Archive DV on computer

[sankaboy]

I have about 10-20 miniDV and would like to achive them on my computer for the future.
As DV takes lot of space I would like to compress them to a smaller format with good quality like H.264 for example.

My question is how I convert my DV's to H.264?

I will first transfer them to my computer via WinDV but then what?

Any good available guides or tutorials?
What software to use?
How do I deinterlace the videos with as good quality as possible?

[guns1inger]

By all means put them on the computer, but also put the tapes away in a cool dark cupboard out of the damp. They are your true archive.

[Vidd]

How do I deinterlace the videos with as good quality as possible?

You are going to want to use AviSynth then. There are many options to choose from and this is open to debate.

One of the cleanest deinterlacers I have found is yadifmod+nnedi.

A script would go something like this:

interp = nnedi()
yadifmod(edeint=interp)

This is painfully slow. 0.5 to 2 fps per second on a fast machine.
yadifmod and nnedi were coded by tritical and can be downloaded from http://bengal.missouri.edu/~kes25c/



Yadif 0.9 by itself will give good results but will show some artifacts in some situations. Yadif is very fast and can deinterlace in real time.

yadif()

Yadif was ported by Fizick and can be downloaded from http://avisynth.org.ru/yadif/yadif.html

[jagabo]

You should be aware that deinterlacing:

1) Removes half the temporal resolution (50 to 25 images per second in PAL video) resulting in jerky motions, especially camera pans in bright scenes.

2) Removes some of the spacial resolution. In the worst case half the spacial resolution is lost. The better deinterlacers will do much better than that but you will not retain 100 percent under any circumstances.

3) Create artifacts like jagged edges, moire patterns, etc. Again, the better deinterlacers will produce less artifacts but they will still create some.

Deinterlacing is ok for viewing but not as a sole and permanent archive. The same goes for h.264 or any other lossy compression format.

[Soopafresh]

Lots of quality lost when you deinterlace standard def DV. The "best" method will run at 1fps on your system x 10 or 20 tapes. I wouldn't convert to h264 or deinterlace if you wish to use your footage for future projects. Just buy a bigger hard drive and store the original footage.

stockholma_0-520_q3_yadif_mvbobmod_tgmca4_tdtmm.avi

[Vidd]

The way I backup DV is to dump the raw DV to an external hard drive for permanent stoarge. You can fit around 40 hours per 500 GB drive. I also do a high bitrate mpeg-2 encode preserving the interlace and burn it to disc. And of course try to preserve the original tape.

[JohnnyMalaria]

Ditto @Vidd

[sankaboy]

Okey, maybe I just should save my films as they are on my computer, but I have some questions.

If I would like to make DVDs from my filmes, don't I have to deinterlace them anyway?

As my films are filmed with a 10 year old camera (Sony PC100) it feels like it doesn't mutter if I lost just little bit of quality as I don't think I would see any different, or would I?

[jagabo]

If I would like to make DVDs from my filmes, don't I have to deinterlace them anyway?

No, DVD supports interlaced video.

[sankaboy]

That is what I am talking about. Don't I have to deinterlace my videos anyway if I want to make a DVD or a H.264 video?

By the way, should I deinterlace my video before using it in for example TMPGEnc DVD Author to create a DVD?

Can I create a H.264 with same software?

[JohnnyMalaria]

You should leave the interlacing as it is for DVD. Let the decoders do the deinterlacing if it is needed. On a regular TV, it will look far better if you do leave it interlaced. If you only ever intend to watch it on a PC or an LCD TV then you could choose to deinterlace but there's no need really since the display hardware will do it.

[sankaboy]

Should I save my DVs in Type1 or Type2 format?

[PuzZLeR]

If I would like to make DVDs from my filmes, don't I have to deinterlace them anyway?

No you don't. To extend on Jagabo's post, even when you use deinterlacers, and even if it has been deinterlaced, the stream still needs to be encoded as "interlaced" to be compliant with the DvD standard. I know, not straightforward, but a "flag" must be raised in the video stream to identify this. Then again, it's not a problem. If you enable DvD encoding, the good encoding software will automatically implement this for you.

And backing up a moment, if I may

The way I backup DV is to dump the raw DV to an external hard drive for permanent stoarge. You can fit around 40 hours per 500 GB drive. I also do a high bitrate mpeg-2 encode preserving the interlace and burn it to disc. And of course try to preserve the original tape.

I agree with your scheme. Personally I manage much of my DV source cut up on several DvD-R/DL in some spindles at the back of one closet (it's psychological, making me feel it's not taking up my "real" storage :P).

However, I will encode some content to high bitrate MPEG-2 (and of course preserve interlace) - and this becomes my new "source". I will opt to delete the DV source of the less important stuff for a massive bitrate saver and small quality hit. Not all, but some.

Can I get your software and settings for this when you yourself encode to MPEG-2 (or even Johnny's)? Just curious. I personally just go with CCE, using one-pass VBR (OBR), Q=10, Min=0, Max=9000 and the quality, although obviously lossy by theoretical means, is at least indistinguishable from the original visually.

[CogoSWSDS]

Don't deinterlace. Store the original DV files on an external hard drive. I make an additional backup to Verbatim single-layer DVDR as well. As Vidd said, you can get about 40 hours of DV on a 500GB external hard drive. Compare this to 15 minutes on a single layer DVDR.

