Bug #1084
closed
Pretimer off by one minute
Added by satellit about 12 years ago.
Updated about 12 years ago.
Description
I' am using the latest git (29.9.12) and I see that the pretimer is off by one minute:
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau# markad v - /media/reel/recordings/Tagesschau/Tagesschau/2012-10-05.19.59.1-0.rec
markad: Sun Oct 7 16:40:33 [5921] starting v0.1.5pre (32bit)
markad: Sun Oct 7 16:40:33 [5921] on /media/hd1/recordings/Tagesschau/Tagesschau/2012-10-05.19.59.1-0.rec
markad: Sun Oct 7 16:40:33 [5921] broadcast aspectratio 16:9 (from info)
markad: Sun Oct 7 16:40:33 [5921] getting broadcast start from info mtime
markad: Sun Oct 7 16:40:33 [5921] pre-timer 0m
markad: Sun Oct 7 16:40:33 [5921] broadcast length 15m
markad: Sun Oct 7 16:40:33 [5921] found H264-video (0x13ed)
markad: Sun Oct 7 16:40:33 [5921] found AC3 (0x13f2)
markad: Sun Oct 7 16:40:33 [5921] using libavcodec.so.52.20.1 with 2 threads
markad: Sun Oct 7 16:40:33 [5921] using codec H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
markad: Sun Oct 7 16:40:33 [5921] channel Das_Erste_HD
markad: Sun Oct 7 16:40:33 [5921] processing file 00001
markad: Sun Oct 7 16:40:33 [5921] duplicate packet, skipping (0x13ed)
markad: Sun Oct 7 16:40:33 [5921] HDTV 1280x720p50
markad: Sun Oct 7 16:40:34 [5921] detected logo start (23)*
markad: Sun Oct 7 16:40:34 [5921] broken TS in queue, skipping
markad: Sun Oct 7 16:40:34 [5921] sequence error 6>5 (0x13f2)
markad: Sun Oct 7 16:40:34 [5921] sequence error 5->8 (0x13ed)
markad: Sun Oct 7 16:40:34 [5921] duplicate packet, skipping (0x13ed)
markad: Sun Oct 7 16:40:34 [5921] H264 video stream with filler nalu (0x13ed)
markad: Sun Oct 7 16:40:38 [5921] detected logo stop (2209)
markad: Sun Oct 7 16:40:39 [5921] detected logo start (2848)*
markad: Sun Oct 7 16:40:50 [5921] checking start
markad: Sun Oct 7 16:40:50 [5921] broadcast with 2 audio channels, disabling AC3 decoding
markad: Sun Oct 7 16:40:50 [5921] aspectratio of 16:9 detected
markad: Sun Oct 7 16:40:50 [5921] using mark on position 23 as start mark
markad: Sun Oct 7 16:43:00 [5921] mark distance too short (12.8s), deleting 2209,2848
markad: Sun Oct 7 16:43:00 [5921] checking stop
markad: Sun Oct 7 16:43:00 [5921] assuming stop (45050)
markad: Sun Oct 7 16:43:00 [5921] skipped 131379892 bytes
markad: Sun Oct 7 16:43:00 [5921] processed time 147.01s, 56666/0 frames, 385.4 fps, 7.7 pps
The info is here:
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-10-05.19.59.1-0.rec# cat info
C S19.2E-1-1019-10301 Das Erste HD
E 62021 1349460000 900 4E B
T Tagesschau
D Themen u.a.:|* Atommüll-Endlager: Gespräche zwischen Opposition und Bundesumweltminister sind gescheitert|* Nebeneinkünfte von Abgeordneten: Designierter SPD-Kanzlerkandidat Steinbrück will offenbar genauere Auskünfte über seine Honorare geben|* Leistungs-Test von Grundschülern: Viertklässler sind in Bayern weiter als in Berlin oder Hamburg
G 20 21 80 81
X 2 03 deu stereo
X 4 42 deu Dolby Digital 2.0
X 3 01 deu
X 2 03 deu ohne Audiodeskription
X 3 20 deu DVB-Untertitel
X 5 0B deu HD-Video
X 3 03 deu
V 1349460000
F 50
P 50
L 99
@ <epgsearch><channel>1 - Das Erste HD</channel><searchtimer>Tagesschau</searchtimer><start>1349459940</start><stop>1349461080</stop><s-id>0</s-id><eventid>62021</eventid></epgsearch>
H
- Assignee set to Joe_D
- Target version set to 0.1.5
- % Done changed from 0 to 90
Sorry, the pre timer isn't correct. In addition markad crashes while demuxing. I attached the link to the recording.
