Pentax K-3 II Field Test Part I

Is Pixel Shift Resolution the best argument yet for shooting raw?

By | Posted:

31mm-equivalent, 1/250 sec. @ f/7.1, ISO 200

Regular readers will probably know that I've been a big fan of Pentax's enthusiast-grade DSLRs ever since the launch of the Pentax K-7 way back in 2009. As soon as I finished my review of that camera, I bought one for myself. The same thing happened with the followup Pentax K-5, and on the rare occasions when I don't have a camera ready for a review, it's now my daily shooter.

Ever since I completed my Pentax K-3 review in late 2013, I've been planning to buy that camera as well. I hadn't yet gotten around to it, though, mostly because I promised myself that I'd sell at least one of the earlier cameras before I bought another. Now, I'm considering making the leap to the Pentax K-3 II instead, as it possibly fits my desires even better than did its sibling.

The K-3 II marks the second time the series has seen a II release, and indicates a less-significant upgrade. The last time it happened was the K-5 II, a camera I praised but didn't personally buy as it was so similar to my K-5. Only relatively few changes in the K-3 II, but they're all ones which have appeal for me -- improved autofocus tracking and image stabilization, built-in GPS with AstroTracer function, and most of all, the brand-new Pixel Shift Resolution mode.

(Curious how Ricoh's flagship Pentax APS-C DSLR lineup has evolved over time, or want to know what you'd gain from upgrading from your earlier model? Here are some links to compare the Pentax K-3 II vs. Pentax K-3, the Pentax K-3 II vs. Pentax K-5 IIs or K-5 II, the Pentax K-3 II vs. Pentax K-5, and the Pentax K-3 II vs. Pentax K-7. And I'd recommend reading my earlier Pentax K-3 review and shooter's reports, as well, because as noted above, the new model is in most respects identical to that camera other than the points mentioned above and the removal of the flash strobe.)

60mm-equivalent, 1/320 sec. @ f/5.6, ISO 200

In-camera geotagging saves me a lot of time

Perhaps the most significant selling point of the Pentax K-3 II for me, as compared to its predecessor, is the newly built-in GPS. I like having my shots geotagged so I can browse a map to find the images from a given location, and it's a major pain in the neck to add tags manually for every shot after the fact.

31mm-equivalent, 1/250 sec. @ f/7.1, ISO 200

Sure, I could pick up the Pentax GPS Unit O-GPS1 to use with the K-3 and get the exact same functionality, but it's pretty pricey, bulky, and can't be used at the same time as either the internal flash or an external strobe -- unless the latter is separated from the camera and connected to the PC sync terminal, anyway.

Sadly, there's no longer a built-in flash...

It's a bit of a shame to lose the built-in strobe of the earlier camera, but I must admit I use it pretty seldom anyway, since direct flash from an on-camera strobe is rather unflattering. And I can certainly understand why Ricoh didn't keep the flash.

Modifying the casting for the magnesium-alloy body to make space for the GPS antenna and related circuitry would've been much more expensive than simply making a mold for the new, slightly taller plastic cover that replaces the popup flash. (And since the antenna likely has to be outside the magnesium-alloy body so as not to block GPS signals, there really wasn't anywhere else it could be put. Unlike its rivals from Canon and Nikon, the Pentax K-3 II has a body that's almost entirely crafted from magnesium-alloy, so putting it behind a plastic panel somewhere else simply wasn't an option.)

...but ditching that flash actually means a lower pricetag

By including the GPS in-camera, the Pentax K-3 II's geotagging capabilities feel much more tightly integrated, and I save about US$250 list on the price of the GPS receiver. That's more than enough to cover the cost of an external strobe like the AF201FG, which is not only more powerful than the built-in strobe was, but also more versatile. (That particular strobe, priced at US$150 list, has an all-weather design and tilting head for bounce flash.)

60mm-equivalent, 1/320 sec. @ f/6.3, ISO 200

Still, I'd really like to see the flash return alongside in-body GPS and Wi-Fi antennas in a future generation of the flagship K-series cameras, because the belt-and-suspenders part of me knows that I'll inevitably leave my flash at home at some point, and find myself regretting that.

GPS setup is quick and simple

Since geotagging was so high on my wishlist, literally the first thing I did on taking the camera out of the box -- after restoring it to factory defaults, dialing in a few preferred settings and inserting a flash card -- was to enable GPS. It's very easily done with a simple press of a button on the side of the camera. (It's actually the same one that used to serve as the mechanical release button for the popup flash on the K-3, but now it's an electronic button instead.)

I got a lock from the GPS -- indicated by a yellow or green satellite icon adjacent to two signal strength bars on the LCD monitor -- quite quickly, even the very first time I took the camera out of the box. With the camera having been sitting unused for at least a couple of days before it got to me and then completely reset, a lock took perhaps ten or fifteen seconds. It's probable that had it been left for a few weeks or months without use, the first lock would take longer, though. Subsequent locks were even faster, on the order of just one or two seconds.

