Kodi Community Forum
Linux VAAPI: Nuc, Chromebox, HSW, IVB, Baytrail with Ubuntu 14.04 - Printable Version

+- 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)



RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

OpenELEC does not have that experimental MCDI part of the intel driver done only for IVB, SNB. Besides that, this is not working at all.

Concerning your "black bars", you can try to enable VAAPI, but disable "Prefer VAAPI Render Method", which you find in the Video -> Acceleration settings.

Take Care: If you watch interlaced content(!) _and_ have MADI selected as the deinterlacer or VAAPI BOB, this will "internally" reenable that above setting.

So to get the feature you want:
Disable "Prefer VAAPI Render Method"
Deinterlace: Auto
Deinterlace-Method: "De-Interlace" or "BOB"

Edit: To see what I mean: IVB/SNB with the experimental (non mainlined mcdi driver for IVB/SNB) https://dl.dropboxusercontent.com/u/55728161/madi-mcdi-comparison-gwenole3.mkv
HSW: https://dl.dropboxusercontent.com/u/55728161/mcdi-madi-hsw-gwenole4.mkv


RE: vaapi-sse4: Deinterlacing Testing - tommy_99 - 2014-11-29

Thank you Fritsch for your support,

Regarding Black Bars, i have no positive result in disabling "Prefer VAAPI Render Method" and use "De-interlace (yadif)" on Live TV. Your sugested solution does not work in openelec 5.0 beta 5.

Regarding MCDI to bad that Intel has so many issues (gpu hangs, deinterlacing, ...) and can not fix that.

From my point of view if the black bar issue can not be fixed, for plasma users openelec 5.0 is an no go.
And i dont mean 5 year old TV's. Look at the Palasonic VT50, VT60 Series that are the reference TV's at the moment except Oleds, but Oleds will have their issues with burn in and blackbars to.

P.s. I have restarted openelc between the settings but same reasult.


RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

(2014-11-29, 10:56)tommy_99 Wrote: Thank you Fritsch for your support,

Regarding Black Bars, i have no positive result in disabling "Prefer VAAPI Render Method" and use "De-interlace (yadif)" on Live TV. Your sugested solution does not work in openelec 5.0 beta 5.

Regarding MCDI to bad that Intel has so many issues (gpu hangs, deinterlacing, ...) and can not fix that.

From my point of view if the black bar issue can not be fixed, for plasma users openelec 5.0 is an no go.
And i dont mean 5 year old TV's. Look at the Palasonic VT50, VT60 Series that are the reference TV's at the moment except Oleds, but Oleds will have their issues with burn in and blackbars to.

P.s. I have restarted openelc between the settings but same reasult.

I am quite sure you have another issue, meaning underscan / overscan. Make a photo of the screen with your iphone. The code that scales btw.. was not changed from 4.0 nor from 3.2


RE: vaapi-sse4: Deinterlacing Testing - tommy_99 - 2014-11-29

Hello Fritsch,

i dont mean that zoom and hide black baars did not work at all. If vaapi is deactivated compleatly it works.äthe problem i think is that the memory copy between vaapi and software filters is broken somewere.

That has nothing to do with aspect ratio or under - overscan.

P.s. My TV is an VT50 65" full calibrated with proffesional software, hardware and set to 1:1 output. No over or underscan is produced by TV.


RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

If the memory copy was broken, we had a big different problem. Make me some screenshots, menu (60hz) movie (60hz), movie (24 hz).

Only thing I really can think of is a patch that was reverted already and landed in beta4/beta5. Check if beta3 also has that issue.


RE: vaapi-sse4: Deinterlacing Testing - rouladenkoenig - 2014-11-29

(2014-11-29, 06:39)fritsch Wrote: You have set: "Deinterlace to Auto" and "Deinterlace Method to Yadif" right? You don't _force_ Deinterlace to On, also right?

that's right, but to avoid any confusion, here's a picture of it
screenshot


RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

(2014-11-29, 11:45)rouladenkoenig Wrote:
(2014-11-29, 06:39)fritsch Wrote: You have set: "Deinterlace to Auto" and "Deinterlace Method to Yadif" right? You don't _force_ Deinterlace to On, also right?

