DV codec and YUY2

[Sillyname]

If the DV codec is ~ 5:1, then why are you guys salivating over higher data rate captures? Isn't the video that you get out of your camera at any given stage along its D/A conversion 5:1 quality whether or not you capture using IEEE1394?

------------------
Deep is not the root word of depression.

[Dr Mordrid]

Not all of us are using DV cams to export video for our YUY2 captues. I use a Sony Hi8 through the S-Video port. The Marvel captures YUY2 off this video stream.

I find this advantageous over using DV because contrary to what many believe DV is a lossy codec. The only thing lossless about it is in the transfer of the video files from the cams tape to the hard drive. In the editor it's really not much better than MJPeg at the same ratio. In fact under certain circumstances when it could be worse;

1. DV has a reduced colorspace (NTSC=4:1:1, PAL=4:2:0) compared to the 4:2:2 colorspace of the YUY2/RGB captures you've been reading about. This means there is a 30% or more likelihood of errors in color reproduction when adding overlays, bluescreening or other compositing effects. Check here for examples;

http://www.adamwilt.com/pix-artifacts.html

2. DV can and will show recompression artifacting when effects are added. DV does have a bit of advantage over MJPeg in that it has minimal error masking mechanisms, but it isn't that great.

This can be greatly improved if DV is processed in a device like the RT-2000 that uses a 4:4:4 RGBA compositing engine, but most DV devices don't have that advantage.

3. There is a great deal of variance in the quality of the various DV codecs. Much of this again comes down to errors in color reporduction, with some showing a marked green discoloration and vertical banding after a few recompressions. Look at this page for examples;

http://members.home.net/dgcom/MiniDV/DVcompressors2.htm

Dr. Mordrid


[This message has been edited by Dr Mordrid (edited 29 August 2000).]

[Meek]

Yes, and if your source is DV, you'll of course get the best quality when capturing with DV, but for edit time recompressions you could use RGB or YUV2, that way you do not get recompression artifacts before the last stage, outputting back to tape in DV format.

M.

[Frankie G]

Doc - where can I go to learn what this "4:1:1" and "4:2:2"
colorspace nomenclature means? I've been to AVI FAQ,
rec.video.desktop FAQ and have not found a good
explanation.

I believe in RTFM, so if anyone points me to a good FM,
I'll be more than happy to R! Thanks.