Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > MPEG-4 AVC / H.264

Closed Thread
 
Thread Tools Search this Thread Display Modes
Old 18th September 2008, 18:16   #221  |  Link
nm
Registered User
 
Join Date: Mar 2005
Location: Finland
Posts: 2,641
Quote:
Originally Posted by crypto View Post
Third good news: The quality of the final encode seems to be higher. I don't know why and have to do more tests.
Perhaps there is some hardware postprocessing going on even when the video is not deinterlaced. Denoising might cause such a difference, for example. You could do some lossless encodes with both hardware and software decoding and compare the output frame-by-frame.
nm is offline  
Old 18th September 2008, 19:23   #222  |  Link
CruNcher
Registered User
 
CruNcher's Avatar
 
Join Date: Apr 2002
Location: Germany
Posts: 4,926
Ahh right crypto you should make sure your settings in the Control Panel dont include Nvidias Denoiseing or Sharpening Post Pro (Shaders) crazy maybe it can applied this way too
__________________
all my compares are riddles so please try to decipher them yourselves :)

It is about Time

Join the Revolution NOW before it is to Late !

http://forum.doom9.org/showthread.php?t=168004
CruNcher is offline  
Old 18th September 2008, 23:21   #223  |  Link
Diodato
Registered User
 
Join Date: May 2002
Posts: 16
A quick test of the Nvidia deinterlacing capabilities. I'm impressed.

The source is a crapy Main@L4.0 ~7.5Mb/s 25fps
(Beijing Olimpics 2008 Opening Ceremony 1080i CCTV.ts)

http://rapidshare.com/files/14643643...rlace.zip.html

All Nvidia processings are turned off (edge=off, noise=off).
Diodato is offline  
Old 18th September 2008, 23:24   #224  |  Link
blubberbirne
Registered User
 
Join Date: Sep 2004
Location: Germany, Hamm
Posts: 161
I don't get the decoder to work

Using Vista SP1 64bit here.

LoadPlugin: unable to load "DGAVCDecodeNV.dll"

The DLL is in the same directory with the avs script.
blubberbirne is offline  
Old 18th September 2008, 23:30   #225  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
Do you have CUDA drivers installed? Do you have an Nvidia card? Which one? Did you put nvcuvid.dll in System32?
Guest is offline  
Old 18th September 2008, 23:31   #226  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
Quote:
Originally Posted by CruNcher View Post
Ahh right crypto you should make sure your settings in the Control Panel dont include Nvidias Denoiseing or Sharpening Post Pro (Shaders) crazy maybe it can applied this way too
Umm, what is this control panel you're all talking about?
Guest is offline  
Old 18th September 2008, 23:31   #227  |  Link
Ranguvar
Registered User
 
Ranguvar's Avatar
 
Join Date: Feb 2007
Location: ::1
Posts: 1,236
NVIDIA driver control panel.
Ranguvar is offline  
Old 18th September 2008, 23:52   #228  |  Link
rack04
Registered User
 
Join Date: Mar 2006
Posts: 1,538
Don't really know how to translate the results but here goes:

Q6600
8800GT

DGAVCDecode


DGAVCDecodeNV


Directshow

Last edited by rack04; 19th September 2008 at 00:06.
rack04 is offline  
Old 19th September 2008, 00:46   #229  |  Link
JK1974
Registered User
 
Join Date: Mar 2005
Posts: 89
Sorry for getting off-topic in-between the lines - between all this discussion on deinterlacing - I will come back to this point later on.

I also checked DGAVCIndexNV on two XP SP3 machines with GeForce 8600 GT, and it is really fantastic. No more artifacts on my AVCHD recordings of my Canon HF100 (which I had with DGAVCIndex before).

When trying to install it on the second machine, the plug-in did not load. I installed the latest drivers 177.92 (which seems to be beta; from the "the force within" promotion section of the site), tried it with the "special" cuda drivers, installed the toolkit, nothing worked until I checked again with Sysinternals Processmon. It was finally the nvcuvid.dll which was missing in the system32 directory - after installing this I could even remove the cuda toolkit and reinstall the latest (beta) drivers.