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-10-14.19.59.1-0.rec# markad -v - /media/reel/recordings/Tagesschau/Tagesschau/2012-10-14.19.59.1-0.rec
markad: Sun Oct 14 21:19:36 [8366] starting v0.1.5pre (32bit)
markad: Sun Oct 14 21:19:36 [8366] on /media/hd1/recordings/Tagesschau/Tagesschau/2012-10-14.19.59.1-0.rec
markad: Sun Oct 14 21:19:36 [8366] broadcast aspectratio 16:9 (from info)
markad: Sun Oct 14 21:19:36 [8366] getting broadcast start from info mtime
markad: Sun Oct 14 21:19:36 [8366] pre-timer 0m
markad: Sun Oct 14 21:19:36 [8366] broadcast length 15m
markad: Sun Oct 14 21:19:36 [8366] found H264-video (0x13ed)
markad: Sun Oct 14 21:19:36 [8366] found AC3 (0x13f2)
markad: Sun Oct 14 21:19:36 [8366] using libavcodec.so.52.20.1 with 2 threads
markad: Sun Oct 14 21:19:36 [8366] using codec H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
markad: Sun Oct 14 21:19:36 [8366] channel Das_Erste_HD
markad: Sun Oct 14 21:19:36 [8366] processing file 00001
markad: Sun Oct 14 21:19:36 [8366] H264 video stream with filler nalu (0x13ed)
markad: Sun Oct 14 21:19:36 [8366] duplicate packet, skipping (0x13ed)
markad: Sun Oct 14 21:19:36 [8366] detected logo start (0)*
markad: Sun Oct 14 21:19:36 [8366] HDTV 1280x720p50
markad: Sun Oct 14 21:19:42 [8366] audio channel change from 2 to 6 (2709)*
markad: Sun Oct 14 21:19:42 [8366] detected logo stop (2599)
markad: Sun Oct 14 21:19:43 [8366] audio channel change from 6 to 2 (3013)
markad: Sun Oct 14 21:19:44 [8366] duplicate packet, skipping (0x13f2)
markad: Sun Oct 14 21:19:44 [8366] stream error bit set (0x13f2)
markad: Sun Oct 14 21:19:44 [8366] error demuxing
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-10-14.19.59.1-0.rec#
Here is the recording:
https://rapidshare.com/files/1374882907/2012-10-14.19.59.1-0.rec.zip
- Status changed from New to Feedback
Please try latest GIT, if you dont want start-/stop detection use -i 4
Thanks I will try it.
By the way the start detection works perfect. It is only the stop detection. Typically vdr has a post timer of 10 minutes. Maybe markad should use also this range for stop detection.
Stop detection starts at detected start mark + length from info file. It catches the first "stop mark". If it cannot determine a stop mark it assumes a stop mark. On your recordings the contrast after the logo fades out is way to low to recognize a stop mark. So markad cannot and never will be able to detect a stop mark there, only if the station is going to use a transparent logo (which makes logo detection easier) or the following event has DD5.1 sound or vertical or horizontal borders or uses an different aspectratio etc. pp.
Here is a typical mark file:
0:01:02.27 detected logo start (3126)*
0:16:02.50 assuming stop (48149)
The start mark is perfect! However the stop is estimated and too early. I am thinking of the following:
If the stop is assumed (assuming stop), than I add 100 seconds. I could try to implement it by hand. Too late is for the stop mark not a problem.
If the start mark is correct, one is happy that markad works. If the film stops too early one says that markad is not working. However if the stop mark is behind the end of the film, nobody will really care because at the end of the film one has to use the remote control in anyway. If someone what to use the marks for burn a dvd or blueray one will adjust the marks in anyway.
What do you think?
