10-bit vs. 8-bit

Visual Quality Difference Between 10 bit and 8 bit precision

A question often arises about whether there are benefits in using 10-bit processing in post-production if your source images are 8 bits in depth from sources such as HDV (or even 8-bit uncompressed for that matter).  The short answer is YES, if you are applying any significant effects to your images, then maintaining 10-bit precision in post will result in much better visual quality for your end product.

When performing operations like color correction, compositing, effects, titling, etc, and especially if you have material with smooth gradients (i.e., blue sky) you will be manipulating your 8-bit source data – adding, multiplying, etc.  Each time you render the source data, precision is truncated back to 8 bits.  The 8-bit truncation operations can introduce “banding” which is commonly visible in smooth gradients.  Instead, if you convert your 8-bit source data to 10-bit data prior to manipulation in post, your gradients will remain smooth during heavy effects work.  Think of it this way, because of the extra two bits you have additional discrete steps in a 10-bit image between each discrete step in an 8-bit image.

To observe, click on the image below.  It should open in a new window.  You should view the image at 100%.  Notice the following:

•  the smooth sky gradient in the 8-bit original image

•  Two 8-bit gamma operations were performed on the uncompressed version of the YUV source image resulting in Slices B and D respectively.  Notice the clearly-evident banding in slice D.

•  Slice C represents an export to a 10-bit CineForm file after application of the initial 0.4 gamma operation.  Slice E results after applying the second (2.5) gamma operation to the 10-bit CineForm source file from Slice C.  Notice the sky is completely smooth.