Log in

View Full Version : Using Lagarith to compress DV


NLight95
13th December 2010, 21:26
I'm trying to compress some DV captures using VDub64 with the Lagarith codec but the target file ends up being much larger than the source. I captured the video using Nero Vision with the DV Type-2 setting.

source: 349 MB
target: 1.00 GB

I'm new to using vdub so perhaps I'm overlooking something.

Usedocne
13th December 2010, 21:48
Lossless encodes will always be much bigger than your source - that's normal.

NLight95
13th December 2010, 21:57
Lossless encodes will always be much bigger than your source - that's normal.

I guess I don't understand. I thought the reason to use any codec (lossless or otherwise) was to "compress" the video. I know this may not be an equal comparison, but I think of FLAC being smaller than WAV.

Dark Shikari
13th December 2010, 21:59
I guess I don't understand. I thought the reason to use any codec (lossless or otherwise) was to "compress" the video. I know this may not be an equal comparison, but I think of FLAC being smaller than WAV.WAV is uncompressed. DV is not. Converting from DV to a lossless format is like converting from MP3 to FLAC.

NLight95
13th December 2010, 22:11
WAV is uncompressed. DV is not. Converting from DV to a lossless format is like converting from MP3 to FLAC.

Does this means that when I captured to DV I lost some video quality in the process?

Can I recapture the video again but to a lossless format instead? The videos are on DV tapes recorded by my old Sony DCR-TRV25 camcorder.

Dark Shikari
13th December 2010, 22:24
Does this means that when I captured to DV I lost some video quality in the process?Yes, yes, you did.

NLight95
13th December 2010, 23:59
Yes, yes, you did.

OK, I've decided to recapture the video but vdub doesn't do capturing.

What I'd like to do is capture the video with scene detection and auto split those scenes and save them as different files into a folder. Nero Vision does this but only in DV format (as far as I know). I'd like to use a lossless codec rather for capture rather than DV.

Is there a freeware program that will do this? If not, then perhaps a commercial program? I've tried AVIcutty but it crashes all to often and it peppers you with nag screens.

frubsen
14th December 2010, 00:24
Yes, yes, you did.

If he captured using firewire, then he didn't lose any quality. What he captured should be identical to what was on the tape. If you want to compress the video further, you need to use a lossy codec such as xvid or h264.

NLight95
14th December 2010, 00:31
If he captured using firewire, then he didn't lose any quality. What he captured should be identical to what was on the tape. If you want to compress the video further, you need to use a lossless codec such as xvid or h264.

I captured to DV (type-2) using a firewire connection. When I tried to compress the video further using vdub64 + logarith the target ended up about 3x larger than the source DV file (see op).

frubsen
14th December 2010, 00:56
I captured to DV (type-2) using a firewire connection. When I tried to compress the video further using vdub64 + logarith the target ended up about 3x larger than the source DV file (see op).


Well that's because DV video already as a compression ratio of 5:1. Essentially you are uncompressing the video by going to a lossless format like lagarith.

henryho_hk
14th December 2010, 01:02
1) DV itself is lossy and compressed.
2) PAL DV, NTSC DV and MPEG4 are using different pixel sampling methods, although all are called YV12. Make sure that u are using a proper PC DV codec to decode the DV materials.

NLight95
14th December 2010, 02:55
I'd like to capture lossless. What software will allow me to capture using logarith and also allow me to use scene detection and save each scene as a separate file in a folder?

mariush
14th December 2010, 04:01
Virtualdub can capture from Firewire, just go to Capture mode and select the Firewire from the devices list.
You can also set custom resolutions, color spaces and codec and so on.

henryho_hk
14th December 2010, 04:33
If your source is actually DV tape, u can use programs like "WinDV" to dump the content bit-identically to your PC.

NLight95
14th December 2010, 05:14
Virtualdub can capture from Firewire, just go to Capture mode and select the Firewire from the devices list.
You can also set custom resolutions, color spaces and codec and so on.

I need scene detection as each DV tape contains videos of which some I want to keep and others discard. Vdub64 can't do that (as far as I know).

NLight95
14th December 2010, 05:33
If your source is actually DV tape, u can use programs like "WinDV" to dump the content bit-identically to your PC.

