Bit depth/color space impact on sharpness?

R

Rick LaLonde

Active member
Hello all, I'm on a quest to discover the answer to this. For example, would 1080 12-bit 4:4:4 yield a sharper image than 1080 8-bit 4:2:0?

Any input appreciated.
 
Bit depth and colour space aren't going to affect sharpness, it's compression artifacts that will. And as a general rule, 8-bit 4:2:0 codec are vastly more compressed than 12-bit 4:4:4 ones. So if you're seeing significant differences in sharpness between the two, compression is most likely to be the cause.
 
Bitdepth does not matter for sharpness however 4:4:4 is sharper than 4:2:0, color 'leaks' with 4:2:0:

d970543fe7b7e86fb63c708a1072ced5.png
 
Thanks, Cary. I suspected this but wasn't sure. It seems to me that the color separation of 444 would in some instances give a perceptible increase in sharpness for example, the side of someone's face against a black background having a harder line of separation. I wouldn't expect the difference to be a big one.
 
Rick LaLonde said:
It seems to me that the color separation of 444 would in some instances give a perceptible increase in sharpness for example, the side of someone's face against a black background having a harder line of separation. I wouldn't expect the difference to be a big one.
Click to expand...

Typically in video the idea of "sharpness" is more closely associated with resolution. With more color information, you're talking about visual contrast, which is more closely associated with the concept of acutance.

Basically, the more difference there is at an edge boundary, the easier it is to see. That difference can be value (dark vs. light), it can be hue (eg. orange vs. blue), it can be saturation (has to be a large difference to make it useful), and it can be slope (how many pixels wide is the transition? Fewer pixels between black and white make the boundary easier to see vs. a lot of pixels between black and white). There are undoubtedly many more ways to create visual contrast.

Then there's noise (remember how people said that film grain made everything seem much sharper? That's a noise effect). Not to mention your optics.

All this stuff adds together to give your impression of resolution. Simply using a CODEC that makes chroma sub-sampling bigger (4:2:0 vs. 4:4:4) doesn't mean that what you are filming will look as if it has higher resolution. What it means is, if your scene can take advantage of the increased color space, it's available. Not the opposite; not because it's available your scene will be forced to use it. Besides, one of the first things that happens in compression is the loss off the full 4:4:4 color space -- the first thing most compression algorithms do is lop off those "extra bits". That's why AVCHD was originally just 4:2:0.

I'm just sayin' that chroma sub-sampling is just a (small) piece of the puzzle. Don't put too much emphasis on any one measurement. Because Canon, Nikon, Sony, Panasonic, Red, Arri, etc. etc. etc. don't. What the camera manufacturers are trying to do is give you the best balance that they can. Which is why you don't see them offering 4:4:4 much in the consumer and pro-sumer lines.

The bottom line seems to be that proper lighting is more important than camera specs. It's amazing to me how good lights in the correct places can make a mediocre camera look razor sharp.
 
Rick LaLonde said:
Thanks, Cary. I suspected this but wasn't sure. It seems to me that the color separation of 444 would in some instances give a perceptible increase in sharpness for example, the side of someone's face against a black background having a harder line of separation. I wouldn't expect the difference to be a big one.
Click to expand...
It is indeed minor factor.
 
You must log in or register to reply here.
Back
Top