How to get highest quality from MiniDV Camcoder to DVD Video

[tomclary]

Are there any guides on how to get the highest quality results when taking video from a digital Camcorder to Video DVD recorded on DVD-R? The focus of most guides seem to be going from a commercial DVD to DVD-R or VCD.

I have a few questions:
1) If the video is captured over Firewire, does it matter much what program is used to capture it? Most things I have seen said it is mostly just transfering the data from the camera to the disk with no processing going on.
2) Interlaced or not interlaced? Does this matter if the final goal is to put it on a DVD to be played back on a TV?
3) Any guides on the best settings for CCE when going from DV video to DVD compliant MPEG2? The video is question is 54 minutes so should fit with no problems.
4) There are mentions of the "mb1 interlaced DV" Quantizer Matrix but most of the discussion seems to be in German. What advantages does this offer? Does it matter much if the MPEG2 is high bit rate DVD? I would like to understand the benefits before blindly using something.

I have successfully created a number of DVD's from my MiniDV camcorder and have been happy with the quality. I just want to know if there are ways to get higher quality. I've used TMPGENC in the past and started giving CCE a try recently. The big step in this process seems to be encoding the video.

[stormcrow]

i do a lot of dv to dvd in filmmaking and also archiving, with much success.

1. the program used doesn't matter, its just a binary transfer from the camera, there is no encoding or conversion involved, only a wrapping in an avi file.

2. i keep everything in interlaced, since that is how it is recorded. the only reason to deinterlace would be for pc playback, but this introduces artifacts no matter what. tv is interlaced, so if thats what you're going for, leave it interlaced.

3. sorry, don't use cce. i used tmpgenc for long time before switching to adobe premiere 6.5's built in mpeg encoder, which does a very nice job.

4. when using tmpgenc, i did use specific interlaced and noniterlaced quantizing matrices. the interlaced matrix that i've had much success with is:
8,13,13,17,17,21,21,28,
13,13,17,17,19,21,23,30,
13,17,19,19,21,23,28,34,
13,17,19,19,21,23,28,48,
17,19,19,19,23,28,34,48,
19,19,23,25,28,32,34,48,
19,21,23,25,28,32,34,48,
21,21,25,25,28,32,34,48

8,11,11,15,15,17,17,24,
11,11,15,15,17,17,21,24,
13,15,15,17,21,21,26,34,
13,17,15,17,21,21,26,48,
17,21,21,23,21,30,34,48,
17,21,21,23,28,30,34,48,
19,19,25,26,28,30,48,48,
19,19,25,26,28,30,48,48

at a high enough bitrate, which you'll be at for only 54 minutes, this should make no difference anyway, but useful information nonetheless.

[vicmarines]

HI

I use the TMPEnc to encode to MPEG-2 I have Premier 6.5 too but I dont know how good are that encoder, is it better quality than TMPenc ?

Can you explain me what those values of quntization mens for the video ?
or where to get some info explaining that process?

I use the guide for High Quality DVD using TMPEnc, but I would like to know what the quantization values are for.

greetings

VM

[tomclary]

The matrix provided by stormcrow appears the be the one refered to as 'mb1 interlaced DV' in other messages. As I mentioned before, most of the discussion on it seems to be in German.

[Thorn]

My advice would be to leave the matrix alone for a while. It's a rather advanced topic from what I can tell, and you need to know what you're doing to chooose an appropriate matrix. TMPGenc has three included: default, mpeg standard, and CG. Obviously, CG is for cartoons and such, but other than that I use default. The pros say the matrix makes a difference, but you need to tailor it to each specific video you encode. The time spent makes sense for a professional job, but for the average home user it may not be worth the extra time (the time to learn it as well as the time to use it).

I'd focus on learning to do VBR well, and using TMPGenc (or your chosen program... but TMPGenc is the cheapest).

[tomclary]

I want to do as professional a job as possible on this as I can. It was shot with a high end DV camera by a professional cameraman. It was edited on a Mac in Final Cut Pro and then mastered to MiniDV. I want to do my best to maintain the quality through the DVD creation process. This is something a friend intends to sell.

Thorn, I tend to agree with your comments about the matrix being more complex to deal with than worth the time.

I have used TMPGENC in the past but I'm focusing on CCE because it seems much faster and produces results as good.

