[Softdevice-devel] viafb + cle266mpegdec + TVout : field parity status and patches

[Laz]

On Saturday 07 October 2006 22:19, Nicolas Huillard wrote:
> Hello,
>
> I'm a bit lost about what's needed for field parity correctness on
> TV-out with CLE266. I still have field artifacts on interlaced streams
> on TV-out.
>
> My setup :
> VDR 1.4.2
> softdevice CVS 20060929 (-P softdevice -ao alsa:pcm=default -vo
> dfb:viatv:cle266)
> libcle266mpegdec 0.4
> DirectFB-2006-09-10-04-25-09-UTC
> DFB++-2006-09-10-04-25-02-UTC
> linux-viafb 20060910 (viafb mode=720x576 bpp=32 refresh=50 TVon=1
> TVoverscan=1)
>
> I think I do not need any patches for linux-viafb (Mark Adams commited
> his patches to CVS)
> I don't know if I need something special in DirectFB/DFB++ (I applied
> libcle266mpegdec-0.3/patches/GetFBOffset.patch)
> I think the latest softdevice filed parity patches have joined CVS
> before I checked it out.

I have nicely working TV-out with CLE266. Looking back, I don't think I've 
updated DirectFB, DFB++, or linux-viafb since about April. I'm using 
libcle266mpegdec-0.4 with vdr-1.4.3 and softdevice CVS 04102006 
(-P'softdevice -vo dfb:viatv:cle266'). My options to viafb are: mode=720x576 
bpp=32 refresh=50 TVon=1 TVoverscan=1. Pretty much the same as you're doing!

However, you mention Mark Adams's patches: I have both viafb and DirectFB 
patched with his field-sync patches (attached).

I've just had a quick scan of the current CVS linux-viafb and can't see any 
obvious signs that they were added: at the time, I don't think Mark wanted to 
commit them because they were a quick bodge and he was going to do it 
properly. Having said that, it is possible that he has done so and that they 
are no longer needed with current CVS... It is a long while since he posted 
them so he has probably improved them by now, even if they haven't made CVS!

Last time I built DirectFB (which I'm sure was more recently than April!) I 
couldn't work out whether I needed the patches or not. Maybe I do!

With my setup, I don't see any noticeable interlace effects with TV-out. Some 
channels aren't quite perfect, i.e. scrolling text on BBC News 24 (a good 
test for this type of stuff!) gives the odd judder every second or so but is 
pretty good. Mark reckoned that this was probably down to a timing problem 
with triple buffering.

Maybe the patches will help your setup: since the hardware decoding has been 
in softdevice, my system has really impressed me, although recently it has 
decided to stop powering off / rebooting when shut down so I can't always 
trust nvram-wakeup!

:(

As to your other post about 352x288 streams, pass! Nothing on UK DVB-T is that 
low a resolution, so I cannot test this.

Cheers,

Laz
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dfb_unichrome_flip_ioctl.patch
Type: text/x-diff
Size: 2542 bytes
Desc: not available
URL: <https://lists.berlios.de/pipermail/softdevice-devel/attachments/20061008/82d74b57/attachment.patch>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: viafb_flip_ioctl.patch
Type: text/x-diff
Size: 4357 bytes
Desc: not available
URL: <https://lists.berlios.de/pipermail/softdevice-devel/attachments/20061008/82d74b57/attachment-0001.patch>