OK you knew it was coming 4:4:4 vs 10 bit

O

optitek

Well-known member
both full sampling and greater digital color depth improve the color fidelity.
The popular opinion here seems to be that 10 bit gives better results over 4:4:4 full sampling.
Does anybody here have a technical explanation?
 
10 bit = far more gradiations available between colors, more colors to deal with. This means you get far more color data to work with. (as to encoding of the signal - 444 at 8 bit is just a very clean representation of a limited color space. So while it's nice to have, it doesn't add too moch over 4:2:0 which is virtually lossless to the eye anyway.)

8 bit color = 16.7 million possible colors - 226x256x256 R/G/B 10 bit = 1024X1024X1024 = 10 Billion + possible colors.

More colors = more to work with.

Check this page out: http://blog.abelcine.com/2011/05/18/hd-formats-bit-rate-vs-bit-depth/
 
Furthermore, the eye can't resolve differences in color nearly as well as it can resolve differences in brightness. The difference between 10-bit and 8-bit is a lot more noticeable (probably dramatically more noticeable) than the difference between 4:4:4 and 4:2:2, at least visually. Mathematically there's advantage to 4:4:4, and in heavy VFX/compositing there's an advantage to it, but in termis of actual eyeballs being able to tell the difference it's very slight, and likely bordering on imperceptible.

Look at it this way: 4:4:4 gives you twice as much color sampling as 4:2:2. 10-bit gives you 4x as much luma sampling as 8-bit. So just between those two, obviously 10-bit gives you more information to work with. Then, furthermore, the TYPE of information it gives you is more perceivable to the human eye.
 
Barry_Green said:
Furthermore, the eye can't resolve differences in color nearly as well as it can resolve differences in brightness. The difference between 10-bit and 8-bit is a lot more noticeable (probably dramatically more noticeable) than the difference between 4:4:4 and 4:2:2, at least visually. Mathematically there's advantage to 4:4:4, and in heavy VFX/compositing there's an advantage to it, but in termis of actual eyeballs being able to tell the difference it's very slight, and likely bordering on imperceptible.

Look at it this way: 4:4:4 gives you twice as much color sampling as 4:2:2. 10-bit gives you 4x as much luma sampling as 8-bit. So just between those two, obviously 10-bit gives you more information to work with. Then, furthermore, the TYPE of information it gives you is more perceivable to the human eye.
Click to expand...

Said it better than I could. :)
 
Thanks all for the input- great info and I think many here will benefit.
Let me play the devils advocate here:
If the perception of image quality is primarily based on contrast than cleaner colors (red and blue since they are fully sampled ) with less noise going deeper into under exposure could possibly give perception of more dynamic range which IMO is the close second component for perceived image quality ahead of resolution.
So the bottom line here is: (a lot) more muddier colors vs smaller number of clean noise free colors.
BTW I agree with Barry's iron clad logic just wanted to throw this up here for consideration....
 
I also agree with Barry. I'd much rather have 10-bit than 4:4:4...which is why the HPX-250 with an external recorder is going to be just swell. :happy: (not to flame bait or anything...just what I'm looking for my next upgrade to be...FS100 looks like a great camera).
 
I thought staying in RGB plus a LUT would help with grading in post. Other than that, not sure what the benefit is.
 
Gary Huff said:
I also agree with Barry. I'd much rather have 10-bit than 4:4:4...which is why the HPX-250 with an external recorder is going to be just swell. :happy: (not to flame bait or anything...just what I'm looking for my next upgrade to be...FS100 looks like a great camera).
Click to expand...

The AF100 should have been the HPX-250 with a 4/3 sensor. THAT would have been awesome.
 
Well we just tested the F3 onboard 8bit 420 vs 10bit 422 kipromini. The color space space dffernce between the two is nominal. The luminance differnce is substantial. You gain so much in the highlights, shadows and you get a smoother silkier look all around.

This jump isn't f3 centric. It will happen wth any true 10bit signal being recorded @ 10bits. The jump up to 10 is the key. Shoot, the epic records in 12bits! It shows in the footage....
 
TimurCivan said:
Well we just tested the F3 onboard 8bit 420 vs 10bit 422 kipromini. The color space space dffernce between the two is nominal. The luminance differnce is substantial. You gain so much in the highlights, shadows and you get a smoother silkier look all around.

This jump isn't f3 centric. It will happen wth any true 10bit signal being recorded @ 10bits. The jump up to 10 is the key. Shoot, the epic records in 12bits! It shows in the footage....
Click to expand...

Makes sense and is one reason to get the F3 if one can afford it and an external recorder (many of us can't unfortunately). Ultimately, more data counts.
 
You must log in or register to reply here.
Back
Top