And looking back over the images from my first shoot, the precision was very, very good. The recorded location was within perhaps 10 feet or less of where I know that I was standing almost all of the time, and the absolute worst outdoor result that I could find was a single shot which was maybe 25 feet from my actual location. Even indoors in my basement, the GPS receiver readily got a position fix that was within 30-40 feet of my position. This will doubtless vary some, though, especially when there's more sunspot activity or if you're shooting with a minimal view of the sky.

60mm-equivalent, 1/250 sec. @ f/5, ISO 100

It's great to have a compass, too!

I really appreciate the fact that the Pentax K-3 II includes a compass as well. A GPS alone tells you where the camera was located, but it doesn't tell you what your subject was -- you could have been shooting a distant subject in any direction, after all. With a GPS and compass together, though, plus the focal length of your lens, you have a pretty good record of precisely what you were photographing.

60mm-equivalent, 1/200 sec. @ f/4, ISO 100

The compass in the K-3 II seems to me to be both very responsive, and impressively accurate -- much more so than the compass on any smartphone I've ever tried. In my experience, smartphone compasses typically either drift a couple of degrees or more without moving them, or lag significantly behind my movements because they're averaging multiple readings to smooth out the hiccups. The K-3 II's compass seems to have much greater precision, because it doesn't drift noticeably but reacts to camera motion near-instantly.

Compass calibration is also simple

You do need to calibrate the compass now and then, though, something which only takes a few seconds to do. All you need to is to turn the camera beyond 180 degrees on all axes. Once you're done, the camera will tell you that calibration was successful, or alert you if it failed. It's important to move away from any large metal objects that could affect the calibration.

(You can see Ricoh's suggested method for calibration in this YouTube video. There aren't any English-language subtitles, but it's pretty self-explanatory. It's also not the K-3 II that's shown, but an earlier model with the optional O-GPS1 accessory mounted. The process is identical, though.)

There doesn't seem to be much effect on battery life

Given that geolocation is one of the primary selling points of the Pentax K-3 II, I've left the GPS enabled at all times when I've been shooting with the camera. Impressively, battery life seems to have been barely affected by the addition of the GPS receiver. At this point I've already captured hundreds of images over a period of days, and spent a good while fiddling with camera settings and reviewing my shots on the camera's LCD.

31mm-equivalent, 1/500 sec. @ f/8, ISO 200

I brought along a spare battery everywhere I went, but I really didn't need to: It's only within the last hour or so of typing and referring to the camera as I went that the battery gauge has even shifted from the full position, and even now it's only one tick below full! That's not really too surprising, though -- I've always found the K-series cameras to have great battery life.

A few thoughts on AstroTracer

I don't have a lot to say on the AstroTracer function in this first field test because I want to test it for a while longer before I render judgement. With that said, I do have some early thoughts on the function. The interface is a little unintuitive, and I had to go back to the user manual to figure out how to get it working. Initially, I thought I'd enabled it correctly, but my shots all had star trails in them, which I couldn't understand.

It turned out, though, that the camera has to be in Bulb mode before AstroTracer can be functioned. I'd incorrectly assumed I could just enable mirror lock-up, switch to manual mode and dial in my chosen exposure, but that's not the case. And even once you've enabled AstroTracer in the GPS menu and made sure you're in Bulb mode, the function still won't be used until you've then pressed the Green button to activate it. It's a little confusing, initially.

And there's one more quirk which is even more curious. Once AstroTracer is enabled, the Calibration function in the GPS menu is replaced by a Precise Calibration function -- but it uses the exact same motions as the regular function does. I'm not quite sure why there's a distinction between regular and precise calibration, as it seemed to be no more difficult to complete the precise calibration than it was for the standard one.

31mm-equivalent, 1/100 sec. @ f/4, ISO 100

My early results with AstroTracer have been hit and miss

I'm still waiting for a cloud-free night with which to give AstroTracer a proper test, but once I figured out how to enable it, I did shoot some test shots through gaps in the clouds. My early results were a bit hit-and-miss. Sometimes, the star trails were frozen impressively well. Other times, I still saw star trails in my images, even though I could hear the image stabilization system moving to try and freeze the star positions.

I should note here that I was well below the maximum available exposure time, as well -- not that the camera will let you exceed the limit when AstroTracer is active. There's a hard limit of five minutes, but the actual time can be much less depending on the focal length of the lens you're using, as well as your physical location and the direction in which you're pointing the camera. (All of these variables affect the speed at which stars move across the frame, and so dictate the length of time it will take for the K-3 II's sensor shift mechanism to move all the way across its range of motion, at which point the exposure has to stop.)

Pixel Shift Resolution, 172mm-equivalent, 1/200 sec. @ f/7.1, ISO 800
See the single-shot version of the same image here

I'll revisit AstroTracer and try GPS logging in my second field test

I will, of course, be trying to get to the bottom of all of this -- and to figure out how to get consistent results with AstroTracer -- the first clear night it gets. So far, having to chase gaps between clouds has made it impossible to figure out. (You don't really realize just how quickly clouds move until you try to shoot a several-minute exposure between them!)