When capturing with WinDV it gives me a bunch of smaller files not based on scenes but some other value I have no idea of. Even if I could get it "working" I'd have to apply compression and split the scenes. I'm wanting to do everything during the capture process. Nero Vision will do everything except use logarith (or Huffy) to apply compression during capture.

Does anyone know if Nero Vision can use a third-party codec?

Hagbard23
14th December 2010, 07:08
If i understand it correctly, even if nero WOULD accept lagarith, you wouldn't get any benefit of that sort you are searching for:

You want to end up in smaller files - or -at least- not so big files compared to your DV-Original. That is not possible with the way you are thinking. You have to use a codec which applies more compression than the 5:1 DV-Codec to get to smaller sizes - this follows strictly logical ways.

Lets play a little mind-game:

If it would be possible to get nero interacting with lagarith, you will do the same thing, you are already doing now, and so end up at the same filesize.

Someone in here said, that DV has a compression ratio of 5:1.

So Nero would decompress the DV Footage to some kind of uncompressed frames(uncompressed=relative to the source-content, which already lost detail due to DV-Compression, uncompressed means here mainly: without B-Frames and so on, quasi an "I-Frame only" Stream). Those files would end up unthinkable big, even if they don't have THAT much detail or noise in 'em. Nobody would feel comfortable with those files. So: The Files are then re-compressed via Lagarith and it does the job quite well. If you would compare those intermediate-decompressed files with the lagarith-compressed files, you would even get along with your WAV<->FLAC comparison. BUT: Those files are still bigger than the -original, high compressed- DV-Files, and that is what you DON't want. It doesn't matter, when, or where Lagarith comes into play, inside nero or inside virtual dub or somewhere else. As long as it takes the uncompressed files, it would end up with the same (big) filesize.

So: The point you are thinking about, will NOT give you the solution you are searching for. As already said:

If you want to end up in smaller files, then:
- Process the Capture as you already have done: Via Firewire (bitcorrect) to Nero.
- Do your cutting and snibbling around... ;)
- After that get the DV (best option for most people) via Avisynth (Filtering!) to a codec like x264.

Or: You prefer pseudo-lossless (lossless relative to DV-Compression) huge files and get along with it. But you would be dumb (truly speaking) if you will do that. It would have been better, to do no fiddling around at all, and keep the files in their native format.

That are your options.

You can't make the capture very better, than you already do. Catching bitcorrect via Firewire is one of the best options you have. Fiddling a Lossless codec in between won't give you any benefits. (Hope i got it right..DV is not my hometown...)

What you MIGHT really want is:
A Camcorder which doesn't use 5:1 Compression. Then you got a real benefit. If you find a camcorder recording in Lagarith or at least a codec with less compression, you have found the big thing you are searching for, and then it would make sense to play around with lagarith.

Otherwise it is just the normal senselessness of upscaling an mp3 to ac3 or wav or FLAC. You will get bigger files with less quality...unfortunately..


BTW: @DS:
He he...in Fact WAV isn't "uncompressed"...compared to analogue original, the WAV Format is always a lossy compression (mostly 44.1 sampleprocesses per second, which isn't that much) - even at 32BitFloat 192000Khz. Digitalization is shit anyways - you only can make the shit less visible... ;) ... analogues are dead, long live the analogues... ;)

Gavino
14th December 2010, 11:12
I'd like to capture lossless. What software will allow me to capture using logarith and also allow me to use scene detection and save each scene as a separate file in a folder?
As others have said, a Firewire DV transfer (a more accurate word than 'capture') is lossless. You get bit-for-bit exactly what is recorded on the tape.
When capturing with WinDV it gives me a bunch of smaller files not based on scenes but some other value I have no idea of.
You can configure WinDV to split the files based on time-code discontinuity (hence scene-change) and/or on max no of frames (hence duration). Check out the 'Config...' panel.

DmitryPopov
16th December 2010, 09:04
a Firewire DV transfer (a more accurate word than 'capture') is lossless.
But DV tape stores video already compressed with lossy DV codec.

Gavino
16th December 2010, 12:31
As I understand it, the OP's source is a DV tape.
A Firewire transfer to PC is a lossless 'capture' of that data and cannot be bettered - the losses occurred earlier when he shot the scenes on his camera.