In response to some recent threads, I have been playing around with three different dodging and burning techniques in Photoshop that don't use the dodge and burn tools. Here's what I have come up with. I would be interested to see whether those here who have more knowledge of photoshop have some corrections or additions to this.
All of this assumes that you dodge and burn in separate layers.
1. Technique 1: create a neutral gray layer, paint with a black or white brush, and use a blend mode like overlay or soft light. I think this is the most common method, and it is the one I used until recently. I think it has nothing to recommend it over the other two, other than its simplicity. It adds or subtracts gray, and as Manfred noted, this affects saturation. AFAIK, there is no way to make a saturation adjustment to compensate just in the painted areas, or to make other adjustments only to the painted areas.
2. Painting on a transparent layer with a black or white brush, using the same type of blend mode. I think John suggested this, but a very nice explanation can be found here. This too adds or subtracts gray, I believe. The big advantage over #1 is that you can adjust the saturation or any other attribute of the painted area. You can do that in either of two ways: creating a selection from the painted area, as shown in the video, or by making the subsequent adjustments a clipping mask.
3. Manfred's technique: create a curves layer with a black mask and paint areas that you want to burn or dodge, with a normal blend mode. With this as well, you can adjust the painted area. One can use a clipping mask, as Manfred showed, or copy the mask from the burning layer to the subsequent adjustment layers.
Manfred showed this to illustrate a way to compensate for unwanted saturation changes. I think one could avoid those saturation changes more simply by changing the blend mode to luminosity. I think this works with #2 as well, although I have just started playing with that.
If I'm right so far, what are the pros and cons of #2 and #3?
I think that in many cases, the choice won't make much difference. However--here I am getting further ahead of myself--I think there are circumstances in which #3 is superior. First, it doesn't require a uniform darkening or lightening. Manfred showed a mid-tone adjustment, which affects most of the histogram, although to varying degrees. However, one could use any curve one wanted and adjust some parts of the tonal range more than others--vaguely akin to using luminosity masks in a local adjustment. Second, if I am right, #3 does not add or subtract gray. It changes the distribution of luminance in the RGB channels. It seems to me that in cases where one wants large changes in luminance, this may be superior.
Corrections? Other reactions?