GT430 + 260.19.21 + VDPAU = artifacts

[ogl]

I have this same problem with my ION board since updating my kernel and nvidia drivers.

I use vdr and the xineliboutput and streamdev streams play cleanly on my windows laptop. The host machine uses vdr-sxfe with VDPAU and the "sparkles" in dark areas are very noticeable. Only happens on SD MPEG2 channels (which I have set to temporal+spatial) and HD H264 (temporal only) appears clear. Definitely not a broadcast/tuner issue.

How do I go about screen capping VDPAU in ubuntu?

Edit: Just switched from temporal+spatial to temporal on SD MPEG2 and the problem is gone.

[micmad]

I can chime in and report that I also have the same problem with these white "sparkles". I'm using GT240, 260.19.26 drivers, xineliboutput, and MPEG2 SD material w/ temporal+spatial enabled.

[eidenscm]

The same problems are arising with my configuration (vdpau + xineliboutput). However I found a workaround that works. Perhaps it helps nearing the Problem.
The workaround is to set lightness and contrast in the video out settings in the menu:
settings-plugins-xineliboutput-video from vdr.

[eidenscm]

I must correct me, setting contrast and brightness didn't help really.
But an update to kernel 2.6.32-25-generic an a rebuild of all relevant kernel modules helped.

[G3RG3R]

fixed in latest version (260.19.29)
Link: http://www.nvidia.com/Download/index.aspx?lang=en-us

[micmad]

Originally Posted by G3RG3R

fixed in latest version (260.19.29)
Link: http://www.nvidia.com/Download/index.aspx?lang=en-us

As much as I wanted to believe that, it does not seem to be the case. Running 260.19.29 and still the white artifacts can find their way to my screen. As if there wouldn't be enough snow already outdoors here in Finland...

[ogl]

Can anybody else confirm it has been fixed in 260.19.29?

My artifacts have returned despite switching to temporal only on 260.19.21

[sandipt]

Some of you facing this issue on TV playing video and it looks like issue related to temporal and temporal-spatial de-interlacing . Can somebody use mplayer to test/repro the issue with video clip ftp://urpla.net/sample.mpeg ?

mplayer -vo vdpau:deint=3 <Video_file> # Temporal
mplayer -vo vdpau:deint=4 <Video_file> # Temporal Spatial

[ogl]

Sandipt,

The artifacts occur for me with both settings, it becomes noticeable towards the end of the clip.

However, the best example I have seen is when the BBC sometimes broadcast their Ceefax service in the early morning. Lots of near black = many artifacts. I must try and capture a clip.

Hope this helps.

[sandipt]

Hello ogl,
Please provide details nvidia_bug_report, connection cables used, mplayer version, screenshot, video clip at the point artifacts occur.

[ogl]

I have also noticed the artifacts on progressive xvid content in XBMC. I think my nvidia ION does not even provide acceleration for this.

I will provide more information as soon as possible.

Edit: Upgrading to 260.19.29 has fixed the artifact problem for me, but now I have lost audio over hdmi (that could be down to me playing with pulseaudio though).

Edit 2: Artifacts ARE still present on some clips, but less noticeable.

[Xeno22]

I have the same issues on my development box (AOpen Intel-Atom HD player DE7000) at work.

The issue: There are white pixels appearing on very dark areas while playing H264 content with mplayer(from SVN) with VDPAU or XV. Especially visible with the free HD trailer of "Season of the Witch" (my colleague downloaded it from Apple in 1080 as mp4).
It does not matter if I enable render acceleration in Xorg server or not, also the composit extension does not matter.
It sometimes appears after reboot, and sometimes it does not.
Sometimes it is worse with VDPAU sometimes with XV. No clue why.



Kernel: 2.6.36.2 (64bit)
CPU: Pentium(R) Dual-Core CPU T4400 @ 2.20GHz
Chipset: MCP79
RAM: 4GB
HDD: 8GB SSD
Monitor connected via: HDMI to DVI Adapter

Nvidia ION C79: 02:00.0 VGA compatible controller: nVidia Corporation Device 0874 (rev b1) (prog-if 00 [VGA controller])

Nvidia driver: 260.19.29

Kernel log with nvidia stuff:

[    2.017245] ACPI: PCI Interrupt Link [LPMU] enabled at IRQ 23
[    2.017251] nvidia 0000:00:03.5: PCI INT B -> Link[LPMU] -> GSI 23 (level, low) -> IRQ 23
[    2.017498] ACPI: PCI Interrupt Link [SGRU] enabled at IRQ 22
[    2.017502] nvidia 0000:02:00.0: PCI INT A -> Link[SGRU] -> GSI 22 (level, low) -> IRQ 22
[    2.017508] nvidia 0000:02:00.0: setting latency timer to 64
[    2.017513] vgaarb: device changed decodes: PCI:0000:02:00.0,olddecodes=io+mem,decodes=none:owns=io+mem
[    2.018028] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  260.19.29  Wed Dec  8 12:08:56 PST 2010

