DG NV tools beta: Multiple file support

[Guest]

Here is a full beta for my NV tools with multiple file support. All 3 indexers are included (AVC, MPG, VC1) and the decoder DLL supports all of them.

I would appreciate your beta testing and reporting any irregularities you may find.

http://neuron2.net/misc/NVbeta1.zip

[Note: To use these tools you need a supported Nvidia graphics adapter and your existing license.]

[Inspector.Gadget]

Excellent. I will get to testing these immediately. Thank you:)

[Audionut]

Tested AVC. Seems to be working fine.

edit: forgot to mention that there's no more pauses while indexing.

[Sharc]

I tested AVC and VC-1. Both seem to work ok.

Note on VC-1 "Play" and "Preview": The first few frames are blocky. I guess this is just a cosmetic issue?

[Guest]

From the user's manual:

The Play/Preview commands decode and display all frames from the starting point, i.e., they do not suppress orphaned frames. if any, at the start of the GOP. So you may see macroblocking on the first few frames of a Play/Preview operation if your stream has open GOPs. Note that this is purely a display matter in DGVC1IndexNV; it never occurs in the frame-served video or in demultiplexed video files. While such frames could have been suppressed, DGVC1IndexNV is intended for analysis and not playback; it can be useful to operate this way because one can easily identify open GOPs in the stream. Note that when navigating on the timeline, orphaned frames are suppressed.

[Sharc]

Very clear .... sorry.... :o

[Kurtnoise]

Don, CUVIDServer executable is missing in your current package...

:thanks: btw

[hajj_3]

very interesting, thought u had made a single app called nv tools instead of 3 seperate ones, would have been incredible having 1 program, but its still nice that you've added multiple file support:)

[Guest]

CUVIDServer executable is missing in your current package Added. Thank you.

[Guest]

very interesting, thought u had made a single app called nv tools instead of 3 seperate ones, would have been incredible having 1 program, but its still nice that you've added multiple file support:) The frame server DLLs have indeed been combined into one and the merger of the indexers is still on the agenda. Some things can't happen overnight.

[trex]

Somehow, i cannot get the new version to work.
I use a template as follows:


loadplugin("...\DGDecodeNV.dll")
avcsource("__vid__")

When I start the avs-file, there is a error saying "no such function avcsource on line 2"

This is really strange, because I never had such problems.

Trex

[Kurtnoise]

DGSource(...)

[Audionut]

there is a error saying "no such function avcsource on line 2"

Don making sure we're on the ball and reading the manuals. :devil:

[trex]

Kurtnoise

Thanks for pointing that out.
Reading the included help files, I had the impression that avcsource() was still the command to be used.

Cheers, trex

[Guest]

Reading the included help files, I had the impression that avcsource() was still the command to be used. What part of which help file gave you that impression?

[trex]

Hello Neuron

The following part:


AVS Template

DGAVCIndexNV can automatically generate an Avisynth script for serving your project. It requires you to specify a template file, such as the following one:

loadplugin("...\DGDecodeNV.dll")
loadplugin("...\Decomb.dll")
avcsource("__vid__")
fielddeinterlace()



Cheers, trex

[Turtleggjp]

Thanks for the multiple file support. I'm trying out a transcoding of my video camera's footage right now.

Question: Would you be willing to implement some sort of simple audio sync correction when working with multiple files? I already have to do this with my video camera's footage, and although my method does work, it is pretty lengthy and complicated to do. This is mainly because I rely on a couple of different tools give me the information I need (number of audio and video frames in each clip). Once you have this information (which your indexer should be able to calculate on the fly while indexing), it's fairly simple to decide when to toss out an audio frame at the end of one clip and when not to. If you could do this, it would certainly help out owners of AVCHD camcorders who want to just simply join several of their clips together without re-encoding the video and audio. Thanks!

Matt

EDIT: My transcode just finished, and the sync actually looks pretty good. Are you already doing some sync correction with multiple files?

[Guest]

No, I'm not doing anything special between files. If you can create the issue let me know and we'll see what can be done.

@trex

Thank you. I will fix the documents.

[Turtleggjp]

Ok, I'm going to be doing some experiments with my latest trip's footage (49 clips). I'll let you know how the sync comes out.

BTW, the automatic sorting of the clips turned out to be unhelpful for this particular set of clips because they start out going from 00976.m2ts to 00999.m2ts, and then go from 00116.m2ts to 00141.m2ts. When I tried to add them in this order, it sorted them all when I added the second group. I then had to move the 900 series of clips one at a time to the top. This situation sometimes come up for me especially if I change memory cards during the trip, as my camcorder uses numbers for the clips based on what card is in there. I'm guessing you are sorting the clips because Windows does not always give them to the program in order, so that's great. However, it would be better if it only sorted the clips that are being added and not the entire list of clips, so that way if you need to add them in a strange order like I did this time, it is easy to do so. Thanks

Matt

EDIT: Ok, it is definitely not in sync, just as I expected. It appeared to be before because I was having trouble playing back a 59.94 fps XviD file. The audio was drifting ahead by quite a bit, so if I played it for a short time to check the sync in a certain area, it was drifting far enough ahead to correct the sync problem in the source file making it look right! When I let it play from the beginning, the audio was about 8 seconds ahead of the video after 18 minutes. According to my calculations, the audio should be about 672ms behind by the end of the clip. Once I corrected this issue by switching MPC-HC to use the standalone XviD decoder instead of it's own internal XviD decoder, I confirmed that this was indeed the case.

