View Full Version : DG NV tools
kypec
2nd December 2010, 10:41
Build 2037
----------
* Fixed use of an uninitialized variable that could cause crashes
and/or bad hinting for MPEG2 and VC1 files.
* Added a string to the first line of the DGI file that identifies the version of DGIndexNV
used to generate the index.
* Fixed counting of leading B frames for AVC streams.
* Improved the accuracy and robustness of the audio delay calculation.
* Revised DGDecodeNV so that serious CUDA errors do not cause the parent application to crash.
These errors now simply cause return of invalid video. Also, CUDA error popups now
time out after 5 seconds so that unattended applications do not hang.
* Added cropping parameters to DGSource(). This gives you the flexibility to define GPU cropping
either in DGIndexNV or in the DGSource() parameters.
* Added my current to-do list to the distribution (also available online).
* Backed out the mod 2 cropping granularity change because the Nvidia driver does not
honor it.
SubJunk
5th December 2010, 11:00
Starting with build 2037 several of my files aren't indexing. It was OK with 2036.
When the file doesn't work, what happens is:
I load it via DGIndexNV, I can use the GDIndexNV video player to skip to different positions, but when I try to save the file it gets stuck partway through.
One I'm looking at now won't go past frame 120, and remuxing it using MKVToolnix doesn't fix it either.
The files are fine, I am able to load and convert them using DSSource2 with no problem.
LigH
5th December 2010, 13:36
Please register in Donald Graft's forum and post your bug report there. He won't look in here much more...
7ekno
5th December 2010, 22:30
One I'm looking at now won't go past frame 120,
Same for me, I get the first 2.5 seconds and that is it ....
7ek
Guest
6th December 2010, 15:36
Same for me, I get the first 2.5 seconds and that is it .... Is that with MKV files only? If so, I've slipstreamed the fix. Re-download 2037 and test again.
If it is happening with files other than MKV, please post a link to a source file that fails?
7ekno
7th December 2010, 08:39
Yeah, I just realised it was MKV only (the GUI tool was dumping to MKV from M2TS!) ...
Thanks, will test 2037 now ;)
*EDIT* Yep, new build works fine now on MKVs, great job Don!
7ek
Hagbard23
7th December 2010, 16:59
Same for me - had the bug of crashed mkv-index-jobs - now FIXED with the newest 2037 Build.
Thanks Donald...keep up the good work...
rica
10th December 2010, 02:31
neuron,
happy to see you here again.
Let's go on. :)
I'm trying to make 3D avisynth scripts but i need your help, will you?
hydra3333
10th December 2010, 03:51
Hey, great to see you Neuron2 ! Cheers !
palwan
30th December 2010, 22:42
Hi!
I have replaced megui's dgindexnv with build 2037, after I made my dgi file and wanted to open it with meui's script creator I get http://img341.imageshack.us/img341/1100/capturefu.png (http://img341.imageshack.us/i/capturefu.png/), than I get dgmultiDecodenv from megui's update server and put dgindexnv folder, but I get missmatch error :confused: What would be the solution for my problem?
Thanks in advance.
Zathor
30th December 2010, 23:02
I assume you are using MeGUI 0.3.5.0? Please switch to the development build if you want to use the recent DGIndexNV. You can change the update server in the MeGUI settings.
Dogway
1st January 2011, 09:24
I normally use ffvideo for loading sources but I just felt I needed something faster if possible, and an elegant way to load AVC streams at the same time, that's when I discover DGNV.
My question is about performance benefits, is there any kind of comparison with different frame accurate AVC decoders?
I know that since its GPU based its gonna defintely be faster, but with a near future $15 in mind I would like to know how faster, in say simple decoder comparison, and another one comparing GPU decoding,cropping,resizing against usual ways in avisynth. Now I just thought whether there is any kind of trial version?
cweb
1st January 2011, 09:33
I normally use ffvideo for loading sources but I just felt I needed something faster if possible, and an elegant way to load AVC streams at the same time, that's when I discover DGNV.