[ronnylov]

Have you tried Canopus Procoder? It does a good job when encoding from miniDV to DVD or SVCD. I get better results with Procoder than using TMPGEnc with custom matrices.

[tomclary]

I have made a number of discoveries this week which may explain a few things. I have seen a number of places that said you needed to specify the bottom field first after encoding with CCE from interlaced MiniDV video. I did not find that to be the case. The video was jittery regardless of Upper Frame First setting if the bottom frame first was set with another utility. The output from CCE always had the upper frame first set when you look at it with changer. Doing a bit more research, I found I was using the MS DV Codec that is part of DirectX 8.1. It always deinterlaces interlaced video! So CCE was always getting a full frame and the bottom first would be wrong. I have not been able to dig up the reference, but somewhere else I read where DirectX 9 has changed the MS DV Codec to not deinterlace the video. This means CCE gets the interlaced video. The person posting the message had good luck upgrading to DirectX 9. I took the plunge and upgraded to DirectX 9.

I encoded a sample with Upper Frame First (UFF) specified and the sample with it not selected. I made copies of each of these and ran changer.exe on them to specify bottom field first. The straight UFF looked OK, the UFF with bottom first look jittery, the no UFF straight was jittery, and the no UFF with the bottom field first looked the best.

[yg1968]

I am not really sure about this but I get the feeling that the MS codec only de-interlaces upon playback but not upon conversion. I would be surprised that it de-interlaces DV upon conversion to mpeg.

[Xesdeeni]

Save yourself the headache of the hit-and-miss MS codec and download the MainConcept DV codec (http://www.mainconcept.com/downloads.shtml second from the bottom of the page). Decode is free in the demo version (encode adds water mark, but I seldom need this).

Xesdeeni

[yg1968]

I was under the impression that the Mainconcept DV codec is only used when using type 2 DV avi files (which are compatible with vfw). Type 1 DV avi uses the directshow built-in MS DV codec. So if your software uses type 1 DV avi, you have no choice but to use the directshow built-in MS DV codec.

See this link for more information:

http://www.microsoft.com/hwdev/tech/stream/vidcap/dvavi.asp

P.S. Interestingly, it seems that Panasonic is making the MS DV codec that is built in directshow. See this link:

http://www.microsoft.com/presspass/press/1997/Apr97/MSMEIpr.asp

[tomclary]

Interesting information. I found this comment on the Mainconcept forum from the Forum Admin "Our DV Codec works with all programs which support Video for Windows codecs." So this seems to confirm the comment it only works with vfw. The captured video I am dealing with is Type 1 DV. GSpot reports DirectShow as the renderer for this data file.

In CCE, there is a mode with you can look at each frame of video. It is now obviously interlaced. I do not recall conclusively what it looks like with the DirectX 8.1 but I don't recall seeing interlaced looking video.

I guess it makes sense that Panasonic (Matsushita) was involved in the creation because they were involved with the DV format. I am curious what, if any, their current involvement is.

The results seem really promising from a quality stand point and so far everything else seems happy enough with DirectX 9. It does beg the question, are you ever better to de-interlace during processing and encode a progressive frame in to the MPEG-2 output. Even though though it will probably be played on an interlaced TV.

[Xesdeeni]

I was under the impression that the Mainconcept DV codec is only used when using type 2 DV avi files (which are compatible with vfw). Type 1 DV avi uses the directshow built-in MS DV codec. So if your software uses type 1 DV avi, you have no choice but to use the directshow built-in MS DV codec.

I don't think there is a direct correlation between type 1/2 and VFW/DirectShow.

I only use type 2 clips (for reasons below) and both the MS and MC codecs can play them back. It depends on which program I use to view them, which codec I get. If I use MediaPlayer, I get the MS codec (naturally). If I use VirtualDub, I get the MC codec (or nothing if it's not installed). Both on the same type 2 clip. This would indicate what we already know: that MediaPlayer uses DirectShow, while VirtualDub uses VFW.

I don't have a type 1 clip to determine compatibility, but I suspect that the MC codec can't deal with type 1, while the MS codec can deal with both. This is probably why I switched to type 2 a couple of years ago, when I started playing with this:

I use VirtualDub and AVISynth for processing the video most of the time. VirtualDub won't work with the MS codec (DS), so I have to install the MC codec (VFW). In AVISynth, I have a choice using AVISource() or DirectShowSource(). But the latter won't give me audio, while the former does. Again, this means I need the MC codec.

