What's the difference between 4:2:0 and 4:1:1?

DavidBeier

DavidBeier

Veteran
Another thread said something like this:

4:2:2 - which the HVX200 samples color 50% of the time.

4:2:0- which the HDV cameras (Canon XLH1) and the JVC HD100 sample color 25% of the time.

4:1:1-which mini-DV uses samples color 25% of the time.

If this is the case, what's the difference between 4:1:1 and 4:2:0? For that matter, what's the difference between mini-DV and HDV?
 
4:2:2 = 100% luma-information (details), 50% (every second pixel) chroma-information (color) is recorded.

4:1:1 = 100% luma-information, 25% (every 4th pixel) chroma-information is recorded.

4:2:0 is actually alternating 4:2:2 and 4:0:0, meaning odd lines are recorded 4:2:2 (50% chroma) and even lines 4:0:0 (no chroma-information).

I believe PAL DV is 4:2:0 and NTSC is 4:1:1.

The human eye is more sensitive to details than to colors, so it makes sense to compress the color.

mini-DV or DV25 is a 25 mbit/s codec, where every frame is recorded discretely.
HDV is a codec for recording HD-pictures on mini-DV tapes. It records group of pictures (GOP) like MPEG, which saves bandwidth, but is less reliable, less easy to edit and doesn't look good when you're shooting fast or complex moving objects (e. g. fireworks, cars).
 
MarcusX is 99% right, but there's one error in there..

4:2:0 is actually alternating 4:2:0 (4[luma], 2[blue], 0[red]) on one line and then 4:0:2 (4[luma], 0[blue], 2[red]) on the next.
On average it's still the same as 4:1:1 (from a distance). - 4:1:1 just goes 25% of each color on every line. 4:2:0-4:0:2 does 50% of each color alternating on each line. (one line has 50% blue data, the next has 50% red data.)

- Mikko
 
There are actually two different types of 4:2:0, so it gets even more confusing.

But, in the simplest of terms, both 4:1:1 and 4:2:0 record one color sample for every four pixels.

4:1:1 does it like this:
AAAABBBB
CCCCDDDD

4:2:0 does it like this:
AABBCCDD
AABBCCDD
 
Barry's explanation is correct, but let me add some extra details.

Within the USA video, 4:1:1 is one color sample over 4 pixels in the horizontal direction.

USA 4:2:0, we have one color sample over 4 pixels, BUT the sample goes down one pixel and to the right one pixel. My feelings are that this is better than 4:1:1, because the samples are placed in a 2 pixel by 2 pixel area, rather than a 1 pixel by 4 pixel area. Thus, 4:2:0 is more uniform in both directions.

One interesting problem with 4:1:1 DV25 is when it's recorded to DVD it acts like 4:1:0. DVDs use 4:2:0 and the one color sample per 4 pixels of 4:1:1 gets dropped to one color sample per 8 pixels.

Still, given an HVX, why record in DV25? DV50 is 4:2:2 and it kicks DV25's ***. I would want to come into an edit system with the best master possible.


Bob Diaz
 
...and the difference beteween 4:1:1 and 4:2:0 is only really noticed why you try and greenscreen with them - the shape of the ugly jaggies changes! :huh:


EDITED TO ADD: Interestign about the color sample offset Bob. That recorded 'sample' is an average of the sampled raw data from the surrounding pixels, so yeah, by changing the group a little it should look better. Cool. You eluded that only NTSC 4:2:0 uses this offset, what about PAL?

- Mikko
 
Last edited:
I believe Greame Nattress, maker of the "G nicer" FCP chroma upsample plug-in, has stated that it is in fact easier to up-sample 4:1:1 than 4:2:0.

But maybe this referred only to interlaced footage?
 
BobDiaz said:
Still, given an HVX, why record in DV25? DV50 is 4:2:2 and it kicks DV25's ***. I would want to come into an edit system with the best master possible.


Bob Diaz
Click to expand...

I plan to only record in DV50 when I get an HVX200. I was just curious.
 
You must log in or register to reply here.
Back
Top