It has been a while since I posted the “How was this picture made #11?” article, where I showcased a very high resolution image of sand particles with tons of detail. The image was massive in size and resolution when I extracted it out of Lightroom. In fact, the image was so big, that I had to downsize it to 4096 pixel long resolution in order to keep the size at less than 10 MB with as much JPEG optimization as I could. When dealing with so much detail, even the highest JPEG compression levels will still yield large files, since there is so much pixel-level data. And that’s what you get when you have an image produced from a sensor that moves one pixel at a time in order to create a super high resolution image! And combined with the power of focus stacking multiple images, you get insane levels of detail from a macro shot like this. So how did I do it? Let’s talk about the specifics of this particular shot.
I have already kind of revealed the answer in the opening paragraph – the image was captured by the Pentax K-1 DSLR, the camera that has the amazing pixel-shift technology capable of creating extreme details, something typical sensors with a bayer filter are unable to reproduce. Stephen Bennion, great job at guessing the setup! :) The lens that I used was the Pentax 24-70mm f/2.8, which is basically a re-branded Tamron 24-70mm f/2.8 with its VC disabled. Although I had the Pentax 70-200mm f/2.8 with me as well, it would have been a bit too long to use for this particular shot, since I was pointing my camera directly down from my tripod. I zoomed to around the 70mm mark to get the details and the tight framing I wanted and at that distance and focal length, it would have been impossible to capture all the details in the frame from the center all the way to the extreme corners, even if I stopped down significantly. I personally try to stay away from small apertures past f/11 for these kinds of shots, because diffraction can seriously hurt resolution, so my choice was to take multiple shots focused at different areas of the frame, then stitch them together for a single shot in Photoshop.
Before I talk about how I did that, let me first talk about the advantage of the pixel shift technology, particularly for such shots. What kind of additional detail can cameras such as the Pentax K-1 add? Just take a look at the following two images (studio scene to soon be published in the upcoming Pentax K-1 review):
I know what you are thinking – one looks like blurred mess, almost like an out of focus image, while the other looks quite detailed in comparison. But I can assure you, both images are 100% in focus. The “before” image was shot with the Nikon D810, whereas the “after” image was shot with the Pentax K-1. The difference is literally in the bayer pattern – even without an anti-aliasing / low-pass filter in place, the Nikon D810, which is supposed to be a camera that can produce tons of detail, is not capable of providing full pixel-level details of the scene. The Pentax K-1, on the other hand, has the ability to shift its sensor one pixel at a time (thanks to its IBIS), in order to resolve every bit of detail available. Now that’s amazing!
Of course this pixel shift technology does not work for every type of scene, especially if there is a lot of subject movement. However, for static shots like this and when there is no wind, pixel shift technology is absolutely amazing. In fact, the detail level one can get with such a mode is pretty darn close to what a medium format camera can do, once images are down-sampled.
Now imagine the potential of using pixel shift, along with focus stacking for a single composite. And that’s exactly what I did here. I started out by focusing on the bottom edge of the frame, then moved up bit by bit, by capturing a total of 6 images. Take a look at the following sequence of the last three images:
As you can see, there is a slight change of framing between the three shots. And that’s the unfortunate issue you are going to get with most lenses – when you focus at different areas of the frame, there is a potential for a change in focal length, which can mess up the framing! This means that if you were to take such images to Photoshop and try to automatically stitch them, they might not stitch properly due to alignment issues.
After right-clicking the six images in Lightroom and selecting “Open as Layers in Photoshop” and letting Photoshop do its initial work, I selected all layers, then went to Edit->Auto-Align Layers:
What this does, is align layers and cut off parts that are not present in all layers. After the process was complete, I cropped the image a little bit to only be left with the full image. Next, I visited Edit->Auto-Blend Layers and picked “Stack Images”. I deselected both “Seamless Tones and Colors” and “Content Aware Fill Transparent Areas”, since I did not need them.
Once blending was done, I ended up with layer masks on top of each layer. Although Photoshop did a wonderful job, I still selected a soft brush and brushed off the abrupt edges manually from some areas. The image was ready and it only took me about 3-5 minutes to get this all done.
The pixel shift technology offered by some modern cameras is amazing. If manufacturers continue to work in advancing this feature a bit further by taking quicker shots and processing them, finding ways to blend images better and take care of other issues and limitations, this will be a very desirable feature to have on every camera.
Pentax has done an amazing job with the K-1. If Pentax had a comparable lens line-up to Nikon, I would seriously consider switching. The Pentax K-1, being a sub-$2K full-frame camera, offers so much amazing technology, that it is now becoming a reference camera when it comes to rich set of features, image quality, dynamic range and overall performance. I hope Nikon and Canon wake up from their deep sleep and shake the industry once again. Abandoning the Nikon 1 mirrorless system or releasing cameras such as the EOS M5 that can barely compete with other mirrorless offers won’t get the two far…