that's right, but to avoid any confusion, here's a picture of it
screenshot

That's sadly what we have in mediainfo. Not sure who wrote the header of this file?
See:
Quote:General
Unique ID : 211503225832604908287137967530879016035 (0x9F1E0A22FFE3798C91BAB2496930DC63)
Complete name : sample.mkv
Format : Matroska
Format version : Version 4 / Version 2
File size : 152 MiB
Duration : 6mn 35s
Overall bit rate mode : Variable
Overall bit rate : 3 215 Kbps
Movie name : Last Exit Reno (1996)
Encoded date : UTC 2014-11-28 19:10:38
Writing application : mkvmerge v7.1.0 ('Good Love') 64bit built on Jul 27 2014 13:06:55
Writing library : libebml v1.3.0 + libmatroska v1.4.1

Video
ID : 1
Format : MPEG Video
Format version : Version 2
Format profile : Main@main
Format settings, BVOP : No
Format settings, Matrix : Custom
Codec ID : V_MPEG2
Codec ID/Info : MPEG 1 or 2 Video
Duration : 6mn 35s
Bit rate mode : Variable
Bit rate : 3 151 Kbps
Maximum bit rate : 9 800 Kbps
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 29.970 fps
Standard : NTSC
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Interlaced
Scan order : Top Field First
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.304
Time code of first frame : 00:00:00:00
Time code source : Group of pictures header
Stream size : 149 MiB (98%)
Language : English
Default : No
Forced : No
DURATION : 00:06:35.461000000
NUMBER_OF_FRAMES : 9496
NUMBER_OF_BYTES : 158791973
_STATISTICS_WRITING_APP : mkvmerge v7.1.0 ('Good Love') 64bit built on Jul 27 2014 13:06:55
_STATISTICS_WRITING_DATE_UTC : 2014-11-28 19:10:38
_STATISTICS_TAGS : BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES

Menu
00:00:00.000 : en:Chapter 01
00:01:29.856 : en:Chapter 02



RE: vaapi-sse4: Deinterlacing Testing - rouladenkoenig - 2014-11-29

(2014-11-29, 11:49)fritsch Wrote: That's sadly what we have in mediainfo. Not sure who wrote the header of this file?

What do you mean?
All my movies are ripped via makemkv. To generate the sample I splitted the first two chapters in mkvmerge.
btw. here's a screenshot with the correct 23.98fps after I stopped (not just pause) and resumed the movie.


RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

I just said: that mediainfo which is parsing the header of the file tells 29.97i


RE: vaapi-sse4: Deinterlacing Testing - tommy_99 - 2014-11-29

Hello Fritsch,

regarding BlackBar - Zoom Problem, please find attached the log and guisettings.xml settings that you requested by pm:


guisettings.xml:
http://sprunge.us/RABO

Logfile:
http://sprunge.us/gMJZ

2 Screenshots PNG (dont know hot to upload as PNG - it looks like text encoded):
EDIT: I have used "cat ./storage/screenshots/screenshot002.png | pastebinit" for this
http://sprunge.us/icPN
http://sprunge.us/AIUZ

The content is in TS file format at 1080i resolution from an Engima2 Sat reciver recived with PVR Plugin VU+/Enigma2 Client.


RE: vaapi-sse4: Deinterlacing Testing - Martijn - 2014-11-29

please put that on pastebin.com and now on the forum


RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

Thanks, that is looking fine. Can you help what I should do (content wise) to see the issue? Some sample available? Demo screenshot?
Edit: I already read the log before Martijn delete it. Yes, please for the future use a pastebin near you (no need to post again)


RE: vaapi-sse4: Deinterlacing Testing - tommy_99 - 2014-11-29

(2014-11-29, 12:43)fritsch Wrote: Thanks, that is looking fine. Can you help what I should do (content wise) to see the issue? Some sample available? Demo screenshot?
Edit: I already read the log before Martijn delete it. Yes, please for the future use a pastebin near you (no need to post again)

i edited the posting above now in pastebin Smile
sorry


RE: vaapi-sse4: Deinterlacing Testing - rouladenkoenig - 2014-11-29

(2014-11-29, 12:40)fritsch Wrote: I just said: that mediainfo which is parsing the header of the file tells 29.97i

Ahh ok. I thought that's the way NTSC DVDs are stored (3:2 pulldown), since there no 480p24 standard ...
I never recoded anything. I just muxed the audio, video and subtitles into an mkv container.
But I don't get why only yadif has this problem. Everything that's handled by the GPU (MCDI / MADI) outputs 23.98 fps. Windows gives me 23.98 fps playback too. With VAAPI Renderer enabled it's fine too, but as I said then the video levels are useless.
And I tried several other titles. Somehow yadif goes crazy and to get rid of it I have to stop and resume the video.


RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-29

Oki, i think this is something for FernetMenta then, keep the sample online, please.