My question is about performance benefits, is there any kind of comparison with different frame accurate AVC decoders?
I know that since its GPU based its gonna defintely be faster, but with a near future $15 in mind I would like to know how faster, in say simple decoder comparison, and another one comparing GPU decoding,cropping,resizing against usual ways in avisynth. Now I just thought whether there is any kind of trial version?
There's no trial as far as I know.. Just do the right thing and give your donation! It will be worth it even in terms of future development.
RiCON
1st January 2011, 16:12
My question is about performance benefits, is there any kind of comparison with different frame accurate AVC decoders?
According to some quick tests I made a few weeks ago, it probably depends on what board you have. With my 9500GT, ffms (using ffmpeg-mt, obv) seemed faster than both coreavc with cuda and dgnv. Might try again one day.
Dogway
1st January 2011, 23:24
Its 9600MGT on laptop, Quadro 4600 on x64 Desktop they both should work if Im not wrong.
TheRyuu
2nd January 2011, 02:36
cpu decoding will probably be faster than gpu decoding if it's threaded (ffmpeg-mt, diavc) unless your cpu is absolutely terrible.
Didée
2nd January 2011, 05:56
I know that since its GPU based its gonna defintely be faster,
Interesting things you know. :)
- If you're encoding to x264 with some "very fast" settings, chances are you'll go faster with CPU decoding. GPUs basically decode "as fast as needed", not "as fast as possible". (It's not even the GPU, but a separate ASIC that does the decoding.)
- If your encoding with some "very slow" settings (x264 and/or Avisynth), then ... it basically doesn't matter. There's the theoretical advantage "CPU cycles not used for DEcoding can be used for ENcoding and/or filtering", but ... if the whole chain is very slow (single-digit FPS speed), then the fraction of CPU used for decoding is very low. In a low-fps encoding scenario, the CPU usage needed for decoding can easily be as low as 1-2%. Now imagine you fully unleash that 1-2% for filtering/encoding. W00t, what a speed boost!! :D
Despite of that, I definetly like DGDecNV for versatility, ease-of-use, reliability, convenience, ... I like to use it for a bunch of reasons, only "speed advantage" is not a member of the bunch.
Dogway
2nd January 2011, 07:27
I see, well most .avs I process are <=1fps so I guess no benefit there, I thought it was going to be quite a speed booster. Actually in ffmpegsource2 indexing is conveniently automatic, so I just asked out of curiosity if its worth the cash related to speed gain, since except (or not) that, Im happy with the rest, talking about .h264. Thanks for the insight Didée.
hajj_3
3rd January 2011, 13:36
I believe DG NV can now add ATI support now that Ati Stream SDK 2.3 is out? http://developer.amd.com/gpu/atistreamsdk/pages/default.aspx
Adub
8th January 2011, 20:27
If that is true, then he may have to change the name. DGGPUTools anyone? ;)
Lyle_JP
9th January 2011, 00:46
If that is true, then he may have to change the name. DGGPUTools anyone? ;)
I get the feeling that DG won't bother unless a few of us pitch in and buy him some new ATI cards to play with. :) Even then, I don't think AMD would give him the same personal support he got from (smaller and more developer-centric) NVidia.
Underground78
9th January 2011, 10:01
See here (http://neuron2.net/board/viewtopic.php?f=8&t=64).
Adub
11th January 2011, 00:59
Thanks for the link Underground78. For those of you to lazy to click a link, neuron2 says "Don't hold your breath". I guess that settles that.
egrimisu
23rd January 2011, 12:34
Hi guys, does dg nv support forced film option? i can't see the setting in the menu. Thanks
Groucho2004
23rd January 2011, 13:20
Hi guys, does dg nv support forced film option? i can't see the setting in the menu. Thanks
Read the manual for DGDecodeNV.
egrimisu
23rd January 2011, 15:05
read it, thanks :)
Floatingshed
25th January 2011, 12:19
I have just made my donation and I am having a play around with DGAVCIndexDI. My goal is to be able to use BBC HD ts streams properly.
