[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [K12OSN] Pixelation with MPlayer



Hmm...that is interesting.  I've used the Xv MPlayer output driver with ATI cards, and things worked fine, even with accursed .WMV's.

Could you try out another video, say, one in MPEG4 format?  That's supported directly by ffmpeg (no Windows codecs needed), so this'll help us determine if you might have a codec issue.  Also, if you can get your hands on an Ogg Theora video (here's one:  http://mediacast.sun.com/share/tmarble/moglen.ogg -- about 2.3MB) and try that.

--TP
_______________________________
Do you GNU!?
Microsoft Free since 2003--the ultimate antivirus protection!


David Hopkins wrote:
Update:  I have now tested an Epia M which has the cle266 video chipset (via driver is loaded by ltsp).  I was hoping that this would resolve the issue but the image still pixelates severely.  Perhaps the issue isn't with the video card, but how I compiled mplayer?

So ... output from mplayer -vo help is:

[student01 ncslts3 ~]$ mplayer -vo help
MPlayer 1.0rc1-4.1.1 (C) 2000-2006 MPlayer Team
CPU: Intel(R) Xeon(TM) CPU 2.40GHz (Family: 15, Model: 2, Stepping: 7)
CPUflags:  MMX: 1 MMX2: 1 3DNow: 0 3DNow2: 0 SSE: 1 SSE2: 1
Compiled for x86 CPU with extensions: MMX MMX2 SSE SSE2
Available video output drivers:
        xv      X11/Xv
        x11     X11 ( XImage/Shm )
        xover   General X11 driver for overlay capable video output drivers
        gl      X11 (OpenGL)
        gl2     X11 (OpenGL) - multiple textures version
        xvidix  X11 (VIDIX)
        cvidix  console VIDIX
        null    Null video output
        mpegpes Mpeg-PES to DVB card
        yuv4mpeg        yuv4mpeg output for mjpegtools
        png     PNG file
        tga     Targa output
        pnm     PPM/PGM/PGMYUV file
        md5sum  md5sum of each frame

and the output from playing a video is:

[student01 ncslts3 ~]$ mplayer Desktop/Shaky_Ground.mov
MPlayer 1.0rc1-4.1.1 (C) 2000-2006 MPlayer Team
CPU: Intel(R) Xeon(TM) CPU 2.40GHz (Family: 15, Model: 2, Stepping: 7)
CPUflags:  MMX: 1 MMX2: 1 3DNow: 0 3DNow2: 0 SSE: 1 SSE2: 1
Compiled for x86 CPU with extensions: MMX MMX2 SSE SSE2

Playing Desktop/Shaky_Ground.mov.
ISO: File Type Major Brand: Original QuickTime
Quicktime/MOV file format detected.
VIDEO:  [SVQ3]  320x240  24bpp  15.000 fps    0.0 kbps ( 0.0 kbyte/s)
==========================================================================
Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family
Selected video codec: [ffsvq3] vfm: ffmpeg (FFmpeg Sorenson Video v3 (SVQ3))
==========================================================================
==========================================================================
Opening audio decoder: [ffmpeg] FFmpeg/libavcodec audio decoders
AUDIO: 44100 Hz, 1 ch, s16le, 31.9 kbit/4.52% (ratio: 3984->88200)
Selected audio codec: [ffqdm2] afm: ffmpeg (FFmpeg QDM2 audio decoder)
==========================================================================
[AO ESD] latency: [server: 0.56s, net: 0.00s] (adjust 0.56s)
AO: [esd] 44100Hz 1ch s16le (2 bytes per sample)
Starting playback...
VDec: vo config request - 320 x 240 (preferred colorspace: Planar YV12)
VDec: using Planar YV12 as output csp (no 0)
Movie-Aspect is undefined - no prescaling applied.
VO: [xv] 320x240 => 320x240 Planar YV12
Shared memory not supported
Reverting to normal Xv
Shared memory not supported
Reverting to normal Xv
A:  14.2 V:  14.3 A-V: -0.018 ct: -0.068 195/195  3%  0%  2.1% 0 0

 I did try invoking xvidix, but that gave an error about not being able to determine which driver to load.  If I run as root, it detects the server's onboard video (8Mb rage XL), but then the video displays a green screen and nothing else. 

So ... any suggestions on how to recompile mplayer? Or try and download the latest rpm again? This has to work since LTSP uses the same mb's that I am using as part of their testing.  It must be something that I am doing wrong at this point.  I can try recompiling mplayer with any suggestions in order to get the playback cleaner.

Sincerely,
Dave Hopkins




On 3/5/07, David Hopkins <dahopkins429 gmail com> wrote:
I went over to the school today and checked my thin clients.  They are Epia533 but they have the PLE133 video chipset.  This is a real downer for me since I don't see where this chipset (which uses the trident video driver) has any support for MPEG acceleration (or whatever it is called).  Apparently later versions have the CLE266 (Unichrome) chipset which does but you have to be careful since they are still selling ones with the PLE133. We have had these clients since Spring 2003 and they run flawlessly (even ordered another 30 Spring 2005), but ... I am now not sure if they can handle video streams well enough.  I guess I will be spending time looking for more recent drivers, maybe a BIOS update or somesuch adds better support for video. Or, see if they support an add-in video card and purchase them but we have 75 of them at this point and that could run ~2K.