For maximum flexibility, I would recommend converting your type 1 clip to type 2 (tools to the left will allow the conversion without any quality loss) and then use of the MC codec.

It does beg the question, are you ever better to de-interlace during processing and encode a progressive frame in to the MPEG-2 output. Even though though it will probably be played on an interlaced TV.

As with all opinions, take mine with a grain of salt ...

I recommend that if the video is to be encoded using a high bit rate (>3Mbps), leave interlaced alone.

For a low bit rate, it may be preferable to deinterlace, but it will depend on the source and content of the video.

Deinterlacing will reduce the temporal data, possibly making the video a bit more jerky, but it will allow the encoder to encode 30 (25) full frames instead of 60 (50) fields. This normally will provide a higher image quality at the expense of fluidity of motion.

Also keep in mind that if the source video has any motion at all (and all hand-held camcorders have this, even with image stabilization, unless they are on a tripod), even smart deinterlacers won't be able to help much. The result will be a loss of vertical resolution as they default to interpolated data. Plus there will be artifacts added by the deinterlacer that may be worse than the MPEG artitfacts present when the video is encoded a interlaced.

Xesdeeni

[yg1968]

Xesdeeni,

We basically agree. I didn't say there was a "direct" corelation between the two. I should have been clearer in my explanations.

Essentially, a type 2 DV avi file is vfw compatible. A type 1 DV avi file is not vfw compatible. Directshow can read both type 1 and type 2 DV avi files. A vfw only program (such as vdub) will not be able to read type 1 files. It will be able to read a type 2 file DV avi file if a DV codec is installed. A program that uses directshow will not need a DV codec for type 1 or a type 2 DV file.

Anyways, as I said before, Microsoft explains all of this in the following text:

http://www.microsoft.com/hwdev/tech/stream/vidcap/dvavi.asp

P.S. Incidentally, you can use the following freeware programs to capture directly in type 2 DV avi: DVapp or DVIO.

[tomclary]

I may have discovered my first problem with DirectX 9. I was thinking I was home free. I had the video the way I wanted. I tweaked the audio stream the way I wanted. I loaded these into Ulead MovieFactory 2 like I done successfully multiple times this week. First I wanted to burn a copy of the straight video so we could later master a VHS before creating menus. Everything will be fine I thought. I will save some time and burn to a 2X DVD-R. No issues with the process and get new disk. Pop in DVD player and no audio. Argggh! $5 down the drain.

Only a few things are different from what I have been doing all week. DirectX 9, play clip first, and no menus. Audio played fine in MovieFactory. Windows XP was nice enough to create a restore point before installing DirectX 9. So it was painless to roll back. I am mastering a new disk with the same MF setup. An DVD-RW this time!

I did confirm one thing. With DirectX 8.1, CCE is getting deinterlaced video with all the negative artifacts of deinterlacing. So that seems to tell me it is using the MS DirectShow Codec for Type 1. When I get the chance, I may try the Mainconcept DV Codec but I don't think it will affect CCE.

When I looked at DVIO, it had a 2gb file limit and did not seem to automatically split up the capture. I'm looking at 11gb of Type 1 video here. Does DVapp deal with larger size files?

[yg1968]

DVIO should not have any size limitation as long as you have Windows XP or 2000 and your hard drive is NTFS. I don't think DVapp has any size limitation either.

[troyvcd1]

CCE at least 2.5 will always encode TFF no matter what is checked. You need to uncheck this (i think). Anyway you need to encode and then use pulldown to convert your video back to BFF. I do not rememeber the pulldown command but it is somewhere here or for sure on doom9.org. Maybe someone else can help out.l

[tomclary]

Thanks, I am aware of using pulldown to set BFF but I have been using the Changer application which comes with the trial version of DVMPEG from Darvision (http://www.darvision.com/) to do this.

It is too early to say but I think my audio problem may be MovieFactory and not DirectX 9 related. Using the old DirectX, I still had problems when I tried to create the video where it played automatically. I did one with a single menu and it worked correctly. I guess this wouldn't be much fun if everything worked perfectly the first time.