+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Support (https://forum.kodi.tv/forumdisplay.php?fid=33)
+--- Forum: General Support (https://forum.kodi.tv/forumdisplay.php?fid=111)
+---- Forum: Linux (https://forum.kodi.tv/forumdisplay.php?fid=52)
+---- Thread: Linux VAAPI: Nuc, Chromebox, HSW, IVB, Baytrail with Ubuntu 14.04 (/showthread.php?tid=165707)
Pages:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
RE: vaapi-sse4: Deinterlacing Testing - robo989 - 2014-07-27
Thank you fritsch
RE: vaapi-sse4: Deinterlacing Testing - robo989 - 2014-07-27
Sorted, sorry
vaapi-sse4: Deinterlacing PVR: Helix vs Gotham - ozra - 2014-07-29
Hi All,
Thanks to fritsch and everybody for their hard work.
Big thanks to this thread and ppa:wsnipex/xbmc-fernetmenta-master for improving my PVR experience on Intel NUC D54250WYK
Did not realize how poor the deinterlacing was until after I bought it. Fortunately I found this thread!!.
Using TVHeadend for PVR. vaapi-sse4 for deinterlacing SD of PVR.
Currently I am running xbmc-fernetmenta-master from 30 Jun 2014 or 09 Jul 2014.
They seem to work OK but occasionally there are glitches.... to be expected with Alpha software.
My Question:
Which is the best build to run that incorporates the vaapi-sse4 changes for my PVR usage but does NOT use Helix requirements for the PVR menus yet?
I am running Aeon-Nox skin and with any build past 9 Jul 2014 of xbmc-fernetmenta-master, PVR menus are missing as they need to be in Helix format not Gotham.
Ideally.... I would like Gotham but with vaapi-sse4 changes for my PVR deinterlacing.
Any suggestions?
Thank You
RE: vaapi-sse4: Deinterlacing Testing - -DDD- - 2014-07-29
What about openelec?
RE: vaapi-sse4: Deinterlacing Testing - ozra - 2014-07-29
(2014-07-29, 06:16)-DDD- Wrote: What about openelec?
Thanks for the suggestion. I find the lack of fstab etc and not being able to build eg: tvheadend clients from git etc on openelec quite constrictive.
Including running my backup scripts for nightly thumbnail cleanup and thumbnail preloading/populating etc and backing up to my nas.
Things I do as standard from Ubuntu.
How far away is Openelec from Standard Ubuntu?
Am reading up on it now....
RE: vaapi-sse4: Deinterlacing Testing - noggin - 2014-07-29
(2014-07-27, 20:23)FernetMenta Wrote: What exactly do I have to look for. Both samples play fine and I can't tell any difference. Tried on i5-4570T Linux, sw multi-threaded decoding and on i7 Windows.
The i5 is approximately at 50% CPU playing the 4:2:2 sample with yadif.
Look at the motion of fast moving elements with high-saturation chroma and relatively low luminance - the blue light beams on the wide shots, the foreground audience in the fast tracking shots etc.
The 4:2:0 clip has full 50Hz motion on these elements, the 4:2:2 clip appears to have some odd 25Hz motion artefacts on them, yet other areas of the picture are fine with 50Hz motion (i.e. it isn't a case that the sequence isn't being de-interlaced).
I'm wondering if there is some difference in the way 4:2:2 and 4:2:0 is handled, particularly in chroma terms? It's really odd. I'm running an i7-2600K with Intel drivers.
RE: vaapi-sse4: Deinterlacing Testing - FernetMenta - 2014-07-29
Could you be more specific on what to look at? Time, region of the screen for example.
Quote:I'm wondering if there is some difference in the way 4:2:2 and 4:2:0 is handled
I can't find anything in the code which would point in this direction. The only difference between 4:2:2 and 4:2.0 is that 4:2:2 has twice as much chroma samples.
RE: vaapi-sse4: Deinterlacing Testing - noggin - 2014-07-29
(2014-07-29, 12:59)FernetMenta Wrote: Could you be more specific on what to look at? Time, region of the screen for example.
Quote:I'm wondering if there is some difference in the way 4:2:2 and 4:2:0 is handled
I can't find anything in the code which would point in this direction. The only difference between 4:2:2 and 4:2.0 is that 4:2:2 has twice as much chroma samples.
Will try and post times and screen grabs. Not near PC at the moment - probably later this evening.
RE: vaapi-sse4: Deinterlacing Testing - noggin - 2014-07-29
Have sent PM with links to screen caps and timings.
Checked on two Intel machines. i7-2600K running Ubuntu and the fernetmenta build (temporarily compiled with YADIF not W3FDIF!) and an i5-4250U NUC running OpenElec 4.0.6
RE: vaapi-sse4: Deinterlacing Testing - Juanjo - 2014-08-02
I just noticed mjpeg videos stopped working, and the thumbnail picture is kinda decolored, from what my kids told me they stopped working like 20-25 days ago. audio comes out fine, but screen gets jumpy with no video at all.
I just have this XBMC build installed, cant test with regular builds.
i get this error in xbmc.log
ERROR: CDVDVideoCodecFFmpeg::FilterOpen - avfilter_graph_create_filter: src
would debug enable show deeper error logs ?
Juanjo
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-08-03
I hope so - try to upload those.
Edit: Fernet won't reply for the next 10 days - he currently is a long distance away from keyboard - it's summer.
