ADOBE AFTER EFFECTS: Forum Expressions Tutorials Creative Cloud

Working in 8 bit vs 16/32bit?

COW Forums : Adobe After Effects

<< PREVIOUS   •   FAQ   •   VIEW ALL   •   PRINT   •   NEXT >>
James CarverWorking in 8 bit vs 16/32bit?
by on Dec 4, 2016 at 9:23:06 pm

Can anyone explain the type of projects or scenarios where you'd need to work in 16 or 32 bit? I have a general understanding of the difference between the 3 but it's still not clicking when I should be using higher bit rates in AE? Do the additional colors you are able to see in 16/32 bit render out during export as well?

iMac 27" Late 2013
32 GB 1600 MHz DDR3
3.5 GHz Intel Core i7
NVIDIA GeForce GTX 780M 4096 MB


Return to posts index

Darby EdelenRe: Working in 8 bit vs 16/32bit?
by on Dec 4, 2016 at 10:20:51 pm

In my opinion higher bit depth is always preferred. However, it does come with a performance penalty in terms of render speed and memory usage. To that end here are some guidelines for when you'd benefit most from higher bit depth.

High bit depth becomes especially important when your project will include subtle gradations of color. An ideal gradient is a smooth continuous change in color value, but in the digital world there are 256 discrete values in an 8 bit color channel. This results in discrete steps in the gradient ("banding"). 16bpc has 65,536 steps available in each color channel. There are still steps, but they're so close together that we perceive them as being more continuous.

Another scenario where higher bit depth is important is when adjusting color dramatically. Every operation that modifies color will result in a value that needs to fit into one of the buckets o' bits. When working in 8bpc you only have 256 buckets for each color channel, which results in color transformations being rounded less accurately than they would in 16bpc or 32bpc. This problem gets compounded with each additional color transform.

The last scenario where I'd say 16/32bpc is actually required is when linearizing your working color profile. Human vision is more sensitive to darker colors so in the interest of pushing more of the 8 bits available into that darker range we use gamma encoded images. When the gamma is linearized in 8bpc it results in dramatic banding in the low end of color values. If you ever linearize your working profile you must use at least 16bpc.

Lastly, some codecs support higher bit depth outputs, but even if you're outputting to an 8 bit image you'll benefit from working at a higher bit depth. Gradients will be calculated internally at the higher bit depth and then dithered to fit into 8bpc, resulting in a more perceptually continuous gradient.

Darby Edelen


Return to posts index

James CarverRe: Working in 8 bit vs 16/32bit?
by on Dec 5, 2016 at 5:09:10 am

Thanks for that refresher, but what do you mean by "linearizing your working color profile"?

Also is it safe to assume I can work in 8 bit and later change it to 16 bit for the final render without issues? Basically switching back and forth to monitor as I go along?

iMac 27" Late 2013
32 GB 1600 MHz DDR3
3.5 GHz Intel Core i7
NVIDIA GeForce GTX 780M 4096 MB


Return to posts index


Darby EdelenRe: Working in 8 bit vs 16/32bit?
by on Dec 5, 2016 at 6:27:58 am
Last Edited By Darby Edelen on Dec 5, 2016 at 6:42:15 am

There should be no negative impact in only switching from 8bpc to 16bpc. Some effects are not supported in 16bpc and 32bpc, they will appear with exclamation points in the effects panel. These effects render in 8bpc so some of the benefit of the higher bit depth is lost, but it shouldn't look worse than it did in 8bpc.

If you switch to 32bpc (floating point) then it's possible to have values outside of the black-to-white range and so the look has the potential change dramatically.

Linearizing the working color profile is an option in the project settings. For pure motion graphics work this is largely unnecessary, but for compositing work it makes it possible to create more physically accurate effects and composites. The explanation is involved, but suffice to say that a linear profile more accurately represents light in the real world.

Darby Edelen


Return to posts index

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
© 2017 CreativeCOW.net All Rights Reserved
[TOP]