I still have some remarks on the tool itself:
- When doing a SeparateFields(), BFF seems to be used although even DGAVCIndexNV itself says that it is TFF. Currently, I would have to set an AssumeTFF() before in any script handling with the fields.
- I think it has not really clearly been said: TempGaussMC is a bobber which makes an 1080i50 being a 1080p50 while the NVIDIA deinterlacer just outputs 1080p25. This might be interesting for Blu-ray ripping but when it is about scaling while preserving the fields e.g. for an AVCHD-to-DVD conversion, this deinterlacing is useless IMHO - or did I miss something?
- One little request: Could you please add the extension "mts" to the list of ts-streams in DGAVCIndexNV? Itīs the extension used by Canon AVCHD-Cams.
JK1974 is offline  
Old 19th September 2008, 01:38   #230  |  Link
Guest
Guest
 
Join Date: Jan 2002
Posts: 21,901
Quote:
Originally Posted by JK1974 View Post
I also checked DGAVCIndexNV on two XP SP3 machines with GeForce 8600 GT, and it is really fantastic. No more artifacts on my AVCHD recordings of my Canon HF100 (which I had with DGAVCIndex before).
Yes, it solves the problem of broken libavcodec.dll.

Quote:
When trying to install it on the second machine, the plug-in did not load. I installed the latest drivers 177.92 (which seems to be beta; from the "the force within" promotion section of the site), tried it with the "special" cuda drivers, installed the toolkit, nothing worked until I checked again with Sysinternals Processmon. It was finally the nvcuvid.dll which was missing in the system32 directory - after installing this I could even remove the cuda toolkit and reinstall the latest (beta) drivers.
Good to know. Thanks for reporting your experiences.

Quote:
I still have some remarks on the tool itself:
- When doing a SeparateFields(), BFF seems to be used although even DGAVCIndexNV itself says that it is TFF. Currently, I would have to set an AssumeTFF() before in any script handling with the fields.
SeparateFields() separates based on the parity of the clip. DGAVCDecode and DGAVCDecodeNV currently blindly set parity to BFF, because I haven't done the code to determine it and set it yet; it's been on my to-do list for a long time. Since most h264 streams are TFF, I will change it to set TFF, atleast until I get time to do it properly. It involves some parsing of PIC timing SEIs when field or frame repeats are present and other things.

Quote:
- I think it has not really clearly been said: TempGaussMC is a bobber which makes an 1080i50 being a 1080p50 while the NVIDIA deinterlacer just outputs 1080p25. This might be interesting for Blu-ray ripping but when it is about scaling while preserving the fields e.g. for an AVCHD-to-DVD conversion, this deinterlacing is useless IMHO - or did I miss something?
You can't make a DVD at double rate, it has to be 29.97 for NTSC, so your point is kind of moot. But anyway, there are a lot of applications of single-rate deinterlacing. Why do you think single-rate deinterlacers exist?

Quote:
- One little request: Could you please add the extension "mts" to the list of ts-streams in DGAVCIndexNV? Itīs the extension used by Canon AVCHD-Cams.
I think I can probably manage that.

Last edited by Guest; 19th September 2008 at 01:45.
Guest is offline  
Old 19th September 2008, 02:52   #231  |  Link
Sagekilla
x264aholic
 
Join Date: Jul 2007
Location: New York
Posts: 1,752
As a note to everything, please remember that my deinterlacer comparison was very naive, I never changed the TempGaussMC settings so it could theoretically do better.
__________________
You can't call your encoding speed slow until you start measuring in seconds per frame.
Sagekilla is offline  
Old 19th September 2008, 03:37   #232  |  Link
lexor
Registered User
 
Join Date: Jan 2004
Posts: 849
Quote:
Originally Posted by rack04 View Post
Don't really know how to translate the results but here goes:

Q6600
8800GT
to me it says one of two things:

1) If AVS is something more than "open file", then avs is a bottleneck negating advantages of hardware decoding

2) If AVS just contains an "open file" statement, then CPU cycles freed up by hardware decoding of source are inconsequential

Either way, at least we get greater stream compatibility (with paff and such). Still too bad we don't get encoding speed boost.
__________________
Geforce GTX 260
Windows 7, 64bit, Core i7
MPC-HC, Foobar2000
lexor is offline  
Old 19th September 2008, 04:03   #233  |  Link
rack04
Registered User
 
Join Date: Mar 2006
Posts: 1,538
Quote:
Originally Posted by lexor View Post
to me it says one of two things:

1) If AVS is something more than "open file", then avs is a bottleneck negating advantages of hardware decoding

2) If AVS just contains an "open file" statement, then CPU cycles freed up by hardware decoding of source are inconsequential

