Today’s Question: If I have a 16-bit file that is flattened (the original layered file being saved in 16-bit mode) and I want to save storage space, is there any image degradation if I convert it to 8-bit mode?
Tim’s Quick Answer: There is some degree of image degradation when converting from 16-bit per channel to 8-bit per channel. However, from a practical perspective this is not a significant concern unless you’ll be applying relatively strong adjustments to that 8-bit image, or the image had been converted to black and white.
More Detail: The bit depth for an image determines the total number of tonal and color values available for the image. For most typical photographic workflows, the benefits of working in the 16-bit per channel mode are largely theoretical. However, there are situations where that higher bit depth can be helpful, and so I do recommend working in 16-bit per channel mode for your master image files.
One of the key risks of an 8-bit per channel image is posterization, which results in the loss of smooth gradations of tone or color in an image. For example, a sky that should appear as a smooth gradation of blue may instead have a banded appearance.
In most cases, with a properly exposed photo and relatively modest adjustments, the risk of posterization is relatively low. However, for monochromatic images (black and white for example) or when strong adjustments are required, the risk of posterization can be very real.
So, first and foremost I recommend retaining the original master image in the 16-bit per channel mode. If you’re going to create a flattened copy of that image, and it is a full-color image that you don’t intend to apply any further adjustments to, then converting that flattened copy to 8-bit per channel mode is perfectly reasonable. Doing so, by the way, will reduce the file size for the image by half.
For images that you may still need to apply strong adjustments to, and for monochrome images in particular, I recommend keeping the image in the 16-bit per channel mode. This will help minimize the risk of posterization for the image, especially if you’ll be using a print workflow that supports 16-bit per channel data, for example.