rwatson767 wrote Why are some functions 8 bit only? This degrades the image... |
First, the question is if that degrades the digital representation of the image or if it degrades the image you are looking at.
Your display, your printer... and your eyes can't tell the difference between your carefully edited in 16-bits image and its converted to 8-bits version at that final stage. The importance of 16-bits precision lies in the degradation that can be produced by stretching the tones so much that you can perceive 'posterization', that is visible differences between contiguous and nearly monochrome areas and gradients like skies or stage backgrounds. The eye can perceive the boundaries between those areas if their digital values are farther apart. A smooth gradient from your camera jpeg (8-bits), let's say 8 to 12 on the 0 - 127 range of one RGB channel can easily be edited abd streched to 4-40; that means you are working with 8 times less 'steps' (5 bits precision). That's the result of editing. You can test it easily on a black to white gradient picture to appraise that difference visually; the 'posterize' filter is a good tool to play with that.
If you shoot jpeg, you start with 8-bits precision.
If you shoot raw, you start with the 12 - 14 bits precision or your sensor.
When you edit in the raw converter, you are working internally in 16-bits precision for all calculations, so there is very little risk of creating posterization due to stretching tones. Even originally 8-bits jpegs do benefit from that ACR conversion... at least in theory (there has been a challenge to submit a workflow to prove that and I was not very successful in the process).
If you keep in mind that avoiding posterization is the single real risk and that it is creating by stretching tones between contiguous areas, it's easy to see which kinds of edits are able to stretch tones excessively, such as strong curves or gradient changes, layer blend modes like difference, filters like equalize... All those edits can be cared for in the ACR module or in the editor itself with its available 16-bits tools.
The risk is greater if you are working in black and white, only 256 values instead of the 256 x 256 x 256 values provided by 8-bits per RGB channels.
Another important factor is noise. With noisy images, the boundaries between posterized contiguous areas is much less visible. That's why adding noise is the second tool to avoid posterization in gradients, beside 16-bits editing.
The last idea is that to be able to see posterization, you need contiguous areas to perceive the boundaries. That never happens with local tools such as spot healing or sharpening. No advantage in 16-bits for those tools.
To summarize: Elements provides everything you need to edit what is necessary in 16-bits, especially the ACR module.
If you want 16-bits for (nearly) all your edits, use the full Photoshop.
1
Reply
AdChoices