You can't view a linear, gamma 1.0, file. At some point you HAVE to give it a higher gamma. When you do, you throw away a lot of highlight information. It's a mathematical certainty that when you process a linear capture you throw away this highlight information. In a histogram of a linear file the entire right half of the histogram represents the highest one stop and you'd be hard pressed to tell there's anything there at all by looking at the histogram because the samples are so spread out. What these compression schemes do is throw away highlight information before the file is even stored rather than have it be lost as part of processing. It's not used or needed, so why keep it around.
Jan, if you're pushing for the absolute-best quality you can get and are using ETTR intelligently, then you've filled up that highlight area as much as you can with an overexposed image that you'll then be bringing back down in post. If so, then some of that data is most certainly used and needed. That's one reason to keep it around.
Not everyone is doing that, and I am not arguing that everyone needs to keep that data. I am simply saying that the opposite (not needed, and not necessary to keep it around) is not always true.
It also depends upon the scene and the likely amount of enlargement of the image. Much less likely to see any tonal variation and detail loss at print sizes 16x20 and smaller with most subjects and scenes and "normal viewing distances" (which is not 12 inches).
The end product is a print and the gating factor is the printer and the conversion of the image data to the data file that is used by the printer. I doubt that in most situations that even 14-bit is going to make a visibly noticeable difference in the finished prints.
elkhornsun wrote:
I doubt that in most situations that even 14-bit is going to make a visibly noticeable difference in the finished prints.
That's not the benefit of shooting 14bit though. To my knowledge, there's no such thing as a 14bit printer anyway. The benefit is the flexibility you gain in post processing. Push and pull it in LR, get it to look correct in the printers color space, and if done correctly... you should see a difference in your final prints.