If anyone knows how to get a better driver, please let me know.  I can't even determine where the trident driver is loaded from: best guess is it is bundled with the Xorg software?  There is a tridentfb.ko, but I wasn't able to get it to load.

Hopefully, tomorrow will be better. :)

Sincerely,
Dave Hopkins

On 3/5/07, "Terrell Prudé Jr." < microman cmosnetworks com> wrote:
Actually, the full-screen bit is where your video chipset takes over.  Decompressing the video stream from, say, MPEG1/2/4, etc., does use the server CPU, but scaling the video, once decompressed, is a function of the video board.  Pop in the right device driver, and your teachers should be happy.

Please, keep us updated!

--TP
_______________________________
Do you GNU!?
Microsoft Free since 2003--the ultimate antivirus protection!


David Hopkins wrote:
I also just noticed from the LTSP wiki that my particular mb's don't necessarily get the correct configuration for the video chipset.  I have min-itx based systems with VIA, so I will try adding the XSERVER = via line and see what happens.

Of course, all the teachers want to run at full screen on the systems which is 1024x768 and I am not sure what that is going to do to the server. :(

Thanks for all the great info.

Sincerely,
Dave Hopkins


On 3/5/07, "Terrell Prudé Jr." <microman cmosnetworks com> wrote:
Yes, it sure could.  I saw that 90-client-per-server figure and went, "WHOA!!!"  I did some server CPU tests on my own server, and here's what I found:

640x480 MPEG4 clip:  21% CPU/session
320x240 MPEG4 clip:  7% CPU/session

The CPU's are 1.2GHz Athlon MP's.  This is per CPU core, BTW, so if you've got four such cores, you get just under 4x the capacity.

--TP
_______________________________
Do you GNU!?
Microsoft Free since 2003--the ultimate antivirus protection!


Petre Scheie wrote:
Could this be an issue of quantity?  That is, Terrel, how many clients have you got simultaneously watching the videos?  If Jim's got an average of 90 clients per server, even though they're big boxes (I've seen 'em), might that number be more than the server can handle for video?

Petre

Terrell Prudé Jr. wrote:
I've watched 640x480 videos (MPEG4) without any problems on my thin client with MPlayer.  The key was to swap the video board in the thin client for something a bit more powerful.  In my case, it was the Matrox Millenium G400; doing that solved all my MPlayer issues.  It also, BTW, made TuxType play faster.  Heck, I'm able to watch Tom's Hardware videos, in MPEG4, at 720x480, without any framedrops, pixelations, whatever--just nice, smooth video.  No, I don't have to throw the -framedrop switch, either.

Folks, watching videos on thin clients WORKS FINE.  I do it all the time.  If you can, please try swapping your video board and see what happens.

--TP
_______________________________
Do you GNU!?
Microsoft Free since 2003 <http://www.gnu.org/>--the ultimate antivirus protection!


David Hopkins wrote:
Didn't mean for this to cause problems, so to speak, but ... what I have observed is that the video clip is mostly watchable at the 320x240 resolution that seems to be the 'native' resolution of the clip.  However, it pixelates which is where the teachers start chiming in about the inferior quality.  It does get worse as the clips get larger:  A 5Mb clip and an 'extended' version of the clip that is 13Mb do not have the same playback characteristics, with the larger and longer clip looking worse.

My clients consist of Epia 533-based systems and donated Compaq EN SFF systems.  Either can play back without issues as long as they are fat clients. I have Gb from the server to the switch and 100Mb to the clients.

 If I specify using X11-only, then MPlayer says that the Compaq does not have the resources to play back the video.  Using Xv, it does but is pixelated on fast motion scenes. :(

MPlayer has alot of options for framedropping, lower fps playback, etc, and I will try and play with these to see if there is some magic combination that provides overall better quality.

I really hope that the Atlanta work goes well since it will help me when I periodically try to the the State of Delaware DOE interested in the thin client approach.  I have a few Reps and Senators that know about the option, but getting them to commit time to really looking at it has been tough and little glitches like lower quality movie playback seems to 'impress' them more than the phenomenal uptime of the systems, ease of use, ease of management, etc.

Slightly off-topic, but the same instuctor that is really vocal about the playback also was vocal about not getting sound on a video clip a few weeks back.  After playing it back in a pure Windows environment, it was finally possible to convince him that the clip in question didn't, in fact, have any sound and was only video.

Sincerely,
Dave Hopkins



------------------------------------------------------------------------

_______________________________________________
K12OSN mailing list
K12OSN redhat com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

------------------------------------------------------------------------

_______________________________________________
K12OSN mailing list
K12OSN redhat com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

_______________________________________________
K12OSN mailing list
K12OSN redhat com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

_______________________________________________
K12OSN mailing list
K12OSN redhat com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>


_______________________________________________ K12OSN mailing list K12OSN redhat com https://www.redhat.com/mailman/listinfo/k12osn For more info see <http://www.k12os.org>

_______________________________________________
K12OSN mailing list
K12OSN redhat com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>



_______________________________________________ K12OSN mailing list K12OSN redhat com https://www.redhat.com/mailman/listinfo/k12osn For more info see <http://www.k12os.org>

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]