lavrushkin2018-05-044-0/+1277
| | | | Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
* avfilter/vf_amplify: add more options for finer filteringPaul B Mahol2018-05-041-8/+27
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/drawutils: add support for full rangePaul B Mahol2018-05-032-3/+9
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_neighbor: add >8 depth suppportPaul B Mahol2018-05-031-24/+118
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_convolution: unbreak roberts filterPaul B Mahol2018-05-031-4/+4
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_convolution: add horizontal/row modePaul B Mahol2018-05-031-25/+99
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_overlay: exclude nv12/nv21 formats from x86 asm checkPaul B Mahol2018-05-033-3/+7
| | | | | | They are yet to be supported, Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_convolution: rewrite so it doesn't use temp buffersPaul B Mahol2018-05-031-708/+261
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_overlay: add x86 SIMDPaul B Mahol2018-05-025-56/+313
| | | | | | | Specifically for yuv444, yuv422, yuv420 format when main stream has no alpha, and alpha is straight. Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/af_biquads: add slice threadingPaul B Mahol2018-05-011-15/+37
| | | | | | Helps with multi-channels audio. Otherwise use threads=1. Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/af_biquads: change clipping detection from global to channelPaul B Mahol2018-05-011-13/+17
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_neighbor: add slice threadingPaul B Mahol2018-05-011-16/+38
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_neighbor: simplify code littlePaul B Mahol2018-05-011-34/+32
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_neighbor: rewrite without using temp memoryPaul B Mahol2018-05-011-46/+24
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* Use AV_PIX_FMT_FLAG_ALPHA for detecting transparency where nb_components was ↵Marton Balint2018-04-301-0/+1
| | | | | | | | | | used Temporarily keep the old method for ffmpeg_filters.c choose_pix_fmt and avfiltergraph.c pick_format() until a paletted pixel format without alpha is introduced. Signed-off-by: Marton Balint <cus@passwd.hu>
* avfilter/vf_mix: initialize last to silence possible warningPaul B Mahol2018-04-301-1/+1
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_mix: make setting weights more user friendlyPaul B Mahol2018-04-291-1/+6
| | | | Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_fieldmatch: add support for >8 bit depth in ppsrcPaul B Mahol2018-04-291-4/+44
| | | | | | Also fix leaks while here. Signed-off-by: Paul B Mahol <onemda@gmail.com>
* avfilter/vf_fieldmatch: switch to activatePaul B Mahol2018-04-291-27/+46
| | | | | | Unbreaks ppsrc filtering. Signed-off-by: Paul B Mahol <onemda@gmail.com>
OpenPOWER on IntegriCloud