In my previous post I wrote that AI-based utilities have no place in my personal photo enhancement toolbox. Their results have been more miss than hit on my digitised snapshots. One set of utilities, however, has managed to generate results that almost border on the magical: Photo stitching software.
Many modern cameras and smartphones today feature the ability to create panoramic or 360-degree images but it is stitching software that is able to take photos, ideally from the same vantage point, of one or more subjects and almost seamlessly stitch them together into one larger picture.
So long as the focal point and lighting are similar, the results can be quite amazing.
The first such tool I regularly used was AutoStitch. It “works from unordered collections of images, automatically finding matches between images [and] then robustly aligns all images and uses advanced blending algorithms to form seamless panoramas.”
Eventually I moved over to Microsoft’s Image Composite Editor (primarily to stitch together record sleeve scans) which recently came in useful when I realised that I had actually been creating old-fashioned analogue panoramas in my photo albums all along!
Here is the result of ICE stitching together scans of the original negatives.
Similarly, below is a more complex mosaic made of prints. Despite its crudeness, this particular page also highlights the benefits of treating a photo album as a scrap book.
Luckily, I still had the original 640 × 480 pixel resolution image files (taken with my first digital camera), so I ran them through ICE as well. Due to the source photos being of lesser DPI, the final result is a little smaller than those from the photo scans (which is another topic entirely) — but I reckon this one came out alright.
All of this is nice and well when it comes to scenes that are static or include only one person who sits still while posing for the camera but as soon as you have moving subjects such as water or a crowd of people, the results can get rather weird and sometimes serendipitous — such as the guy with the red backpack in the image below:
And once again we deviate from the reality of a scene.
Photo stitching software ultimately becomes an invitation to various forms of creativity since it’s effectively delving into what the world of analogue photography knows as “multiple exposure”. There’s nothing wrong with creativity, but if the aim is to accurately represent a scene or location and there are technical or spatial situations to be overcome, then stitching images together helps capture the complete view that the human eye would see. This legitimises the practice to the point where one would have to accept it as being “close enough” (though I make sure to mark stitched images as such).
Google Street View and the like already offer a pseudo-virtual experience that is based on near-seamlessly stitched photos. None are perfectly truthful, and not only have we gotten used to it, but we’ve come to accept it as “good enough” for its purpose.
The old adage that “the camera never lies” thus no longer holds true.
In fact, it hasn’t for a while. With (generative) AI soon to be unleashed on smartphones, it might be that the very instant you click the shutter, the saved image will already be a glorified, beautified, and enhanced distortion of the reality in front of your very eyes.
Nothing is real.
All photos by Herby Hönigsperger.