DGAVindexDi handles the pictures perfectly but no audio!
In the preview audio pane...
DGAVCIndex states: 157d: AC3 2/0 48 192
DGAVCIndexDI is blank and nothing obvious will change this.
Re-muxing with TSremux makes the audio appear but with an enormous sync error.
Any thoughts most welcome.
kypec
25th January 2011, 12:59
Floatingshed, I advise you to post your problems over at neuron's own support forum (http://neuron2.net/board/viewforum.php?f=6). He doesn't seem to visit this board anymore.
Clumpco
25th January 2011, 13:02
You might get a quicker answer on DG's own revitalised forum http://neuron2.net/board/index.php
It would be a good idea to tell us/him what you are capturing the ts file with and what (if anything) you are using to edit it before using DGIndexNV.
BBC HD captured with DVBViewer works fine here.
[Edit] Kypec beat me to it!
Floatingshed
25th January 2011, 13:29
Thanks. I'll chase him over there!
kypec
3rd February 2011, 18:15
Build 2038
----------
* Restored full update of the Info dialog during Save Project. Also added a Full_Info
option (default enabled) that controls this. Refer to the DGIndexNV user manual for details.
This option revises the INI file, so after installing 2038, open DGIndexNV and then close it
immediately to create the new INI file.
* Removed obsolete "DGMultiDecodeNV User Manual" option from the Help menu.
* Corrected erroneous frame count in the Info dialog when doing Save Project.
* Corrected parsing for profile_idc.
* Added an error message for missing PMTs.
VincAlastor
23rd March 2011, 13:22
kypec: "Is there any chance that QS video-decoder engine be used in players/source filters (MPC-HC, VLC, ffdshow, ffmpegsource) in near future?" - i thought you were the man for this interesting and important job :)
could you build a quicksync or opencl 1.1 avisynth source filter? and what do you think will the x264 developers speed up x264 encoding with quicksync, like the mainconcept encoder? - i don't know if it is necessary to wait for the z67 mainboards.
LigH
23rd March 2011, 13:51
The developer of DGDecNV is Donald Graft (neuron2), not kypec.
DGDecNV is no encoder, only a decoding plugin using Nvidia PureVideo V2.
This thread is not related to x264.
And hardware assisted AVC encoding usually is so severely limited in quality that it is beyond serious competition...
VincAlastor
23rd March 2011, 13:56
The developer of DGDecNV is Donald Graft (neuron2), not kypec.
DGDecNV is no encoder, only a decoding plugin using Nvidia PureVideo V2.
This thread is not related to x264.
And hardware assisted AVC encoding usually is so severely limited in quality that it is beyond serious competition...
thank you for information... i thougt he changed his user name.
egrimisu
31st March 2011, 10:34
Hi guys,
the delay resulting after demuxing audio using dgindex is allways accurate? i have splitted the title set (vts_1_01) using vobblanker in multiple cells(none of the cels are blank or contain only video or only audio streams) and after that import multiple in dgindex and demux it, audio result with a delay of +250, after the encode of the video (same framerate, only ivtc applyed) and muxing it with mkvmerge and adding the delay of 250 results in out of sync. The strange thing is that i apply 0 delay the ep is still out of sync, and i manualy tryed to find the sync and it -100. Any ideas? for some eps the delay is the one given by dgindex, for other 0 and for other eps i have to find it manualy. Now what's wrong ? :) thanks
laserfan
21st April 2011, 03:11
Just tried it--I love the new IDR indications! :)
Build 2039
------------
* The info overlay DGSource(debug=true) was modified so that IDR and I frames are
now properly distinguished.
* The DGI file format was modified to add a frame number indication to the FRM lines.
The file format version number was not bumped because 2039 is backward compatible with
DGI files created with 2038.
* Worked around a CUVID decoding bug for MPEG2 dual prime prediction. The workaround should
also improve decoding performance for mid- and high-end graphics cards.
