Active Topics

 


Reply
Thread Tools
Estel's Avatar
Posts: 5,028 | Thanked: 8,613 times | Joined on Mar 2011
#691
Good idea... Erm, go ahead?
__________________
N900's aluminum backcover / body replacement
-
N900's HDMI-Out
-
Camera cover MOD
-
Measure battery's real capacity on-device
-
TrueCrypt 7.1 | ereswap | bnf
-
Hardware's mods research is costly. To support my work, please consider donating. Thank You!
 

The Following 3 Users Say Thank You to Estel For This Useful Post:
Posts: 2,154 | Thanked: 8,464 times | Joined on May 2010
#692
Originally Posted by felipec View Post
Finally, I have pushed the maemo6-patches branch to gitorious, so you should be able to see the code. There's really not much of a difference from upstream, as I have tried to make the diff as small as possible.
Thanks very much! Here are also needed hd files.

Originally Posted by felipec View Post
Now, off to make latest linux kernel vanilla to work properly on N900
See http://elinux.org/N900#Status :-)
 

The Following 4 Users Say Thank You to pali For This Useful Post:
Posts: 65 | Thanked: 143 times | Joined on Dec 2009 @ Helsinki
#693
Originally Posted by pali View Post
See http://elinux.org/N900#Status :-)
Yeah, but that's old. A lot of stuff broke on 3.2-rc1.
 
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#694
First post updated with new libgstdsp.zip, containing new library which fixes broken videocall functionality.

Please all of you unzip and copy libgstdsp.so in /usr/lib/gstreamer-0.10

Sorry for not providing .deb, but I still have to figure out how will all this new stuff be distributed. Along with other CSSU developers of course

Thanks again to felipec for his great work on gst-dsp project.

@felipec - BTW maybe there is a minor glitch in h264 encoder. I will post it here as I am not really sure whether the bug is in maemo6 or maemo5 gst-dsp (and if it is a bug at all). Anyway maemo6 gst-dsp sets nal_mode = 2 while maemo5 one sets nal_mode = 0 for bytestream format.
 

The Following 16 Users Say Thank You to freemangordon For This Useful Post:
Posts: 65 | Thanked: 143 times | Joined on Dec 2009 @ Helsinki
#695
Originally Posted by freemangordon View Post
@felipec - BTW maybe there is a minor glitch in h264 encoder. I will post it here as I am not really sure whether the bug is in maemo6 or maemo5 gst-dsp (and if it is a bug at all). Anyway maemo6 gst-dsp sets nal_mode = 2 while maemo5 one sets nal_mode = 0 for bytestream format.
Nope, its 0 in maemo6 as well: tidsp/td_h264enc.c. But tidsp/td_hdh264enc.c is a completely different algorithm.
 

The Following User Says Thank You to felipec For This Useful Post:
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#696
Originally Posted by felipec View Post
Nope, its 0 in maemo6 as well: tidsp/td_h264enc.c. But tidsp/td_hdh264enc.c is a completely different algorithm.
Nope, it is not https://meego.gitorious.org/maemo-multimedia/gst-dsp/blobs/maemo6-patches/tidsp/td_h264enc.c#line84
 

The Following User Says Thank You to freemangordon For This Useful Post:
Posts: 65 | Thanked: 143 times | Joined on Dec 2009 @ Helsinki
#697
Oh, the value of 2 is clearly documented as bytestream, with NALU sizes, and it's not specific to maemo6, it has been in upstream gst-dsp for a long time:

https://github.com/felipec/gst-dsp/c...6c76623b574aa2
 

The Following User Says Thank You to felipec For This Useful Post:
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#698
Originally Posted by felipec View Post
Oh, the value of 2 is clearly documented as bytestream, with NALU sizes, and it's not specific to maemo6, it has been in upstream gst-dsp for a long time:

https://github.com/felipec/gst-dsp/c...6c76623b574aa2
Sure, but when using 2 for nal_mode and make a videocall to another n900 with stock maemo5 DSP binaries and stock maemo5 gst-dsp, the video shown on second device is like drugs addict dream, i.e. it cannot be properly decoded. While setting it to 0 fixes things. Anyway, as I said, maybe the problem is in stock codec/binaries. For now I will keep maemo5 build with value of 0.

EDIT: I suspect if this remain 2 in Harmattan we will have broken videocall between Maemo5 and Harmattan (once Harmattan supports it)

Last edited by freemangordon; 2011-12-09 at 16:42.
 

The Following User Says Thank You to freemangordon For This Useful Post:
Posts: 65 | Thanked: 143 times | Joined on Dec 2009 @ Helsinki
#699
Originally Posted by freemangordon View Post
Sure, but when using 2 for nal_mode and make a videocall to another n900 with stock maemo5 DSP binaries and stock maemo5 gst-dsp, the video shown on second device is like drugs addict dream, i.e. it cannot be properly decoded. While setting it to 0 fixes things. Anyway, as I said, maybe the problem is in stock codec/binaries. For now I will keep maemo5 build with value of 0.
The problem is not the nal_mode, the problem is the lack of SPS/PPS, you probably can remove the call to h264venc_strip_sps_pps_header() and it will work the same as nal_mode=0. But that's not proper.

The lack of SPS/PPS should not be a problem, and in any case, rtph264pay has an option 'config-interval' that should be set to some value if you want to re-send SPS/PPS.

Also, you need to make sure mode=1 is set, and maybe play with intra-refresh.

Originally Posted by freemangordon View Post
EDIT: I suspect if this remain 2 in Harmattan we will have broken videocall between Maemo5 and Harmattan (once Harmattan supports it)
I doubt that.
 

The Following User Says Thank You to felipec For This Useful Post:
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#700
Originally Posted by felipec View Post
The problem is not the nal_mode, the problem is the lack of SPS/PPS, you probably can remove the call to h264venc_strip_sps_pps_header() and it will work the same as nal_mode=0. But that's not proper.

The lack of SPS/PPS should not be a problem, and in any case, rtph264pay has an option 'config-interval' that should be set to some value if you want to re-send SPS/PPS.

Also, you need to make sure mode=1 is set, and maybe play with intra-refresh.
Well, I am not sure I want to go deeper in H264, video transport protocols and telepathy-stream-engine. Just not skilled enough . I am sure you are correct in what you are saying but it seems stock gst-dsp decoder (or rtph264pay or whoever) is missing proper H264 parsing.
 

The Following User Says Thank You to freemangordon For This Useful Post:
Reply

Tags
balls, gpl violation, hackjobs, lgpl violation, nokia, upgrade..., video player


 
Forum Jump


All times are GMT. The time now is 21:32.