Canon XL-1, compressions and a computer
 

On another thread I asked about the Canon 14x manual zoom lens and wow what a response - great. So now I have another question. Again it's with my Canon XL-1 and capturing frames or pictures directly into a computer for stop motion. I've read that with Mini-DV the sampling is 4:1:1 and according to Scott Billups book Digital Moviemaking that's not too good, and the compression is 5:1 he also talks about data rate 25 MBS. Alot of this is Greek to me. My question is this. I use the Canon as a way to get the digital image into the computer, saving each image as a pict, jpeg or tiff file. Not compressed for DV. Then I create a quicktime movie in AfterEffects. I save as DV-NTSC. So does all this compression, sampling etc. take place say after I've created a film and printed it via firewire onto a Mini-DV tape. If the same Quicktime movie could be printed to another formatt like say Digi-beta or DVCPRO-50 would the quality be better. Not that I don't think Mini-DV isn't good. To my eye watching a Mini-DV tape, using S video cable into a good monitor looks like DVD quality to me. One of the reasons I ask is I have a grant to start developing a animated video and if at all possible would like to use my XL-1 ( with my new 14x lens) to film it. I'd of course like the best image I can get but don't want to use the money to buy expensive, new equipment. The final product would be on VHS tape for schools but who knows with any luck maybe broadcast at least I'd like to keep that in mind.

Thanks Dan

the following compression stages occur in your scenario:

1. transfer to firewire over your computer is in DV (5:1 compression
has been done here!)
2. saving it as a still. Depending on which format there can be
a lossy (quality loss!!! -> jpeg), lossless (no quality loss ->
targa, tiff) or no compression (bitmap). Make sure YOU DO NOT
LOOSE QUALITY HERE!!!
3. creating a quicktime movie. This is compressed again in DV
format. You will loose another round here....

So... you can have 3 compression steps if your doing it badly
and you will have at least 2. This is not very nice. What might
be a better thing to do for you is to fire up Adobe Premiere and
use its stop motion capture! If I am not mistaken this will save
the frame directly as DV format in your movie which will only
yield you ONE compression instead of the TWO you will get
oterhwise. I have some plans to test this myself in my vacation
next week. If Premiere does this the wrong way I'll build my own
capture application which does it the right way.

Will let you know how the testing goes!

Rob:
Thanks for the reply. Here are a few more questions and thoughts. I save the still images as TIFF files which are approx 1 mg files each. If I take that same image bring it into Premiere, create a quicktime movie (which is only one frame) that is DV compressed the frame is 120k. So is capturing it via firewire to the computer compressing it? If I import a whole sequence of frames (animation) into Aftereffects (this can also be done in Premiere) make a movie with no compression I would imagine that would be saving on one of the compressions too. I have used Premiere to capture the animation but I use another program, BTV pro, that frame averages which produces an image with no noise - it's really clean.
Just to keep my self inline with this form on the XL-1 and not make it too much about non-XL-1 topics I love my XL-1 and it looks really cool with the new 14x manual lens on it.
Dan

You are correct in stating that it is 120K (which it should be for
NTSC). DV uses a fixed bandwidth comression algorithm.

Capturing through firewire is not compressing it. This has already
been done in the camera. Capturing through firewire is merely
transferring the data and storing it inside a file, that is all.

If you save it to an uncompressed movie you will ofcourse have
no extra compression indeed. But you will get a very large file!

I haven't had the time to test Premiere's motion capture myself
yet. I'll see if I can do that tonight and let you know how it goes.

The less extra cycles in your production the better for time, speed
and quality!

Can you tell me how this compression business applies when using Vegas Video 3? I'd like to get the best image possible when outputting (mastering on minidv). What can I do to keep my stuff looking its best?

Okay. I have looked at Premiere stop motion capture as promised
and Premiere does it wrong (at least 6.0 version, I am waiting
for the release of 6.5 to see if they changed anything). What they
do is store each frame as a seperate BMP (bitmap). Not only is
this clumsy and cubersum, it also means you have to do an
extra compression cycle if you want to stay in DV. This will not
happen ofcourse if your output is MPEG for example because
you will have to re-encode anyways that way. I'll see what
happens with Premiere 6.5. If that still does not fix it I'll build
a stop motion capture program myself. If that is done I will let
everyone know.

Josh,

If you are just capturing footage, editing it and then outputting
it all within the same application you do not have to worry much.
The only thing you might loose quality then is in the output part,
but that depends on which format your outputting to etc.

We are here talking about what if you shoot it in one format
(DV) then output to another (still frames) and then put it back
into a movie format (DV). Then you are loosing quality due to
one (or two) extra compression rounds.

What I'm planning to do is shoot stop motion sequences, then capture each "shot" (set of motions from one camera angle) separately, editing it to my satisfaction and putting it back on a minidv tape, possibly later to be transferred to VHS. So you're saying if this is all I'm doing, I'll be fine?

The first stage of compression, ie 5:1, occurs the minute the light photons hit your CCD. The camera processing compresses the data to DV format before it writes it to the mini-dv tape. Nothing you can do will ever change that fact. Once that initial compression has been done by the camera, copying over firewire to your hard drive is nothing more than that...a COPY!! It is a perfect digital copy of the compressed data stream in your mini-DV tape. IN ALL NLE's, even VV3, the digital data is only a copy of the original UNTIL you perform a command like a fade, a title, or some other manipulation of the original image. In this case, the data is uncompressed, the change is made, then the data is recompressed into the DV format. This is called rendering. So, ANYTIME you do a render, you're introducing a slight degredation from the original....sorry there is no escape from reality. AT a minimum, you're stuck with the original compression that happened in the camera....no way around it.

Is this slight degredation noticable? I'd assume not, otherwise everything everywhere would be straight cuts and nothing else.

There are several websites around that show the result of multi-generational re-rendering. I think one of them is Adam Wilt's site. They show that after 10 renderings(or so) the Microsoft DV codec is HORRIBLE....color shifts and loss of rez. The best DV codec that they tested seems to be the Main Concept CODEC, with some slight degredation, but, ya gotta look close to see it. I haven't seen tests for the Sonic Foundry CODEC in VV3, but, I am led to beleive it's better than the MainConcept one. There are those that will claim the AVID codec is best, which may be true, but, the only people who can play it are those who coughed up the dough for the entire NLE package from AVID....not an acceptable solution IMHO.

I think it's noteworthy to remember that, no matter what we got, it's better than the old consumer video tape cameras...that was truly BAD!!

You may want to look here also:

http://www.onerivermedia.com/codecs/index.htm

Martin posted this a day or two ago and it opened my eyes.

Jeff