* Added an error popup to notify users when successive IDR frames with the same idr_pic_id are
encountered. This is an illegal stream condition and results in loss of IDR frames.
laserfan
8th August 2011, 13:01
A new update is now available at neuron2.net:
Build 2040
----------
* Fixed a problem with Info display of frame structure.
* The code that suppresses nondisplayable frames at the end of an AVC stream was
not robust, i.e., it failed if the first GOP contained only one frame. Fixed.
* Added a workaround that fixes AVC streams that have successive IDRs with the
same value of idr_pic_id. This fixes frame dropping for these streams.
* Fixed incorrect picture display when navigating on the timeline with
double-rate deinterlacing enabled.
* The disabled state of the single-step button was solid instead of the correct hollow. Fixed
* A rewind (to start of project) button and corresponding menu item were added.
Support is via http://neuron2.net/board/viewforum.php?f=8
ndkamal
10th August 2011, 21:35
Thanks !!!
hydra3333
13th August 2011, 02:25
Is the double-framerate bug here http://forum.doom9.org/showthread.php?p=1391242#post1391242 fixed in the latest release (of NV or the nvidia drivers for the 8800GT) ? A quick test appears to suggest not.
$in0bi
30th December 2011, 23:29
You chose ignore pulldown. That will cause the pulldown to not be honored (you can see that because there are no combed frames with fieldop=2) but the frame rate will not be changed. What you should use is force film (fieldop=1). It's all explained in the manual.
Give me please the link to PM at this manual, because I can't find it. I found something, but I didn't find there anything about presets fieldop and how to ignore pulldown.
Groucho2004
30th December 2011, 23:40
Give me please the link to PM at this manual, because I can't find it. I found something, but I didn't find there anything about presets fieldop and how to ignore pulldown.
"DGDecodeNVManual.html", comes with the package.
Shevek
31st December 2011, 11:20
Give me please the link to PM at this manual, because I can't find it. I found something, but I didn't find there anything about presets fieldop and how to ignore pulldown.
It is available from the Help menu in the application.
colinhunt
28th February 2012, 19:30
neuron2, I'm testing BD-RB 0.40.08 with DGDecNV downloaded from the licensing page and properly licensed with your generator. My new laptop has a GTX 570M/3GB with latest drivers from Nvidia. OS is 64-bit Windows 7 Ultimate and all power saving features are disabled for max performance.
I'm getting "ERROR: cuInit failed" error messages on the laptop while DGDecNV is running perfectly on another 64-bit Windows 7 Ultimate rig. That one's got a GTX 275 and dual Xeon CPUs.
Is this a driver issue or something else?
Guest
28th February 2012, 20:20
Have you tried any other software that uses CUVID? DXVA? LAV Filters?
colinhunt
28th February 2012, 21:39
Have you tried any other software that uses CUVID? DXVA? LAV Filters?
Nope. I got the laptop only a few days ago and haven't had time to do much with it yet. Can you recommend some tests I could run?
Guest
28th February 2012, 21:54
DXVA and LAV filter. :)
colinhunt
28th February 2012, 22:29
DXVA and LAV filter. :)
By DXVA do you mean using DXVA in a player app like MPC-HC or something else? I have no experience with LAV at all but I suppose there's first time for everything :)
Guest
28th February 2012, 22:40
Yes, exactly.
Another possibility is try CoreCodec in its CUDA mode.
colinhunt
28th February 2012, 23:24
Yes, exactly. Another possibility is try CoreCodec in its CUDA mode.
Installed MPC-HC (32bit) along with MadVR and LAV filters. On running MPC-HC I learned that someone (ahem) had forgotten to install DirectX on the laptop. Oops. Plonked DX in and MPC-HC plays Blu-ray .m2ts files very merrily at less than 10% CPU load. GPU Monitor reported GPU load at 17% during playback.
Guess I'll run BD-RB next to see if cuInit error is a thing of the past. Thanks!
Guest
29th February 2012, 00:00
Just fire up DGIndexNV and try to load a file.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.