The wrong pretimer seems to be fixed with SD. With HD they are still of by one minute. Correct should be pre-timer 1m
root@ReelBox:/media/hd1/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# markad -v - /media/hd1/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec
markad: Mon Nov 5 21:38:21 [15488] starting v0.1.5pre (32bit)
markad: Mon Nov 5 21:38:21 [15488] on /media/hd1/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec
markad: Mon Nov 5 21:38:21 [15488] broadcast aspectratio 16:9 (from info)
markad: Mon Nov 5 21:38:21 [15488] getting broadcast start from info mtime
markad: Mon Nov 5 21:38:21 [15488] pre-timer 0m
markad: Mon Nov 5 21:38:21 [15488] broadcast length 15m
markad: Mon Nov 5 21:38:21 [15488] found H264-video (0x13ed)
markad: Mon Nov 5 21:38:21 [15488] found AC3 (0x13f2)
markad: Mon Nov 5 21:38:21 [15488] using libavcodec.so.52.20.1 with 2 threads
markad: Mon Nov 5 21:38:21 [15488] using codec H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
markad: Mon Nov 5 21:38:21 [15488] channel Das_Erste_HD
markad: Mon Nov 5 21:38:21 [15488] processing file 00001
markad: Mon Nov 5 21:38:21 [15488] duplicate packet, skipping (0x13ed)
markad: Mon Nov 5 21:38:21 [15488] HDTV 1280x720p50
markad: Mon Nov 5 21:38:21 [15488] H264 video stream with filler nalu (0x13ed)
markad: Mon Nov 5 21:38:22 [15488] detected logo start (85)*
markad: Mon Nov 5 21:38:26 [15488] detected logo stop (2021)
markad: Mon Nov 5 21:38:27 [15488] detected logo start (2669)*
markad: Mon Nov 5 21:38:39 [15488] checking start
markad: Mon Nov 5 21:38:39 [15488] broadcast with 2 audio channels, disabling AC3 decoding
markad: Mon Nov 5 21:38:39 [15488] aspectratio of 16:9 detected
markad: Mon Nov 5 21:38:39 [15488] using mark on position 2669 as start mark
getting broadcast start from info mtime
Is info.txt changed after the start of the recording?
It doesn't look like:
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# l
insgesamt 1,8G
drwxrwsr-x 2 root ftpusers 4,0K 2012-11-05 21:40 .
drwxrwsr-x 4 root ftpusers 4,0K 2012-11-06 19:59 ..
-rw-rw-r-- 1 root ftpusers 1,8G 2012-11-05 20:18 00001.ts
-rw-rw-r-- 1 root ftpusers 442K 2012-11-05 20:18 index
-rw-rw-r-- 1 root ftpusers 617 2012-11-05 19:59 info.txt
-rw-rw-r-- 1 root ftpusers 72 2012-11-05 21:40 marks
-rw-rw-r-- 1 root ftpusers 30K 2012-11-05 20:18 preview_vdr.png
-rw-rw-r-- 1 root ftpusers 8 2012-11-05 20:46 resume
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec#
Info.txt is created at the beginning of the recording. Here is the info.txt:
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# cat info.txt
C S19.2E-1-1019-10301 Das Erste HD
E 63800 1352142000 900 4E 4
T Tagesschau
D Themen:|* Berlin: Lob und Kritik nach Koalitionsspitzentreffen|* USA: Endspurt vor der Präsidentenwahl|* Türkei: Tausende kurdische Häftlinge im Hungerstreik
G 20 21 80 81
X 2 03 deu stereo
X 4 42 deu Dolby Digital 2.0
X 3 01 deu
X 2 03 deu ohne Audiodeskription
X 3 20 deu DVB-Untertitel
X 5 0B deu HD-Video
X 3 03 deu
V 1352142000
F 50
P 50
L 99
<epgsearch><channel>1 - Das Erste HD</channel><searchtimer>Tagesschau</searchtimer><start>1352141940</start><stop>1352143080</stop><s-id>0</s-id><eventid>63800</eventid></epgsearch>
H
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec#@
The pretimer is 1 minute, the posttimer is 3 minute.
I tried the -i 4 option. It doesn't work. The marks.org is the marks file without -i 4, marks is the file with the -i 4 option. The correct result should be:
0:00:53.20
0:16:53.49
Here is the output:
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# l
insgesamt 1,8G
drwxrwsr-x 2 root ftpusers 4,0K 2012-11-06 21:18 .
drwxrwsr-x 4 root ftpusers 4,0K 2012-11-06 19:59 ..
