I've met the extremes where everything must be shot perfect in camera, and you should only ever need to shoot JPEG becuase why on Earth would you ever need to adjust anything... ever.
I've also met the extremes where everything can be Phothoshop fakery and yet they call the resulting image a "photograph" when it's not at all what was present in the scene when a camera was used.
I'm happy when someone presents "art" so that others can enjoy it and I'm even pleased with digital art that isn't photography. But I draw the line when I see someone present a "photograph" when the image never actually existed in reality and someone used Photoshop (or other editing tools) to fake the contents of the image and still try to pass it off as this amazing photograph that just happened to capture because they are that good with a camera and their impressive photographic skil!
In photo contests, John is correct in that generally they expect that what you are submitting is, in fact, actually a "photograph" and not digital art created by building a composite image. Even journalistic standards now establish that documentary photography may not use Photoshop to change the reality of the scene (the infamous war scenes where they added more smoke to the sky than was actually there.)
Some photo contests require that you must shoot RAW and that while you do get to submit a JPEG for judging, the original RAW file must be made available to the judges so that they can evaluate if you went overboard while editing the image (did you change things other than the basics.)
But put the debate aside... because the "debate" assumes that you could achieve the same (or very similar) result using either the physical filter vs. the Photoshop effect filter. What I'm trying to point out is real physical limits of technology. There are still barriers that digital technology has not overcome.
Imagine you have camera with a crude digital sensor which each "pixel" only has a bit-depth of 1. This means each pixel can either be a "0" representing "black" or a "1" representing white. You can't capture any tonality.
Now imagine you improve this sensor and give it a bit-depth of 2. Now you can represent four possible values... both bits off (00) is "black", both bits on (11) is "white", but "01" can represent dark gray and "10" can represent light gray. Now you start to get some tonality.
JPEG images are contrained to 8 bits because that's the standard. But RAW images are saved as 14 bit images (because that's the sensor technology).
But it goes a little more complicated than that....
The camera sensor has a concept called "well depth" that the manufacturer's don't publish. But through some testing, it's derived to likely be somewhere around 32000 for most Canon 18MP sensor cameras. You can think of the sensor as being 18 million little test-tubes and each test tube can hold 32000 drops of water before they are full. If you've got a pixel which is represented a reading of 20000 in one well, and you've got another pixel that needs to be twice as bright, then that needs to be represented as 40000 and there's a problem because the camera sensor only has a well depth of 32000... so as soon as you achieve 32001 the water drops start spilling over and can't register any value higher than the maximum well depth. If the camera had a bigger well-depth (say... 50,000) then it could handle more exposure range.
Also.. if for any reason you boost ISO, then you increase the reading for each pixel but that reading can't exceed the maximum that can be represented. That means the darks will all get lighter... but anything that would have already been close to the maximum value the sensor can represent will now get "clipped" (boosting ISO decreases dynamic range.)
In other words... there are technical limits to the capability of the sensor to represent everything accurately.
If you shoot an image and your image contains pixels that exceed those limits then the data is lost. If you have white puffy cloud with lots of details and tonality variations... as soon as you exceed the maximum limits of the sensor... the white puffy cloud that used to have detail now turns into a white outline with no detail inside it. You can go into Photoshop and try to reduce the exposure, but now you'll have a gray outline with no detail inside it (the detail is lost forever.)
So the point of the physical filtter is to bring the bright areas of the image (usually that's the sky) down a bit WITHOUT bringing down the foreground part of the exposure. This changes the shooting conditions of the shot such that an image that used to require 13 stops of dynamic range (which the camera can't handle) might now only require 10 stops of dynamic range (which the camera can handle without losing any data.) The gradient ND filters typically come in 1, 2, and 3 stop densities but you can "stack" two filters if you need more than 3 stops. If I meter my sky at 4 stops brighter than the foreground... then I might slide in a 3 stop filter AND a 1 stop filter (for a total of 4 stops).
Now... having captured that image WIHTOUT losing any data... if you want to perform FURTHER adjustments to that photo via software such as Lightroom or Photoshop you can still do that and make a good photo even better. But if you did NOT use the physical filter and the image exceeded the dynamic range of the sensor then it's too late... nothing you can do will recover the missing information.
So I mentioned that we have these barriers that technology hasn't overcome and that's not entirely true... while a Canon sensor might have somewhere around 32000 well-depth, there are sensors with significantly higher well-depth (some of the better astro-imaging sensors are at 100,000 well-depth that are available to consumers (albeit these CCD imaging sensors are wll above $10,000) and the scientific research-grade sensors are at or above 500,000 well-depth. The problem is they are so extremely expesnive that very few would be able to afford to buy the camera. In the world of consumer electronics it's not really a viable product to sell if you can't build it for a price that consumers can actually afford.
So maybe someday... those sensors will get to a point where consumers CAN afford such sensors and we wont need the physical filters. But that day is has not yet arrived.
Tim Campbell
5D III, 5D IV, 60Da
Filters time has come and went. The advantages in PS so far out weigh filters it isn't much of a decision any more. I'll concede polarizers and ND for now. But just for now.
