DrPablo
Offline
• • •
Upload & Sell: Off
|
 p.2 #5 · Dynamic Range on Canon D-SLR's | |
I don't know about huge losses under normal conditions, but Thomas is right that even hue and saturation editing is 'destructive' in the same sense that levels and curves are (i.e. you get compression and separation of certain levels and you can get blackpoint or whitepoint clipping). The difference when you edit color is that these changes happen to individual channels on the histogram rather than the RGB (or Lightness) composite. The saturation slider is exactly the same as the contrast slider, except that the changes happen differentially in each channel.
Thomas, as far as 16-bit editing goes, I also edit almost exclusively in 16-bit (and in huge color spaces like LAB and ProPhoto RGB). However, you'd actually be hard pressed to show unambiguous evidence of where it makes a difference. By this I mean editing in 16-bit space per se, not shooting in RAW (which is a different issue, and RAW images need not deliver you a 16-bit file). Dan Margulis, whom many consider to be the Yoda of Photoshop elaborates about this, that testing has failed to show a meaningful difference in which bit depth you use for editing. While there may be some cynicism in that, I do think you'd need to show a fairly extreme edit in order to demonstrate a visible difference between 8-bit and 16-bit editing.
The dynamic range of a single capture is easily held in 8-bit space, and 8-bits is more than enough if you don't do much editing. If you want to edit to any degree, 16-bits are more than enough to compress and separate certain tones without posterization. If you're talking about actual HDR, then you need 32-bit space to contain all that information (from multiple bracketed captures), but then again the 32-bit space is clearly just a temporary holding area where you organize the exposure information for a 16-bit conversion.
In the future we may someday have true HDR cameras (which I actually think would take a lot of the fun and creativity out of metering scenes and choosing exposures). This might require a direct 32-bit capture (with all the memory and processing requirements that go along with that). With enough processing capacity, for instance, a 1/60 exposure might allow the camera to calculate and assimilate the info in the first 1/4000, 1/2000, 1/1000, 1/500, 1/250, and 1/125 of that total 1/60. So you'd end up having a 32-bit file with the equivalent of 7-full shots worth of exposure information. Or if shutter speed is a critical component (i.e. for moving water) you could use a sensitivity-priority that can calculate fractions of the total light received in the equivalent of ISO-stops, all with the same shutter speed. This might be a true HDR camera, that like film would allow you to expose for the deep shadows without fear of losing highlights -- and overexposing shadows is a great way to eliminate noise. Of course this is a sort of a fantasy camera, and who knows if it will ever exist.
I also believe that the art of photography is 99% about making choices. So a true HDR camera might be sexy, but it might actually be antithetical to the art within photography. It would pull the decision-making process out of the original scene and stick it on your computer -- because why make choices at the scene if you can just capture everything? You could just do it all in post processing. That doesn't seem very fun to me, because for me photography is mostly about being somewhere and finding a way to capture that place -- there's a joy in looking at things in a way that demands evaluation, interpretation, and choice. And also, there's simply no way a digital file or silver halide emulsion can hold everything in a scene -- a digital file may hold 15 stops of exposure information, but that's vastly different than standing in a scene where 15 stops of actual light are there for your rods, cones, and optic nerves to behold.
|