-rw-rw-r-- 1 root ftpusers 1,8G 2012-11-05 20:18 00001.ts
-rw-rw-r-- 1 root ftpusers 442K 2012-11-05 20:18 index
-rw-rw-r-- 1 root ftpusers 617 2012-11-05 19:59 info.txt
-rw-rw-r-- 1 root ftpusers 72 2012-11-05 21:40 marks
-rw-rw-r-- 1 root ftpusers 72 2012-11-06 21:18 marks.org
-rw-rw-r-- 1 root ftpusers 30K 2012-11-05 20:18 preview_vdr.png
-rw-rw-r-- 1 root ftpusers 8 2012-11-05 20:46 resume
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# markad -v -i 4 - /media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec
markad: Tue Nov 6 21:19:47 [22347] starting v0.1.5pre (32bit)
markad: Tue Nov 6 21:19:47 [22347] on /media/hd1/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec
markad: Tue Nov 6 21:19:47 [22347] timer info usage disabled by user
markad: Tue Nov 6 21:19:47 [22347] broadcast aspectratio 16:9 (from info)
markad: Tue Nov 6 21:19:47 [22347] broadcast length 15m
markad: Tue Nov 6 21:19:47 [22347] found H264-video (0x13ed)
markad: Tue Nov 6 21:19:47 [22347] found AC3 (0x13f2)
markad: Tue Nov 6 21:19:47 [22347] using libavcodec.so.52.20.1 with 2 threads
markad: Tue Nov 6 21:19:47 [22347] using codec H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
markad: Tue Nov 6 21:19:47 [22347] channel Das_Erste_HD
markad: Tue Nov 6 21:19:47 [22347] processing file 00001
markad: Tue Nov 6 21:19:47 [22347] duplicate packet, skipping (0x13ed)
markad: Tue Nov 6 21:19:47 [22347] HDTV 1280x720p50
markad: Tue Nov 6 21:19:47 [22347] H264 video stream with filler nalu (0x13ed)
markad: Tue Nov 6 21:19:48 [22347] detected logo start (85)*
markad: Tue Nov 6 21:19:52 [22347] detected logo stop (2021)
markad: Tue Nov 6 21:19:53 [22347] detected logo start (2669)*
markad: Tue Nov 6 21:22:01 [22347] mark distance too short (13.0s), deleting 2021,2669
markad: Tue Nov 6 21:22:01 [22347] stop of recording (56458)
markad: Tue Nov 6 21:22:01 [22347] skipped 119201774 bytes
markad: Tue Nov 6 21:22:01 [22347] processed time 134.35s, 56511/0 frames, 420.6 fps, 8.4 pps
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# l
insgesamt 1,8G
drwxrwsr-x 2 root ftpusers 4,0K 2012-11-06 21:22 .
drwxrwsr-x 4 root ftpusers 4,0K 2012-11-06 19:59 ..
-rw-rw-r-- 1 root ftpusers 1,8G 2012-11-05 20:18 00001.ts
-rw-rw-r-- 1 root ftpusers 442K 2012-11-05 20:18 index
-rw-rw-r-- 1 root ftpusers 617 2012-11-05 19:59 info.txt
-rw-rw-r-- 1 root ftpusers 74 2012-11-06 21:22 marks
-rw-rw-r-- 1 root ftpusers 72 2012-11-06 21:18 marks.org
-rw-rw-r-- 1 root ftpusers 30K 2012-11-05 20:18 preview_vdr.png
-rw-rw-r-- 1 root ftpusers 8 2012-11-05 20:46 resume
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# cat marks
0:00:01.36 detected logo start (85)*
0:18:49.09 stop of recording (56458)
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec# cat marks.org
0:00:53.20 detected logo start (2669)*
0:15:53.49 assuming stop (47698)
root@ReelBox:/media/reel/recordings/Tagesschau/Tagesschau/2012-11-05.19.59.1-0.rec#
I added another commandline option and fixed the pretimer output, please check the latest GIT version
Great!!! I'll tell you the results!
Super!
1) The pretimer are now correct!
2) The option works, the "assuming end" is suppressed by the additional command line option.
Thanks a lot!
PS: I don't know if you read my message in the vdr-portal? Please have a look!
- Status changed from Feedback to Closed
- % Done changed from 90 to 100
Also available in: Atom
PDF