Non-Destructive Geometric Corrections: Yes, Please

One of the major advancements in the current generation of photographic image editing software has been the advent of non-destructive editing, whereby user modifications to an image do not actually alter the image source data (pixels) in the master file, but are recorded instead as separate actions that are applied to the data when viewing and/or outputting the image. There are many advantages to preserving the integrity of the source data for future use, while still achieving the necessary modifications for an immediate application. These non-destructive edits have been available to alter a wide range of image parameters such as exposure, color balance, contrast, sharpening, noise, and many other adjustments. But there are other parameters, such as geometric distortion manipulation, that are not yet offered in a non-destructive form. These corrections generally require the creation of a separate new version of a file in order to preserve the modified data. For a variety of reasons this is not an ideal solution, with issues related to storage capacity, application interoperability, and versioning. None the less, these types of image adjustments are very important for subjects such as architectural photography where control of visual perspective is highly desirable. In these cases photographers have generally learned to live with these multi-file issues.

On another photography blog that I regularly read the moderator speculated that geometric distortion manipulation of photographic images would always be a destructive editing process, presumably due to the complexity involved to maintain modified data in a state accessible for further editing. However, I'm not so quick to discount the possibility of these as non-destructive corrections in the future. Any image correction is just 1.) a user-selectable value and 2.) an application-generated view of that transformation. There is no reason why that view MUST actually transform the original data, as long as the software can always regenerate the edited view on-demand from the user settings. Even distortion adjustments are correlated to user settings, and these can be recorded, stored independently, then recalled and applied to an image, no different from any other non-destructive value. It probably takes a lot of processing power to manage those kinds of "live" transformations in real time, and that may explain their current lack of availability, but the ever increasing level of computing horsepower is assured and ultimately this should not be a limiting factor. I fully EXPECT for software developers to offer non-destructive distortion corrections in the future. Perhaps not in their next updates, but certainly sooner, rather than later.

For architectural photographers the advantage of being able to perform all our mostly commonly used image corrections in one application, and without needing to create alternate versions of files, will be a significant improvement to our workflow and image management strategies. I have seen examples (on the Web) of prototype image editing programming like this from one of the big graphics laboratories, though I can't remember where now. I do remember that it was impressive and, of course, I wanted it immediately. I have no doubt we'll see it -- and be paying to get it -- before too long.