Xorg log with the NVIDIA lines:

[    14.236] (II) NVIDIA GLX Module  260.19.29  Wed Dec  8 12:24:30 PST 2010
[    14.250] (II) NVIDIA dlloader X Driver  260.19.29  Wed Dec  8 12:10:14 PST 2010
[    14.250] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[    14.270] (**) NVIDIA(0): Depth 24, (--) framebuffer bpp 32
[    14.270] (==) NVIDIA(0): RGB weight 888
[    14.270] (==) NVIDIA(0): Default visual is TrueColor
[    14.270] (==) NVIDIA(0): Using gamma correction (1.0, 1.0, 1.0)
[    14.270] (**) NVIDIA(0): Option "NoLogo" "true"
[    14.270] (**) NVIDIA(0): Option "RenderAccel" "true"
[    14.270] (**) NVIDIA(0): Option "AllowGLXWithComposite" "false"
[    14.271] (**) NVIDIA(0): Enabling RENDER acceleration
[    14.950] (II) NVIDIA(0): NVIDIA GPU ION (C79) at PCI:2:0:0 (GPU-0)
[    14.950] (--) NVIDIA(0): Memory: 524288 kBytes
[    14.950] (--) NVIDIA(0): VideoBIOS: 62.79.77.00.00
[    14.950] (--) NVIDIA(0): Interlaced video modes are supported on this GPU
[    14.950] (--) NVIDIA(0): Connected display device(s) on ION at PCI:2:0:0
[    14.950] (--) NVIDIA(0):     Idek Iiyama PLT2250MTS (DFP-0)
[    14.950] (--) NVIDIA(0): Idek Iiyama PLT2250MTS (DFP-0): 165.0 MHz maximum pixel clock
[    14.950] (--) NVIDIA(0): Idek Iiyama PLT2250MTS (DFP-0): Internal Single Link TMDS
[    14.980] (II) NVIDIA(0): Assigned Display Device: DFP-0
[    14.980] (II) NVIDIA(0): Validated modes:
[    14.980] (II) NVIDIA(0):     "1920x1080"
[    14.980] (II) NVIDIA(0): Virtual screen size determined to be 1920 x 1080
[    15.003] (--) NVIDIA(0): DPI set to (101, 101); computed from "UseEdidDpi" X config
[    15.003] (--) NVIDIA(0):     option
[    15.003] (==) NVIDIA(0): Disabling 32-bit ARGB GLX visuals.
[    15.003] (--) Depth 24 pixmap format is 32 bpp
[    15.003] (II) NVIDIA: Using 768.00 MB of virtual memory for indirect memory access.
[    15.004] (II) NVIDIA(0): Initialized GPU GART.
[    15.007] (II) NVIDIA(0): Setting mode "1920x1080"
[    15.031] (II) Loading extension NV-GLX
[    15.058] (II) NVIDIA(0): Initialized OpenGL Acceleration
[    15.069] (==) NVIDIA(0): Disabling shared memory pixmaps
[    15.069] (II) NVIDIA(0): Initialized X Rendering Acceleration
[    15.069] (==) NVIDIA(0): Backing store disabled
[    15.069] (==) NVIDIA(0): Silken mouse enabled
[    15.076] (==) NVIDIA(0): DPMS enabled
[    15.077] (WW) Warning, couldn't open module dri2
[    15.077] (II) UnloadModule: "dri2"
[    15.077] (EE) NVIDIA: Failed to load module "dri2" (module does not exist, 0)
[    15.077] (II) NVIDIA(0): The X server will not be able to send the VDPAU driver name to
[    15.077] (II) NVIDIA(0):     libvdpau.
[    15.077] (==) RandR enabled

nvidia module options:

options nvidia NVreg_DeviceFileMode=432 NVreg_DeviceFileUID=0 NVreg_DeviceFileGID=27 NVreg_ModifyDeviceFiles=1

I tried to capture a photo of it, but the pixel issue did not appear until now, will try to reboot a few times.

Ok, took a photo (with my mobile, very bad quality). And I found out something interesting:
The wrong pixels are just transparent. I was able to see the underlying video of a different mplayer session thru the pixels (were color changing all the time when something in the underlying movie was moving).

Here are the pictures:




BTW: I have tearing issues with VDPAU output, they do not appear always, but often. Will there be any fix soon?
At the moment I'm stuck with XV output which needs a bit more CPU cycles ...