The benefits of using 32 bit processing for compositing are not always apparent to superficial examination. If you create a project in After Effects, drop a few clips in a comp and turn the switch to set the projects processing at 32 bit you will not see anything happening. Of course 32 bit processing takes much more memory and processing time than the standard 8 bit so what’s the point?We know that cameras shoot at 8 bits per channel, that is 8 bits for Red, 8 bits for Green and 8 bits for Blue. The combination of those three values provides enough colors to depict the world as we see it with human eyes. 8 bits is integer processing, 32 bit is floating point. Floating point provides a much finer amount of values than integer calculation. 8 bits means that you have a maximum of 256 values possible for each channel, from 0 to 255 ( 2 to the power of 8 = 256 – 1 = 255 ). While the combination of RGB leads to a high number of colors that can be represented by the screen, the single channels can “overflow” pretty easily. Overflowing means that the software tries to store a number that is higher than the allowed precision and as a result only a portion of it is stored, the portion that didn’t fit is truncated or clipped. Think about your car’s odometer. When you reach the maximum number the odometer “clips” any subsequent number. Say that your odometer has 5 slots. After you drive for 99,999 miles all numbers will roll over and you’ll be back to 00001. Same thing for binary arithmetic. Using floating point allows us to map color and luminance operations in a much more precise way.This becomes useful as we manipulate our images by adding effects like Trapcode shine or by using blend modes like Add and Screen. All these options can cause luminance values to go beyond the 8-bit limit and get clipped. But there is another advantage of 32-bit processing. If you look at the way a movie shot on film handles crossfades and fades to black you will see that the highlights of the fading clip tend to linger on till the end. If you use your typical “Fade to black” effect in your NLE you’ll notice that the effect lowers the luminance of every pixel in the image turning the highlights into greyish mud. Here is an example, click on the image for the unscaled version:
Uncompressed TIFF file originally created by After Effects.
And here is the exact same image at 32-bit precision. Notice how the highlights are much brighter while the rest of the image is exactly the same as before.
Uncompressed TIFF file originally created by After Effects.
It’s important to note that both images have been converted to 8-bit JPEGs, as a demonstration that the storage and acquisition formats do not determine the final quality of the picture, the precision used to process the effects does.Yes, you acquire your footage at 8 bits and you will very likely deliver it at 8 bits. What you do in the middle will affect the final result and having the ability to work at higher numeric precision is fundamental in order to obtain great images or create believable visual effects. Precision is everything in this field. Remember digital, after all, means “based on numbers.” Everything that we do in the digital domain is numbers, if we chop them up with no attention to precision we cannot expect to see good images at the end of the chain. On the other hand when we use higher numeric precision the final result will be virtually indistinguishable from film. That alone is a reason to try 32-bit processing and it’s amazing that we can access that kind of power with a desktop application like After Effects. In the not so distant past this feature was not available to the general public. Another way the digital revolution brings the power of making great movies to the masses.