Pentax K-1 Pixel Shift Resolution Mode

An analysis of the Pentax K-1's "Pixel Shift Resolution" mode

by Mike Tomkins | Posted: 05/06/2016

When it launched its crop-sensor flagship camera, the Pentax K-3 II, back in April, 2015, Ricoh debuted a brand-new feature that was hugely useful to anyone wanting the maximum per-pixel sharpness and dealing primarily with relatively static subjects. Dubbed Pixel Shift Resolution (or in some markets, Real Resolution), this function captured multiple images and combined them in-camera to create a single shot with higher resolution.

Pixel Shift Resolution worked similarly to the High-Res Shot mode of the Olympus E-M5 II, which predated the K-3 II by a couple of months, although there were some important differences between the two. (We'll come back to those in a moment.) Olympus has since taken its rival feature into an even smaller camera with the handsome Olympus PEN-F. Now Ricoh heads in the opposite direction with the Pentax K-1, the subject of this review and the first full-frame camera to include such a feature.

The K-1's Pixel Shift Resolution can now compensate for moving subjects

In our earlier K-3 II review, we noted that both Ricoh and Olympus' competing technologies shared the same key limitation: They only worked with static subjects. (At least, unless you were willing to roll your sleeves up and get down and dirty in the digital darkroom.) Interestingly, that is no longer the case for the Pixel Shift Resolution function in the Pentax K-1. Although you do still need to use a tripod, just as you do with Olympus' cameras, Ricoh says that its uprated function in the K-1 can optionally compensate for moving subjects.

The way in which it does this is essentially automating the same technique I proposed in my first Pentax K-3 II field test, but with a whole lot less fuss on the user's part. The camera locates areas of subject motion, and then simply doesn't apply the Pixel Shift Resolution technique here. That means that these areas don't gain the benefit of improved resolution, but as it happens that's not really an issue. Your brain expects subjects in motion to look softer, and so if it's noticeable at all, simply interprets that softness as a moving subject. The remainder of the image gains the benefit of greater per-pixel sharpness, and all is well, in theory at least.

I have, of course, tested this theory out. To see how it performed, you'll want to read my real-world Pentax K-1 Field Test. (Spoiler alert: It works better for larger moving subjects than it does for more subtle motion like foliage and ripples in water.)

The K1 still can't compensate for handheld shooting, however

Of course, the reason for the tripod requirement is pretty simple: If it's the camera that's moving rather than the subject, then most or all of the image frame (depending upon how the camera and subject moves) will show motion relative to your subject. That means that most or all of the scene will be rendered using the single-image capture, and effectively the Pixel Shift Resolution function will do nothing at all.

Thus far at least, it seems that the challenge of correcting for handheld motion -- compensating accurately enough to freeze that motion while simultaneously managing the single-pixel shifts required between frames -- is too tough a challenge for the Pixel Shift Resolution function to work without a tripod, but that's not really surprising.

Even with the requirement to shoot on a tripod, with its new motion compensation function Pixel Shift Resolution should be noticeably more useful than in the K-3 II, though, and that's great news. As in the that camera, for predominantly static subjects it can give you access to even greater resolution than can be derived from a single shot, reduce the incidence of moiré and false color, and -- although still we can't think of a good reason you'd need to do so -- can also improve noise levels and grain size.

How The Pentax K-1's Pixel Shift Resolution differs from Olympus' High-Res Shot

Both Ricoh and Olympus' cameras create their high-resolution images by capturing multiple shots in sequence, and adjusting the sensor position just fractionally between shots using the sensor-shift stabilization mechanism so that the light at any given location falls on different sensor pixels. Where typically a camera using a Bayer-filtered sensor must interpolate (or in essence, guess) the values for two out of the three red, green and blue channels for every pixel, this technique allows full color information to be recorded directly at every pixel.

So what separates Ricoh's approach in the Pentax K-1 and K-3 II from that selected by Olympus in the E-M5 II and PEN-F? The answer is fairly simple: Olympus is moving the sensor in steps that are smaller than its pixel size, and then capturing a total of eight frames to create a single output image. Ricoh, by contrast, is using full pixel-sized steps and combining half as many frames per image, with a total of four images captured per output frame.

Olympus' approach to the problem gives it more data to work with, and allows resolution to be improved well beyond what the sensor resolution would suggest. Final output resolution for the E-M5 II is 40 megapixels in JPEG mode or 64 megapixels in raw, while that for the PEN-F is even higher at 50 megapixels in JPEG mode or 80 megapixels in raw. It also presents a much more complex problem for the cameras, though, and requires finer sensor positioning.

With less data to juggle for every pixel, Ricoh's approach as used in both the K-1 and K-3 II is much more easily solved, and yet it still triples the amount of information for each pixel compared to a single-shot capture shot with the same sensor. And as an added advantage, it does so without significantly increasing file sizes, at least so long as you're shooting in JPEG mode.

In raw capture, though, the Pentax K-1's Pixel Shift Resolution images have about 3.7 to 3.8 times the file size of a standard raw file. You can expect some extremely large file sizes if you intend to shoot in raw format using Pixel Shift Resolution with the Pentax K-1, on the order of 185-190MB even at base sensitivity, and likely even larger as your ISO sensitivity setting climbs.