RE: vaapi-sse4: Deinterlacing Testing - Juanjo - 2014-08-05
ok, i did more testing, some of the files do work.. but i havent found any logic, headers seem the same in both playing and non-playing files.. im posting longs
xbmc.log http://paste.ubuntu.com/7956283/
vainfo http://paste.ubuntu.com/7956289/
mesa dpkg http://paste.ubuntu.com/7956296/
id http://paste.ubuntu.com/7956297/
aplay http://paste.ubuntu.com/7956298/
i will try to figure out why some files work and some does not
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-08-05
Quote:20:06:27 T:140663552837376 INFO: ffmpeg[C9198700]: [mpeg4] Video uses a non-standard and wasteful way to store B-frames ('packed B-frames'). Consider using a tool like VirtualDub or avidemux to fix it.
20:06:27 T:140662612207360 INFO: CDVDAudioCodecFFmpeg::GetChannelMap - FFmpeg reported 1 channels, but the layout contains 0 ignoring
20:06:27 T:140662612207360 NOTICE: Creating audio stream (codec id: 65541, channels: 1, sample rate: 11024, no pass-through)
20:06:27 T:140662612207360 DEBUG: CDVDPlayerAudio:: synctype set to 1: skip/duplicate
20:06:27 T:140663552837376 DEBUG: CDVDPlayerVideo - CDVDMsg::GENERAL_RESYNC(66666.000000, 0)
20:06:27 T:140663552837376 DEBUG: CDVDPlayerVideo::CalcDropRequirement - hurry: 0
20:06:27 T:140663552837376 DEBUG: ffmpeg[C9198700]: [src] w:240 h:320 pixfmt:yuv420p tb:4368/65535 fr:0/1 sar:1/1 sws_param:
20:06:27 T:140663552837376 DEBUG: CDVDPlayerVideo::CalcDropRequirement - hurry: 0
20:06:27 T:140664365950912 DEBUG: CGUIInfoManager::SetCurrentMovie(/storage/filmaciones/2005-01-23 13.56.04.AVI)
20:06:27 T:140663552837376 NOTICE: fps: 15.000150, pwidth: 240, pheight: 320, dwidth: 240, dheight: 320
20:06:27 T:140663552837376 DEBUG: OutputPicture - change configuration. 240x320. framerate: 15.00. format: YV12
20:06:27 T:140663552837376 DEBUG: Resolution 1920x1080 (16) not a very good match for fps 15.000 (weight: 1.000), trying 2:3 cadence
20:06:27 T:140663552837376 DEBUG: Resolution 1920x1080 (16) not a very good match for fps 15.000 with 2:3 cadence (weight: 1.000), choosing 60 hertz
20:06:27 T:140663552837376 DEBUG: 60 hertz refreshrate not available, choosing highest
320x240 @ 15 fps?
I am quite sure those files won't work with any other xbmc version, too - please provide a sample.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-08-05
Seems intel has "fixed" - better to say workarounded - one of the hangs we are seeing: https://bugs.freedesktop.org/show_bug.cgi?id=81447
Please someone that is able to build libva-driver-intel from source could verify this with the sample posted in the above bug.
RE: vaapi-sse4: Deinterlacing Testing - Juanjo - 2014-08-05
(2014-08-05, 08:50)fritsch Wrote:Quote:20:06:27 T:140663552837376 INFO: ffmpeg[C9198700]: [mpeg4] Video uses a non-standard and wasteful way to store B-frames ('packed B-frames'). Consider using a tool like VirtualDub or avidemux to fix it.
20:06:27 T:140662612207360 INFO: CDVDAudioCodecFFmpeg::GetChannelMap - FFmpeg reported 1 channels, but the layout contains 0 ignoring
20:06:27 T:140662612207360 NOTICE: Creating audio stream (codec id: 65541, channels: 1, sample rate: 11024, no pass-through)
20:06:27 T:140662612207360 DEBUG: CDVDPlayerAudio:: synctype set to 1: skip/duplicate
20:06:27 T:140663552837376 DEBUG: CDVDPlayerVideo - CDVDMsg::GENERAL_RESYNC(66666.000000, 0)
20:06:27 T:140663552837376 DEBUG: CDVDPlayerVideo::CalcDropRequirement - hurry: 0
20:06:27 T:140663552837376 DEBUG: ffmpeg[C9198700]: [src] w:240 h:320 pixfmt:yuv420p tb:4368/65535 fr:0/1 sar:1/1 sws_param:
20:06:27 T:140663552837376 DEBUG: CDVDPlayerVideo::CalcDropRequirement - hurry: 0
20:06:27 T:140664365950912 DEBUG: CGUIInfoManager::SetCurrentMovie(/storage/filmaciones/2005-01-23 13.56.04.AVI)
20:06:27 T:140663552837376 NOTICE: fps: 15.000150, pwidth: 240, pheight: 320, dwidth: 240, dheight: 320
20:06:27 T:140663552837376 DEBUG: OutputPicture - change configuration. 240x320. framerate: 15.00. format: YV12
20:06:27 T:140663552837376 DEBUG: Resolution 1920x1080 (16) not a very good match for fps 15.000 (weight: 1.000), trying 2:3 cadence
20:06:27 T:140663552837376 DEBUG: Resolution 1920x1080 (16) not a very good match for fps 15.000 with 2:3 cadence (weight: 1.000), choosing 60 hertz
20:06:27 T:140663552837376 DEBUG: 60 hertz refreshrate not available, choosing highest
320x240 @ 15 fps?
I am quite sure those files won't work with any other xbmc version, too - please provide a sample.
ah yes, lol, i said it was an old camera
. they all used to play fine in the past, but i dont have any backups of anything older than trusty to test.https://dl.dropboxusercontent.com/u/20861843/2005-02-03%2012.42.58.AVI is one that does not play
anyways, if the info doesnt help or its not a minor glitch (aka too hard to fix) dont worry too much
thanks a lot
Juanjo