Computational Photography

I mentioned in an earlier post that I was trying a Fujinon XF 14mm F2.8.  It is now my widest lens that I obtained for a few specific uses, but I would like to expand its use, so I have been experimenting with it.

If the lens isn’t used in a horizontal and level position it tends to distort the image as shown in the first view above.  This isn’t unique to this lens.  It happens with all lenses, but it is more obvious due to the wide angle when used up close as above.

Adobe LR CC has a computational capability to automatically correct for it.  With one click of the auto button it very quickly gave me the second version.  I used the above example to show what it achieves and how much gets cut out of the original along the edges.  The photographer who wishes to use this capability just needs to make sure that they go a little larger in their composition to allow for the loss around the edges.

What amazes me is how fast and well it readjusts the 24 million pixels to achieve this change.  After it is done I still have 24 million pixels, but each has been rearranged with no obvious image degradation.  Since it still has the same number of pixels, it obviously also creates pixels.  The photographer needs to be careful to not ask for too much of a change or else some objects might get noticeably distorted in a different manner from the original.

The following is a different example of computational powers.  The first image was originally 6000 x 4000 pixels in size that has been reduced to 2100 x 1400 pixels for display here.  I also made a virtual copy of the full-size image and then cropped out a portion of the center and then increased the size of the crop to achieve an image for display with the same number of pixels as the first.

This demonstrates the power of computations to allow zooming in on a subject.  I look forward to the day when I can use one camera and one prime lens for all of my photography and then rearrange and zoom in on any portion that I choose.

I fully expect that in the not too distant future most cameras will utilize a lot of computations within the camera to allow for vastly different lens designs that reduces the size of the lenses.  I also expect that some of this is currently being done within smart phone optics.  We will also have to stop referring to them as cameras or smart phones and call them for what they are … computers.