I sometimes wonder about how much postprocessing is, for lack of a better word, "honest", in creating a photograph. When working with an image composed of a bit vector, which is only interpretable through the use of complex hardware, I feel free to modify the image as much as desired; unlike working with a traditional negative, in which the image has a concrete physical form, one sequence of bits is, in some sense, as good as any other. This lack of permanency, of a link to the exposure itself, is in some ways liberating, but can also feel dishearteningly trivial.
For most images, I perform mild color (usually, just levels) correction, rotation, and cropping only. I feel that these modifications are not only traditionally acceptable, having analogous processes in the darkroom, but do not change the photograph in a way that misrepresents having been there. That is, I suppose, the most important aspect of photography for me: relaying the experience of seeing something in the world to someone else. Drift too far from that experience, and the photograph communicates a dream, not reality.
The funny thing is, I'm not troubled by paintings, sketches, or detailed computer graphics for not conveying reality. Likewise, I enjoy fantastic effects at the movies, where people interact with fabricated images so realistically that it is difficult to identify what in a scene actually existed in the first place. The idea that photographs, by virtue of their apparent realism, must somehow be constrained to that reality, is, when I think carefully about it, quite arbitrary.
Anyway, there are times when the image I want to create is factually accurate, but impossible to achieve without the use of significant modification. For example, this image of Vernal Falls is a composite of two matched exposures, one above, and one below. I was on a narrow trail inset into a cliff face, and didn't have a lens wide enough to capture the scene in any complete way. I took two carefully aligned exposures, overlaid the images in an editor, and erased the upper exposure along fault lines in the rock face and forest to combine the two. The resulting image is almost indistinguishable from what one would see if they were to stand on that trail, but it is not quite the same image: instead of having only minor color variations, the structure is different. Lens distortion makes such errors impossible to avoid.
What I wonder is, given how "true" the photograph feels visually, does it still communicate reality? How much change in an image is required before it no longer documents what is, and begins to describe what one imagines? And how much modification of that original exposure can I be comfortable with as a photographer?