CogoSWSDS

[sankaboy]

Type-1 or Type-2?

[CogoSWSDS]

Type 2 is generally supported by more video apps, so I'd go with type 2.

CogoSWSDS

[gadgetguy]

Type-1 or Type-2?

Irrelevant. They are exactly the same quality. Use whichever type you capsferred it in.

[sankaboy]

Thanks for all answares.. I will go with Type2 and save my raw DV to my HD.

If I would like to create a DVD. Should I prepare the films in any way or can I just import it to my DVD authorizing software and create my DVD?

I was thinking of using TMPGEnc DVD Author as I like it simple with lot of templates to create DVDs easy. Is there any other software I should try for my needs?

[PuzZLeR]

If I would like to create a DVD. Should I prepare the films in any way or can I just import it to my DVD authorizing software and create my DVD?

You first have to encode it to a DvD compliant MPEG-2 stream otherwise an authoring program cannot author a compliant DvD from DV source.

Do it separately from your authoring program. TMPGEnc Xpress is very good if you have that too with Author. I personally love CCE, but a good free app arguably as good is HC Encoder.

I would prefer to encode any LPCM streams to AC3 to save a little space. This is subjective.

Once you have the encoded streams ready, just drop them into your favorite author app and burn. Other than the menu making, the rest should be real fast if the streams are compliant

BTW - TMPGEnc DvD Author has a well known bug that says some DvD compliant streams have too high a bitrate. Ignore it. If the streams come from a respectable DvD encoder, they will be compliant.

[edDV]

Thanks for all answares.. I will go with Type2 and save my raw DV to my HD.

If I would like to create a DVD. Should I prepare the films in any way or can I just import it to my DVD authorizing software and create my DVD?

I was thinking of using TMPGEnc DVD Author as I like it simple with lot of templates to create DVDs easy. Is there any other software I should try for my needs?

First I agree DV source should be preserved for future software and displays. You may not see the difference but your grandchildren will with future software.


As for MPeg2, most consumer DV camcorders record exposure a bit hot using 16-255 (often with some clipping at 255). They do this to cheat signal to noise and make them appear brighter in low light. The rest of your videos (broadcast, DVD, VHS etc.) are recorded properly to 16-235 (nominal white) with some overshoots into 236-255.

For this reason I pull nominal white down closer to 235 so that camcorder source looks more like normal video on a properly adjusted TV. This is done before encoding MPeg2.

Avoid the Panasonic DV codec. It converts DV to RGB space and in the process clips off the 236-255 white detail. Instead use the Cedocida (or Sony, Adobe, Mainconcept) DV codec to preserve 0-255 white detail.

[sankaboy]

Should I deinterlace the film too or is MPEG-2 interlace compatible?

I noticed that date and time is saved on my tape. Can I get that data from my DV saves on my computer?

If so, how?

[sankaboy]

For this reason I pull nominal white down closer to 235 so that camcorder source looks more like normal video on a properly adjusted TV. This is done before encoding MPeg2.

Thanks for your suggestion, but how do I do that?

[edDV]

Should I deinterlace the film too or is MPEG-2 interlace compatible?

I noticed that date and time is saved on my tape. Can I get that data from my DV saves on my computer?

If so, how?

Define "film".

MPeg2 handles interlace well. AVC formats are not ready for prime time yet for interlace, except for pro VC-1.

[sankaboy]

How about getting shootdate?

[edDV]

For this reason I pull nominal white down closer to 235 so that camcorder source looks more like normal video on a properly adjusted TV. This is done before encoding MPeg2.

Thanks for your suggestion, but how do I do that?

Most any consumer edit program has luminance exposure control (you don't need to touch color). In Virtualdub there are HSV/HSL (V=Luminance) filters.

[jagabo]

For this reason I pull nominal white down closer to 235 so that camcorder source looks more like normal video on a properly adjusted TV. This is done before encoding MPeg2.

Thanks for your suggestion, but how do I do that?

Most any consumer edit program has luminance exposure control (you don't need to touch color). In Virtualdub there are HSV/HSL (V=Luminance) filters.

Many (and VirtualDub in particular) consumer editing programs convert to RGB with a rec601 matrix for filtering -- losing the highlights he is trying to preserve even before filtering begins.

Assuming the DV source has luma levels from 16 to 255, using VirtualDub with Cedocida he can get the desired levels shift by doing the following:

1) Set up Cedocida so it converts to RGB with the pc601 matrix. In the Decoder section of its config dialog set the Output Formats to RGB24 and RGB32. Set the YUV To RGB conversion to [0-255] --> [0-255]. Now when Cedocida converts to RGB the whites will be ok, but the blacks will be too high. So the blacks have to be pulled down.

2) Start VirtualDub and make sure its internal DV decoder is disabled. Options -> Preferences -> AVI -> Prefer Internal Video Decoders... (unchecked).

3) Open your DV AVI file with VirtualDub.

4) Add the Levels filter. In the Input Levels section change the leftmost entry from 0 to 16. The mid levels are now pulled down a bit to. You can compensate by setting the gamma (the middle of the three input fields) to 1.1.

5) The output RGB levels are now in the full range from 0 to 255. Most compression codecs will use the rec601 matrix to convert RGB back to YUV with levels from 16-235. If you're using Cedocida to recompress the video make sure the RGB To YUV conversion setting is set to [0-255] --> [16-235] so you produce a DV AVI file with standard luma levels.