Understanding Color Sampling

WoW! I just read through that! It has totally opened my eyes to what all of that stuff means! Thanks!
 
Simply fantastic article. Color sampling was one of those technical things I always assumed I would just never fully understand. Thanks for writing an article so even people like me could get it.

Chris
 
Barry - Rods and Cones. Rods are your dark vision. They don't really play a part in normal daylight vision as they're saturated and don't add to the visual picture until it gets really dark. Cones come in three kinds, L, M and S, for receptive to long, medium and short wavelengths. They give us our colour and brightness vision under normal viewing. In the fovea, or central section of the eye, we have only cones present - no rods. There is debate on the L:M:S cone ratios, but is thought to be about 6:3:1 Our black and white "luma" vision comes mostly from both the L and M cones. So we do end up with lower chroma resolution, but it's not based on a rod / cone ratio, but on a L:M:S cone ratio.

Gosh, you can tell I've been reading up on way too much colour science stuff recently!

The example are great Barry - must have taken a while to figure out all the chroma averages. However, any correctly engineered video camera can't do such an abrupt 4:4:4 luma/chroma transition between neighbouring pixels due to the optical low pass filtering etc. Interlaced formats can also contribute to chroma problems, especially with 4:2:0, but they also, by their very nature, have lower vertical chroma and luma resolution anyway, which can make 4:2:0 a better choice in some ways, especially in the PAL world of SD broadcast. Of course, this is rapidly becoming less important.


And thanks for the recommend - The filters are in Film Effects and G Chroma Sharpen uses quite a clever luma adaptive chroma fixing method.
 
Well, I'll admit my high school biology was a long time ago... :)

The color sampling wasn't created by a camera, it was done by the codecs. And yes, interlaced chroma on 4:2:0 can be most hideous; the article was meant to be introductory, so I kind of sidestepped the whole interlaced chroma nightmare...
 
Well, I'wouldn't have known about cones and rods either until I recently had to swot up on colour stuff. :-)But yes, article meets the goal of being an intro to chroma sampling and gets everyone thinking of the nasties it can do. It's funny though, that if the codec doesnt' average, but just "picks" one colour from the 4 in 4:1:1, you could do chroma reconstruction to produce a better result than you'd get from the averaging. Also, for the most part, when you're up at a decent bit rate, it makes more sense to just compress the chroma more heavily rather than sub-sample it.....

Graeme
 
I make the banners Capt... that was some photoshop gradients with a pixelation.

nice point Graeme about compressing chroma.. im surprised more people dont do it.
 
WoW! That was amazing! I've had people try to explain this to me before and my eyes just rolled in my head! That was SO CLEAR! Thank you!

So.... does that mean that there will be no "averaging" of the pixels in 4:4:4 (as in the Red Camera)?

BTW, I LOVE that banner!!! Very nice!

Thank you, Gentlemen! Well done!

-Ted
 
Last edited:
Theodore J Arabian said:
So.... does that mean that there will be no "averaging" of the pixles in 4:4:4 (as in the Red Camera)?
Click to expand...
If it's recording 4:4:4, then correct, there will be no averaging of the color values.
 
You must log in or register to reply here.
Back
Top