The largest DNG file we've shot thus far, captured at ISO 204,800-equivalent, weighs in at a truly staggering 264MB for a single scene. The hefty file sizes are also true of Olympus' cameras, if not quite to the same extent. The E-M5 II's High Resolution-mode raw files weigh in at around 104MB, about five times the size of its standard raws. The PEN-F's, meanwhile, tip the scales at around 131MB, some seven times the size of its standard raws. Interestingly, file sizes don't vary even slightly for either of Olympus' cameras, no matter the subject or sensitivity used.

Also, it's worth noting here that even with the 20-megapixel PEN-F, Olympus has quite a resolution deficit to make up compared to the full-frame competition. Since its sensor has around 2.3x more pixels than that in the E-M5 II, and 1.8x as many as in the PEN-F, there's less need for the Pentax K-1 to increase resolution.

Comparing the Pentax K-1 to its full-frame rivals, only the Sony A7R II, Canon EOS 5DS and 5DS R offer higher resolution in single-shot mode, and by much more modest margins. The Sony has about 16% higher resolution on paper, and the Canon models around 38% higher.

(Note that the K-1 crops and linked images below were taken with the Natural Image Tone setting, instead of the default Bright setting.)

The Pentax K1's Standard / Pixel Shift Resolution modes compared

The Pentax K-1's Pixel Shift Resolution versus the competition

Now we've seen the gain over single-shot imagery, what about a comparison to the competition? We were only too keen to find out how Pixel Shift Resolution imagery from the Pentax K-1 compares to that of its rivals, so we pulled together a raft of our standardized test samples to find out.

For this comparison, we decided that a number of comparisons were in order. First and most obviously, we wanted to see how the K-1 fared against the highest-resolution full-frame offerings from its rivals -- the Canon EOS 5DS R, Nikon D810 and Sony A7R II. Then, of course, we wanted a comparison against Olympus' High-Res Shot mode. Here, we opted for the higher-res PEN-F, for a fairer comparison against the much higher-res K-1 than you'd get from the E-M5 II. And finally, just as in our roundup for the Pentax K-3 II, we wanted to see where Pixel Shift Resolution placed the K-1 versus its own siblings, and so decided to include the K-3 II's High Res Shot mode, as well as medium-format imagery from the Pentax 645Z.

For all cameras except the Olympus PEN-F, which has a base sensitivity of ISO 200 equivalent, we opted to stick with ISO 100 for all models, that being the base sensitivity of the Pentax K-1. (And indeed, of most models in the roundup). All cameras were also set to their default noise reduction settings, that being Auto for the Pentax cameras. To ensure we got the most from the Canon 5DS R, we opted for its Fine Detail picture style. And since the Olympus PEN-F offers different resolution depending on whether imagery is created in-camera or not, we decided to include not just the in-camera JPEG results for High-Res Shot mode, but also images processed on the desktop at higher resolution with default settings in Olympus Viewer 3.

Of course, lenses matter too, and as a reader pointed out we missed mentioning which were used. (You can always check the EXIF info by clicking the link from the sample page if you want to check settings, but I did mean to mention the optics here myself.) So which lenses were used? The Pentax K-1 and K-3 II used the Sigma 70mm f/2.8 EX Macro, and so did the Canon EOS 5DS R and Nikon D810. The Olympus PEN-F used the Zuiko Digital ED 50mm F2.0 Macro, and the Sony A7R II used the FE 55mm f/1.8 ZA. Finally, the Pentax 645Z used the D FA 645 55mm F2.8 AL [IF] SDM AW.

Given the number of comparisons to make here -- and therefore the scope of the task at hand -- we decided to make a few concessions relative to our last test. For one thing, we're going to let the images do the talking here: We're simply providing the crops and links to full-resolution imagery, and allowing readers to draw their own conclusions as to which cameras are doing the best job. We're also providing one crop from the mosaic label area of our standardized test scene, because it's near the center of the image frame (minimizing potential lens-induced issues), includes both a mixture of extremely fine detail and smooth, flat areas with minimal detail, and can crop down nicely for easier side-by-side viewing.

Also, while last time we provided comparisons both upsampling the lower-res camera and downsampling the higher-res one, that honestly seemed like overkill to us for a comparison this big. Doubly so because the resampling and sharpening methods used will affect the results, making it to some extent a comparison of the resampling methodology, and not of the cameras themselves. If you want to make the comparisons for yourself, full-res out-of-camera JPEGs and raw files are available on our samples pages for all of these cameras, so you can again draw your own conclusions.

With that out of the way, let's take a look at the results!

The Pentax K1's Pixel Shift Resolution's as relates to sensitivity

Another advantage of this multi-shot technique, as we said at the outset, is that it can significantly reduce noise levels. For our final comparison, we offer a side-by-side look at how noise levels and detail differ between single-shot and Pixel Shift Resolution modes across the Pentax K-1's ISO sensitivity range!