To do this we are asked to take any JPEG or TIFF and first convert it to 16 bits per channel using Photoshop. I did a bit of googling to find out how to do this in Lightroom before posting to the OCA Photography forum for clarification. It turns out LR processes in 16 bit and exports JPEG at 8-bit. JPEG bit depth cannot be adjusted (so far as I can tell), but TIFF can be set at 8 or 16. So for the purpose of this exercise I exported a 16-bit TIFF. This is the image, a souvenir collection from a student who went to Mecca for Hajj.
We are then asked to use the curves tool to make a right-rising curve similar to the one displayed in the course notes. This maneuver essentially recreates the initial sensor capture, which as can be seen is quite dark.
Finally, we are tasked with creating a curve approximating the original image, which as you can see came out as a close mirror image (and a bit lighter than the original).
There appear to be at least two implications here.
Exposure setting. The standard wisdom is to expose for the light in high contrast situations. The results of this exercise suggest why. In linear capture (before gamma correction), more data is collected for brighter tones than for darker. When trying to recover shadows, there is less data to work with and thus greater opportunity for noise. By exposing just enough to prevent highlight clipping, we can preserve as much data as possible for dark tones.
Post-processing auto-correct. Some people feel that corrections should be made manually. But this exercise shows that at the level of gamma curves we already accept and use auto-correction. In order to be consistent about the use of auto-correct in post-processing, should we not then make manual adjustments to gamma curves?
A lengthy discussion of these issues ensued after posting my inquiry the OCA photography forum. One useful document referenced (thank you, Vicki) is from Adobe discussing the effect of linear capture on exposure.
#
No comments:
Post a Comment