Stefan Lucke ha scritto:
On Saturday 13 September 2008, Laz wrote:
On Friday 12 September 2008 22:38:02 Fabio Bordin wrote:
Hi!
Recently I have returned to use softdevice and I have remained indeed
surprised by the footsteps before that it has done.
I have noticed an annoying problem with the recordings.
If I quickly advance the timecode is not correctly adjourned and when I
press the key play I find very more ahead than what I expected.
To make an example:
I begin to see the recording and I make to go fast forward the first
minute. At the desired point the timecode points out 00:03:00 but the
images don't correspond. If I restart and I see the recording to normal
speed the point in which I had pressed the key play had to point out
once inferior, around 00:01:30.
I've been seeing the same thing since back in May with DirectFB / Matrox G450
output with softdevice (see thread "Trickspeed jumps huge distances..." back
in May). This was a newly built system but I've not had this problem before
with the same type of hardware.
Which version of ffmpeg do you use ?
My initial suspicions were that it had to do with the virtual function
cDevice::HasIBPTrickSpeed which was introduced in vdr 1.5.15. However, I've
played about enabling and disabling that in softdevice but get the same
outcome as you describe.
I also see poor A-V synch after using fast or slow speeds: this can be
corrected by jumping backwards or forwards! I think there is a limit to the
A-V offset so that if the offset is greater than that value it is set to
zero: I'm pretty sure this is connected with the "position" in the file and
the current frames being shown being so far apart after using a fast or slow
speed.
For sync issues, you may use attached patch, see below.
If I jump forward or backwards by a minute, the actual position jumped (as
shown by the counter) is something like 1:12 forwards and 50 s backwards,
rather than a minute as it should be. Also, when setting a cut mark, the
position jumps forward quite a few I-frames, rather than just to the next
one.
I'm pretty sure that all of these are connected but whether it's down to
changes in vdr or in softdevice...
Any clues?
Can you both try attached patch ?
There are 2 changes:
1st Makes use of av_read_frame() selectable via OSD.
Thats because I noticed some serrious sync issues with my old ffmpeg
version I use on my prod system. Deactivating av_read_frame() and
using av_read_packed() instead solved that for me.
2nd Feature makes field selectable which is displayed in still picture
mode, but unfortunately this doesn't work at the moment.
I just tried your patch and it works. Speed is not high but for the
moment is ok for me.
Some info about my system:
Debian Etch
vdr 1.6.0
FFmpeg SVN-r15200
libavutil 49.10. 0 / 49.10. 0
libavcodec 51.71. 0 / 51.71. 0
libavformat 52.22. 0 / 52.22. 0
libavdevice 52. 1. 0 / 52. 1. 0
libpostproc 51. 2. 0 / 51. 2. 0
softdevice showed with xv
_______________________________________________
Softdevice-devel mailing list
Softdevice-devel@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/softdevice-devel