Either way, at least we get greater stream compatibility (with paff and such). Still too bad we don't get encoding speed boost.
So are you saying that they results are skewed because I resized in the avs?
rack04 is offline  
Old 19th September 2008, 04:31   #234  |  Link
Sagekilla
x264aholic
 
Join Date: Jul 2007
Location: New York
Posts: 1,752
I suppose when it comes to using faster settings, the GPU decoding definitely comes out on top. I'm curious how this would benefit a Core 2 Quad overclocked to 3.2 GHz Those would be -really- starved for data between my decoding Blu-ray, filtering, and encoding.
__________________
You can't call your encoding speed slow until you start measuring in seconds per frame.
Sagekilla is offline  
Old 19th September 2008, 06:25   #235  |  Link
Sulik
Registered User
 
Join Date: Jan 2002
Location: San Jose, CA
Posts: 216
Unless decode is the bottleneck (ie encode runs at 50+fps) and/or CPU is pegged at 100%, the speed boost for having even infinitely fast decoding would be small. As the numbers from rack04 show, a ~10fps encode gets a 2-5% speed boost.
Sulik is offline  
Old 19th September 2008, 06:29   #236  |  Link
Sagekilla
x264aholic
 
Join Date: Jul 2007
Location: New York
Posts: 1,752
As I said, I'm curious what the results would be on a workhorse like I just described I plan on upgrading to such a system in the near future so this will be helpful having one thing offloaded from my chain.

Also, the encode doesn't need to run at "50+ fps" for it to be a bottleneck. The decode eats up a significant portion of my cpu time on my desktop computer, where I have a lowly Opteron 170.
__________________
You can't call your encoding speed slow until you start measuring in seconds per frame.
Sagekilla is offline  
Old 19th September 2008, 06:30   #237  |  Link
JK1974
Registered User
 
Join Date: Mar 2005
Posts: 89
Quote:
Originally Posted by neuron2 View Post
You can't make a DVD at double rate, it has to be 29.97 for NTSC, so your point is kind of moot. But anyway, there are a lot of applications of single-rate deinterlacing. Why do you think single-rate deinterlacers exist?

I thought that the deinterlacers are just for all those TV-to-DivX encoders that donīt care about fluid movements.
Concerning the DVD at double rate:
For creating high-quality downconversions of the HD material to SD material, the HD material has first to be bobbed to 1080p50 at best quality (this takes huge time with TempGaussMC). Just then, you can scale it down at high quality. The resulting 576p50 has then to be weaved to 576i50.
Any possibility to bob it to 1080p50 at a comparable quality like the deinterlacer would also be a great improvement for a lot of users - at least at PAL land
JK1974 is offline  
Old 19th September 2008, 06:33   #238  |  Link
Sagekilla
x264aholic
 
Join Date: Jul 2007
Location: New York
Posts: 1,752
@JK1974: There was a thread on this before. Yes, bobbing with a very high quality bobber like TGMC would produce the best results in both 1080p and 576p, but the reality was that there was VERY little difference between a dumb bob and smart bob when downsizing to 576p.
__________________
You can't call your encoding speed slow until you start measuring in seconds per frame.
Sagekilla is offline  
Old 19th September 2008, 07:33   #239  |  Link
JK1974
Registered User
 
Join Date: Mar 2005
Posts: 89
Quote:
Originally Posted by Sagekilla View Post
@JK1974: There was a thread on this before. Yes, bobbing with a very high quality bobber like TGMC would produce the best results in both 1080p and 576p, but the reality was that there was VERY little difference between a dumb bob and smart bob when downsizing to 576p.
Thanks a lot for your clarification - donīt have the time to check the forum daily, so I must have missed it. With dumb bob you mean a simple SeparateFields()? Going to check if I find this thread...
JK1974 is offline  
Old 19th September 2008, 08:04   #240  |  Link
blubberbirne
Registered User
 
Join Date: Sep 2004
Location: Germany, Hamm
Posts: 161
Quote:
Originally Posted by neuron2 View Post
Do you have CUDA drivers installed? Do you have an Nvidia card? Which one? Did you put nvcuvid.dll in System32?
Yes, i have 177.84 installed, and also the Cuda SDK. My Card is a GTX260. And i also put the nvcuvid.dll in the System32 Folder.
blubberbirne is offline  
Closed Thread

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 18:09.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.