[pvrusb2] WinTV PVR USB2 No video

[Tamas Szecsi]

Hi Mike,

Thank you very much for your help, the sysfs interface did get the 
device work indeed. Since the many folders to tune sysfs disappear every 
time the device is unplugged, I have created a script to update the 
.../ctl_input folder's 'cur_val' value to s-video. Little bit odd as 
those folders are owned by root so you need root privileges to do that, 
but it does work. Btw, is there any way to change those default values 
in the ../ctl_... folders so that you do not have to use the script 
every time? These folders seem to be generated by V4L so perhaps you can 
instruct V4L what the default values should be.
Using V4L applications is not an option for me as the camera will be 
used in an imaging application with an embedded computer so I need to 
get the video stream for image processing purposes, not just for 
watching the video on screen. But this way it should work.

Thanks a lot again for the fast response and for your determination to 
help people out,

Tamas

Mike Isely wrote:

>On Tue, 21 Apr 2009, Tamas Szecsi wrote:
>
>  
>
>>Hi all,
>>
>>I am trying to use a Canon VC-C4 analogue videocamera in Linux 
>>(Ubuntu). I have used it with a Belkin frame-grabber in Windows but 
>>there seems to be no Linux support for that. So instead I bought a 
>>WinTV PVR USB2 (as there appears to be an excellent support group and 
>>discussion forum for it).
>>
>>I have connected the camera to the SVideo input of the WinTV, in the 
>>Windows WinTV programme I set up the SVideo input for PAL_BGHIDK (I 
>>live in Ireland). Then I tested the setup in Ulead Videostudio (for 
>>Windows), it was abled to capture the video, so everything seems to be 
>>working fine with Windows. However, I have problems with the Linux 
>>installation.
>>
>>The 'mplayer /dev/video0' command only produces a snowflake-like 
>>image, as if I am using a channel without a signal on it. The 'cat 
>>/dev/video0 >test.mpg' command does produce a video file, but again, 
>>when watched on mplayer it is still the same 'empty' video.
>>
>>There seems to be a conflict between the 'Detected format' (NTSC-M) 
>>and the 'Specified standard' (PAL-BDGHI). Any help would be 
>>appreciated. I have attached below the 'standard' printouts for the 
>>device that appear in most mailing lists.
>>    
>>
>
>   [...]
>
>My first guess is similar to yours - a mismatch of the video standard.
>
>When you cat from /dev/video0 (or use mplayer), you're not using the 
>actual V4L API.  That itself is not really a problem, however it means 
>that you're not doing anything to configure the driver for things like, 
>for example, the video standard.  It's entirely possible that the wrong 
>video standard was used.  However I'm not 100% because if you have a PAL 
>vs NTSC foulup, you'll generally see *something* though it won't be 
>usable.  But you're describing snow which suggests no signal at all.  As 
>for the cx25843 log messages, the "detected format" doesn't carry any 
>meaning if there's no signal.  And the fact that the specified standard 
>looks right means that this aspect seems not to be the problem.
>
>Another (very likely) issue is that you don't have the right input 
>selected.  Again, unless you did something to set up the driver before 
>cat'ing from /dev/video0 then the driver is just going to use its 
>current settings.  And certainly in that case the driver defaults to the 
>RF tuner, not the S-video input.
>
>The way to solve this is through the sysfs driver interface.  Read here:
>
>http://www.isely.net/pvrusb2/usage.html#sysfs
>
>Then look at ctl_input/cur_val and ctl_video_standard/cur_val.  You can 
>do all of this from a simple shell.  Check that those are correct (and 
>if not, correct them).  Then try mplayer again.
>
>If you run a V4L app, like MythTV, xawtv or the tv-viewer app I had 
>mentioned a short while ago, then you don't have to worry about the 
>sysfs interface.  In those V4L cases, the controls all are available via 
>the V4L API, and your app will take care of it.
>
>Let me know if that helps.
>
>  -Mike
>
>
>  
>