4:2:0 - whats that all about?
 

I shall probably regret asking...and many of you wondering why I am! but I've seen several posts referring to 4:2:0 and 4:2:2 what does it mean / refer to?

Thanks

You can read up on chroma subsampling here: http://en.wikipedia.org/wiki/Chroma_subsampling

Basically it signifies compression of the color data in a video stream.

George/

WOW! I wish I hadn't asked :-)
I hope I never need to know :-))

Dear Phillip,

I will try to help.

In order to compress a video image, so that it can fit on a tape, or be broadcast in a limited bandwidth environment, certain compromises are made.

Someone figured out the humans are more sensitive to brightness levels than color.

So, in order to compress the large volumes of data, some of the color information is removed. How much is removed, depends on the compression scheme.

Since this gets pretty technical, most people just use these "Chroma Subsampling" codes, such as 4:2:0 and 4:2:2, and 4:4:4 as as way to communication easily.

The following is an over simplification, but I think it is technically accurate:

4:2:0 video can look nice, but is more of a compromise than the following.

4:2:2 has more detail and more color information.

4:4:4 has even more detail and all of the color information.

Thanks Dan, I kinda got that from the link George supplied, it just seemed a bit too technical for my needs and abilities.

I must say your explanation was rather more precise than the wiki explanation - thankfully

Dear Phillip,

You are most welcome.

I do not know if mine was more precise, but it was more concise.

Haha... Well ask a technical question, get a technical answer (or a link).

So, it has to do with how much of the original images' information is present in a video format.

4:4:4 R'G'B' is typically used on the very high-end as a film alternative and is not subsampled.

4:2:2 This is considered broadcast acquisition quality and is also used for many productions with 'less than studio' budgets

4:2:0 Is used by many standards like MPEG/MPEG2, PAL DV and HDV, but is by some considered to be less suitable for high quality acquisition and more of a delivery format (DVD, PAL/SECAM TV). That said, 4:2:0 is pretty good when you have a high enough bitrate (HDV and up) and is then generally accepted as broadcast quality.

Other subsampling schemes like 4:1:1 (NTSC DV, DVCAM, DVCPRO) and 3:1:1 (HDCAM) are still in use but less relevant here in Europe. 4:1:1 is sort of on par with 4:2:0 and more common in the US for its use in DV.

Other things that make up the image quality are:
- Bitrate: HDV is 25Mb/s, the Sony EX HQ is 35Mb/s, some broadcasters 'require' 50Mb/s
- Codec: Efficiency of the codec and supported resolution renders different (quality) results at the same bitrate
- Color resolution/bitdepth: High end broadcast camera's have sensors that actually see 14-bit values and can record 10-bit values for each color. A typical HDV camera records in 8-bits. Every bit doubles the number of color values. So 8-bit is 256 shades of Red, Green and Blue and 10-bit is 1024 shades (4x as much).

So now you know.

George/

P.S. you may think I forgot about the SENSOR... but that's a whole different story altogether.

Dear George,

Thanks.

Nice Post!

Yeah, Thanks George, you can stop now - go have a lie down you deserve it!!!!!!

I am dusting off my old standard 8 Eumig - so much easier!!! in those days

That wikipedia link is quite helpful. That's where I learned about chroma subsampling several months ago when I started getting into video.

Want to make a quick correction/addition to a comment earlier about most 4:2:0 HDV being accepted for broadcast. While that may be true over in Europe, its not here. Most broadcasters require 15% or less of your package to come from a 4:2:0 HDV camera--partly because of the chroma and partly because of the low bitrate of HDV.

Bottom line?
 

Hi George,

Excellent summary-- and thanks. But in the pursuit of the highest quality, are we talking about parameters that only a trained eye would notice? For example, some audiophile friends of mine are constantly debating the merits or one audio system over another. They talk about purity of frequencies which only the most perceptive canine could possible hear.

Given the different compression methods being discussed here, is it likely that the average (whatever that is) viewer could really tell the difference in a real world application?

Tim

Really depends on the footage, and the camera. But it can certainly be noticeable. Might not notice much filming static subjects with little camera movement. But put the camera on a dolly, or start filming faster subjects and the difference become more obvious.

Well, it is actually not about that...

For example a digital HD broadcast may only be the equivalent of a ~1-2Mb/s stream. With DVD and HD-DVD/BD the bitrate (ie. detail) can be much higher (Blueray tops out at an AV bitrate of 48 Mbit/s). And 4:2:0 is fine for delivery to viewers. However this is the delivery side of things. And we're at the other end of the 'pipeline', in acquisition, with possibly a whole gauntlet of processing between us and the viewers.

In the end the technical stuff is just a very small part of a production, and you make choices based on what is required to get a compelling story to the audience.

If you're shooting a straight up documentary with basic editing your acquisition format can be a lot closer to the delivery format(s) than if, for example, you were shooting a project with (lots of) color correction and chromakeying. But even if you shoot at 8-bit bitdepth you can edit and do effects at 10-bit, so it may not even be a problem.

George/

It's mostly the lower bit rate and small sensor size of typical HDV cameras that broadcasters don't want. XDCAM HD is 4:2:0 @35mb and is acceptable to broadcasters. That extra breathing room in the data rate allows for less compression artifacts in high motion scenery.

-gb-

as well as the VBR encoding instead of CBR in HDV.