I was going to post an example of how I correct the sync, but I may have just discovered a slight bug in it. I'll look into this and report back later.

EDIT2: The bug turned out to be irrelevant. Here is a sample of my calculations. The data in the first two columns is what needs to be discovered about each clip. The rest are calculations based on that data. In this case, my video frame rate is 29.97 fps and each audio frame (Dolby Digital 2.0 256kbps) is 32ms. All times are give in seconds.

Video Audio Total Audio Video Excess Adjust For Excess Audio Round Actual Total Actual Total Excess
Frames Frames Video Time Time Audio Error Frames Audio Audio Time Video Time Audio

675 704 675 22.528 22.52252252 0.005477477 0.005477477 0.171171171 0 704 704 22.528 22.52252252 0.005477477
495 517 1170 16.544 16.51651652 0.027483483 0.032960961 1.03003003 1 516 1220 39.04 39.03903904 0.000960961
390 407 1560 13.024 13.01301301 0.010986987 0.011947948 0.373373373 0 407 1627 52.064 52.05205205 0.011947948
645 673 2205 21.536 21.52152152 0.014478478 0.026426426 0.825825826 1 672 2299 73.568 73.57357357 -0.005573574
735 767 2940 24.544 24.52452452 0.019475475 0.013901902 0.434434434 0 767 3066 98.112 98.0980981 0.013901902
1185 1236 4125 39.552 39.53953954 0.01246046 0.026362362 0.823823824 1 1235 4301 137.632 137.6376376 -0.005637638

When the number in the "round" column is not zero, an audio frame needs to be dropped in order to maintain sync. Since the difference between the audio and video tracks has exceeded half an audio frame (16ms), dropping 1 frame will cause the sync to now be no worse than -16ms.

I got this information from my Excel spreadsheet that I use to calculate the sync info. If you would like to see this as well, I'd be happy to post it. Also, if you would like to see the clips that these results came from, I can upload them too. The six clips from the above sample total 269MB.

[Kurtnoise]

@Turtleggjp: what about to concatenate all your files w/ TsMuxer to recreate a m2ts file ?

@Neuron2: I'd like to submit you a feature request: to be able to select multi-ranges. Sometimes, we can have some ads somewhere in the input file (like TS files from recordings) and to avoid loading it many times that'd be great to trim such things with these tools...:)

[Turtleggjp]

@Turtleggjp: what about to concatenate all your files w/ TsMuxer to recreate a m2ts file ?

I haven't tried that with any of the recent versions, so I don't know if that program has the code in it to handle sync issues with AVCHD camcorder files. I do know that it does not work with the version I have been using since I got my camera (1.8.4). The resulting file is playable, but the audio drifts out of sync after several files.

[lansing]

CUVID server keep running when I tried exit it

[Guest]

Use task manager or process explorer to kill it.

[deets]

I cant get it to load in megui via avisynth creator, megui gives an invalid license error?

i know this isnt a megui thread, but is it just me? I cant convert my rips easily until its sorted :(

[Guest]

@deets

Do you have a license?

[deets]

@deets

Do you have a license?

yes, i wouldnt be able to make the file without one i thought?

[Audionut]

Have you changed/updated OS or network card/drivers?

[Guest]

yes, i wouldnt be able to make the file without one i thought? Has your machine ID changed? If so, email me the new ID so I can send you a new license.

I assume you copied your license.txt into the same directory as this new executable.

[deets]

Has your machine ID changed? If so, email me the new ID so I can send you a new license.

I assume you copied your license.txt into the same directory as this new executable.

yep i had. i dont know if it was the megui update or the fact i restarted it all with a new folder etc. but its all working again now :) so im happy!

[[)370|\|470!2]

What part of which help file gave you that impression?

http://img67.imageshack.us/img67/5461/000x0.jpg (http://neuron2.net/dgavcdecnv/AVCQuickStart.html)

How ironic. And i've bumped into this too.

[Guest]

If you look at the quick start file that is included in the current NV beta 10 distribution you will see that it has been updated for some time now.

[[)370|\|470!2]

Well, you asked. :)

[Guest]

Yeah, over three months ago. Thank you so much for your timely contribution.

[[)370|\|470!2]

Timely or not, but that sorta confusing page haven't changed since then. Perhaps it's just me checking an online documentation favorably out assuming it'll be more likely up-to-date. Never mind.

[Guest]

Closing as discussion of the beta moved here some time ago:

http://forum.doom9.org/showthread.php?t=147945