[mythtv-users] Jumping frames while editing

John Pilkington J.Pilk at tesco.net
Wed Nov 11 08:40:31 UTC 2009


John Pilkington wrote:
> Jim Stichnoth wrote:
>> When I edit a recording to set cut points, there is some odd behavior
>> when skipping backward or forward by N seconds.  Instead of skipping
>> immediately to the desired frame, often it jumps through several
>> frames until it converges on the desired frame.  For ATSC recordings
>> (mpeg2), there might be a couple of these intermediate frames.  For
>> HD-PVR recordings (h.264), there could be up to 6 or 7.
>>
>> I understand that this is how seeking works -- find the closest
>> keyframe and then advance frame by frame until the exact frame is
>> found.  But is there something I can do to keep these intermediate
>> frames from being displayed to the screen?
>>
>> This has been the case in 0.21 and 0.22, for many versions of the
>> nvidia driver through my current version (190.42).  I see this with
>> the VDPAU profile and the Slim profile.  I tried setting and unsetting
>> the "Seek to exact frame" setting, with no difference.  The hardware
>> is a diskless single-core IONITX box.
>>
>> Jim
> 
> I'm seeing something very like this on my box running 0.22-fixes from 
> ATrpms under CentOS 5.4, but not with the corresponding build on 64-bit 
> fc10.  The CentOS box didn't do this in 0.21-fixes.  Recordings are from 
> dvb.  Both use nvidia, neither uses VDPAU. Drivers are 96.43.13, 190.42. 
>  I'll do more experiments, but suggestions would be welcome.


What I'm seeing looks as if it may be

http://svn.mythtv.org/trac/ticket/7523

although my first attempts at tweaking nvidia haven't entirely restored 
normal operation.

Thanks to Mike Dean for pointing to this in

http://www.gossamer-threads.com/lists/mythtv/users/406836

> 
> While on the subject of the editor:  I've often wondered why the time 
> counter isn't linked directly to the image I see; there's usually a 
> one-step hysteresis  - i.e. an offset depending on the recent direction 
> of stepping.  It works, but it feels correctable.  I haven't identified 
> a ticket, or tried looking at the code :-(
> 
> John P
> 
> 
> 
> 
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
> 




More information about the mythtv-users mailing list