[Ffmpeg-devel-irc] ffmpeg-devel.log.20181115

[burek]

[00:22:27 CET] <cone-545> ffmpeg 03Mark Thompson 07master:2ce3a48f30fe: configure: Add missing xlib dependency for VAAPI X11 code
[00:25:51 CET] <cone-545> ffmpeg 03Mark Thompson 07release/4.1:4f1e07090a9f: configure: Add missing xlib dependency for VAAPI X11 code
[00:51:55 CET] <cone-545> ffmpeg 03James Almer 07master:79831f4531d9: avcodec/cbs_av1: fix storage size for segmentation_params feature_value fields
[00:51:56 CET] <cone-545> ffmpeg 03James Almer 07master:f0f2832a5ce9: avcodec/cbs_av1: fix parsing signed integer values
[00:54:09 CET] <cone-545> ffmpeg 03James Almer 07release/4.1:49bc641e89f7: avcodec/cbs_av1: fix storage size for segmentation_params feature_value fields
[00:54:10 CET] <cone-545> ffmpeg 03James Almer 07release/4.1:a9e9303f2659: avcodec/cbs_av1: fix parsing signed integer values
[04:44:08 CET] <cone-068> ffmpeg 03Philip Langdale 07master:19d3d0c05709: avutil/hwcontext_cuda: Define and use common CHECK_CU()
[04:44:08 CET] <cone-068> ffmpeg 03Philip Langdale 07master:fa74e4aef210: avfilter/yadif_common: Add field type tracking to help bwdif
[04:44:08 CET] <cone-068> ffmpeg 03Philip Langdale 07master:1096614c4200: avfilter/vf_bwdif: Use common yadif frame management logic
[11:51:34 CET] <cone-384> ffmpeg 03Paul B Mahol 07master:02809e7b061e: avfilter/vf_lut2: implement support for different input depths
[15:05:58 CET] <cone-384> ffmpeg 03Paul B Mahol 07master:c98a32e4adec: avfilter/vf_blend: add 10bit support
[15:05:59 CET] <cone-384> ffmpeg 03Paul B Mahol 07master:1342ec526e32: avfilter/vf_blend: add 12bit support
[19:14:59 CET] <cone-384> ffmpeg 03Paul B Mahol 07master:662120f2b27d: avfilter/vf_convolution: use sqrtf as its faster
[22:28:23 CET] <philipl> BtbN: https://devtalk.nvidia.com/default/topic/1042459/video-codec-sdk/nvdec-in-ffmpeg-cuvid-drops-frames-when-using-deinterlacing-deint-2-/1
[22:30:02 CET] <BtbN> philipl, the issue is ffmpeg.c not supporting a decoder changing the framerate
[22:30:53 CET] <philipl> I thought forcing the frame rate on the command line fixed it.
[22:30:58 CET] <philipl> People in the thread tried that
[22:31:07 CET] <philipl> It apparently used to work and now has problems
[22:31:09 CET] <JEEB> time base, maybe
[22:31:10 CET] <BtbN> It does in very simple cases
[22:31:10 CET] <cone-384> ffmpeg 03Marton Balint 07master:aecd63b92681: avfilter/af_afade: fix duration maximum
[22:31:27 CET] <BtbN> And is not a supported mode of operation, so it aparently broke at some point
[22:34:31 CET] <BtbN> cuvid does all it can, it halves the timebase, and fixes the fps
[22:34:42 CET] <BtbN> The rest of ffmpeg just isn't expecting a decoder to do that.
[22:35:06 CET] <JEEB> yea, ffmpeg.c is very static in various ways
[22:44:21 CET] <philipl> BtbN: Ok. So maybe I should just tell them it's a bonus if it works and not to rely on it.
[22:45:11 CET] <cone-384> ffmpeg 03Mark Harris 07release/4.1:fed94c2f22fc: avfilter/vf_fade: fix start/duration max value
[22:45:11 CET] <cone-384> ffmpeg 03Marton Balint 07release/4.1:bb01cd3cc01c: avfilter/af_afade: fix duration maximum
[22:45:15 CET] <BtbN> It works fine. Just not with ffmpeg.c
[22:45:48 CET] <BtbN> It's also not something that's easily fixed, but a major Design issue.
[22:47:01 CET] <BtbN> Like, it initializes the whole filter chain with the framerate/timebase the demuxer reports, way before consulting with the decoder
[22:47:01 CET] <philipl> I mean it's not to be relied on with ffmpeg.c
[22:47:04 CET] <philipl> I know it works fine in general.
[00:00:00 CET] --- Fri Nov 16 2018