DV Editing vs. Low-Compression M-JPEG Editing
 

I am about to set up a Digital Editing workstation and was all set to go with a full start-to-finish DV setup for documentary work, but I don't want broadcast options limited by technical specs of how I made the piece (PBS requires Digital Beta or D-5 masters, don't know if cable channels are this strict or not). Then I read this, posted by Adam Wilt, a very knowledgeable video engineer, on his website (www.adamwilt.com)...

So now I have two questions regarding editing in dv versus editing in low-compression M-JPEG (or MPEG-2 or DV50) before going out to D-9 or DigiBeta tape (all source material will be shot in DV):

1. Is there really a meaningful quality difference in the final product?

2. Is there a big cost difference?

Thanks!

I would say it depends on the complexity of your post production chain.

Will you be doing chromakeying, multi-layered compositions requiring more than one re-rendering (i.e., performing an After Effects comp, then using the rendered comp as an element within another comp), or CG elements? Or, as your quotation suggests, will there be sharp graphics, a la a PowerPoint presentation with black text and lines on a white background?

If the answer to any of these is no, you're just as well off keeping everything in native DV. Lossiness is only really a factor when you'll be doing multiple re-encodings on the material, and I can imagine very few cases in which that would occur.

I'm not even sure what NLE packages support low-compression MJPEG or MPEG editing. There will be additional memory, hard drive space, and processing overheads to contend with, increasing your costs if your intended project is long (e.g., a feature).

Matrox Digisuite supports lossles mjpeg video editing. It supports 8bit 4:2:2 color also (DV50).

Hi there,

as Adam Wilt says, shooting DV and editing MJPEG or lossless is okay, as long as you realize you will be limited by your original footage. This means 4:1:1 for NTSC, and 4:2:0 for PAL DV. It is very difficult to improve this footage in colour resolution in post. Period.
If however you are doing multi-layer comps etc then lossless or low-compression MJPEG is definately worth looking into, especially if you'll be layring graphics.
If you can live with not having totally real-time previews etc you can get lossless compression editing very easily by converting your footage to HuffYUV before you import it into your editing app (assuming you're not using ExpressDV). Once you've got your footage as huffyuv you do need a lot of space though (I think it compresses 3 or 2.5:1, so you're looking at close to 8 or 9 Megs a second).
4:1:1 and 4:2:0 do impose problems not just when doing multiple re-encodings btw. They also mean that your colour resolution is reduced, which in laymans terms means colour fringing, or slight colour blocking around high-contrast edges. This results in really difficult to control dissolves sometimes, and really gives you headaches when doing chroma keying. There are ways to alleviate the problems, but try to use media that are appropriate for it's use - this means stay well away from DV if you will be doing keying...

Oh, and you can find huffyuv at http://math.berkeley.edu/~benrg/huffyuv.html and ... it's FREE!

Kai.

I've only skimmed the specs, but huffyuv looks good. What can you use to to recode DV? And which popular NLE packages is it compatible with?

Huffyuv is compatible with anything that uses vfw (Video for Windows) codecs. Premiere springs to mind, so do most other pre-DV era editing and compositing packages.
What you would have to do is one of three things;

a) you devide you have looooaaaaads of disk space to play with - you would transcode your footage into huffyuv avi's before actually importing into Premiere et al. In Premiere, you make sure that in your editing prefs you choose the Huffyuv codec also, so that multiple transcodes won't touch your images at all.

b) you decide you have some disk space to play with, but not enough to have an orgy on; in this case you would probably import your material as DV, but set the editing codec to be huffyuv anyway. This means not using one of the standard DV presets, and building your own VFW preset from scratch. You will loose some quality, but if you're not doing much with your footage this is actually identical to what you do in a)

c) you're just playing aroung; well, do exactly that - play around, see how your codec settings affect your quality. There is much to be learned that way, and much time to waste doing this learning also (I know - I've been there and transcoded countless bits of DV footage...)

Please remember that all of this takes you up to, but not past, outputting your final product. This is a whole nuther story and is probably bigger by quite a lot in terms of know-how needed to do it right, let alone well...

Also, please remember that native DV editing apps will not benefit from doing any transcoding. Avid Express spring to mind here, as all it will cause you to do is wait for even longer for Avid to import your footage, having wasted a humongous amount of diskspace in the process.

Make sure (!) you use VirtualDub http://www.virtualdub.org to do the transcoding for you. It's pretty straightforward to learn, has a few really helpful filters, and again, it's FREE.

HTH,


Kai.

Please see my post in the Vegas DV codec selection.
The Black Magic Quicktime codec is available for Quicktime for Windows users for free. It's a lossless 10/8 bit 4:2:2 codec that installs as a Quicktime plug-in. Any NLE and/or Comp app that uses Quicktime can take advantage. AE, Premier, Vegas Video...
all can use this.