Subjects with movement are notoriously tricky for pixel shift but static subjects are more suited to the feature
Getting things right in-camera capture is often the best way to shoot because it provides the best possible raw file for editing, as opposed to crossing your fingers and hoping that you can fix problems in post. So, with this in mind, the idea of using pixel shift as an in-camera method of capture logically suggests that it will provide the best results.
To compare the results between native resolution, Adobe Super Resolution, pixel shift images processed using the camera manufacturer’s software, and pixel shift images processed using a convoluted manual Photoshop technique, I shot with a Sony Alpha A7R V. This camera captures 16 images in quick succession which have to be merged together in Sony’s Imaging Edge software to create a new high-resolution raw file. This takes the file dimensions from the native size of 9504x6336 pixels (121MB raw files) to a whopping 19008x12672, 1.82GB file.
We’ll discuss the results of all the approaches in a moment, but this file size is obviously excessive. In contrast when using Adobe Super Resolution, which also increases the image file size from this camera to 19008x12672, the resulting DNG file that’s created is a much more palatable 229MB. My computer is powerful, but it did struggle when combining 16 raw files in Imaging Edge and using a manual merging technique in Photoshop.
Moving subject results
Looking at the results of a seascape with a rough sea, the Imaging Edge software has done a fantastic job of making the rough sea look natural. To be honest, this is easy for the software to deal with because this area of the image lacks detail, but with an image with finer moving detail such as foliage, there would be a visual stutter/ghosting present. In terms of sharpness, this is the least sharp version of the image with a haziness present in brighter detail areas.
The images taken were shot at blue hour, and blue is the noisiest RBG channel, so grain is present in all images except for the version merged manually in Photoshop. This is because noise is irregular so this method smoothed out during the merge. Movement is simply blended, so it’s not as effective as Imaging Edge, but it works with the water here because it looks like a longer exposure than it was; but for movement of finer details, it would inevitably produce an inferior blend. The overall result is slightly sharper than that from Imaging Edge.
The version that’s arguably the best is the one created by applying Super Resolution to a standard raw file, because this provides the sharpest result. More artefacts are visible in some areas though due to the noisiness of the blue channel. But the fac