60mm-equivalent, 1/250 sec. @ f/5.6, ISO 200

I've also not yet tested the K-3 II's GPS logging function, but I'll give this a try in my second field test, too. Given the excellent accuracy I've seen from the GPS receiver thus far, though, I'd expect GPS logging to perform pretty well!

A couple of very minor GPS complaints

Really, I only have a couple of very minor complaints about the Pentax K-3 II's GPS functionality. The first and most important is that the new GPS button seems a bit wasteful -- and potentially problematic -- if, like me, you intend to shoot with the GPS active almost all of the time. I can't imagine too many people will be wanting to disable or reenable it regularly on a per-shot basis. More likely, they'll want to adjust it once per trip, or just leave it enabled all the time. (If they wanted it disabled all this time, they'd just buy the Pentax K-3 instead.)

And as-is, while it hasn't happened to me yet because the camera hasn't gone in a bag and I haven't used the shoulder strap, that button seems far too easy to bump and disable the GPS by mistake, leaving me with a batch of images to tag by hand until I realise that the button got bumped.

I'd really like to see Pentax give the ability to override the default function of this button, and instead to let it serve as a second customizable button. (And if its function was changed, then enabling or disabling of the GPS receiver would become a function of the GPS menu, instead.)

31mm-equivalent, 1/160 sec. @ f/6.3, ISO 100

The other point is much more minor: In Playback mode when reviewing captured images, or on the Electronic Compass status screen, altitude units are fixed to metric. The latitude and longitude, meanwhile, are written as degrees and decimal minutes. Neither can be changed, which isn't great if, like me, you prefer the imperial system and locations as degrees, minutes and seconds.

On the road to historic Jonesborough

For my first field test of the Pentax K-3 II, I decided to visit the historic town of Jonesborough, Tennessee's oldest city. My six-year old son Geoffrey came along for the day trip, serving as my enthusiastic tripod caddy. With Pixel Shift Resolution being another of the big changes in the K-3 II, I was clearly going to need one!

As well as the tripod, I grabbed a bag and brought along a selection of my own lenses. I opted only for one zoom, the smc Pentax-DA 18-135mm F3.5-5.6 ED AL [IF] DC WR, a versatile and far-ranging weather-sealed lens that many advanced amateurs and enthusiasts would likely choose to start their kit. I also brought both the smc Pentax DA 21mm F3.2 AL Limited and smc Pentax DA 40mm F2.8 Limited, two superb pancake primes that are both well-known as very sharp lenses.

Great performance and excellent daytime image quality

Starting off, I set the camera at its defaults other than enbling the GPS, switching to raw+JPEG file formats, and bracketing my shots by +/-2/3 EV. I spent a while wandering around town, shooting hand-held to get a feel for how the camera would behave. Not surprisingly, the answer to that question was "basically identical to the K-3".

31mm-equivalent, 1/80 sec. @ f/4, ISO 100

I wasn't shooting sports or low-light here, so the tweaks to the K-3 II's autofocus and stabilization systems didn't enter the equation yet. Exposure was very accurate, and images were filled with boatloads of detail accompanied by good if quite snappy, vivid color. Performance was excellent, by enthusiast-grade standards. The Pentax K-3 II is really enjoyable to shoot with, and ergonomically speaking it's right at the top of its game.

Pixel Shift Resolution hugely improves per-pixel sharpness...

After getting some shots wth the K-3 II under my belt, I switched the program line to MTF mode, letting the camera prioritize the aperture it felt would yield the sharpest result. I shot handheld for a little bit more, and then set up my tripod to give Pixel Shift Resolution a try. I have to say that the results were immediately impressive, even when reviewed at 1:1 on the camera's LCD monitor.

(Just like the K-3 before it, the K-3 II cleverly indicates when you're viewing at 1:1 resolution, which is at the 8.3x zoom level. When zoomed in to this extent, you can really get a good feel for sharpness even on the camera's own monitor.)

With Pixel Shift Resolution enabled, I saw a very definite increase in fine detail in the in-focus areas of my images, and that wasn't just true with my really sharp primes. Even with the 18-135mm kit lens, there was a definite improvement on offer. And unlike cameras of days gone by that could perform pixel shifting, there were no noticeable artifacts -- at least, so long as my subject was completely static.

Pixel Shift Resolution, 90mm-equivalent, 1/125 sec. @ f/7.1, ISO 400
See the single-shot version of this image here

...but your subjects must be absolutely static

But shooting in the real world is certainly a much harder test than it was back in the lab. Subtle motion is everywhere in the great outdoors, even in places where you might not expect it. It's not just people, animals, vehicles and foliage that you have to worry about. More than just the immediately obvious can cause artifacts.

For example, when shooting Pixel Shift Resolution photos of an attractive downtown church, I noticed that the windows had a stippled, pixelated look -- and then it suddenly occurred to me: The reflections in the windows had changed between shots! Another surprise was that parked cars showed some artifacts, because reflections of clouds in the sky had moved.

Pixel Shift Resolution, 27mm-equivalent, 1/100 sec. @ f/5.6, ISO 200
See the single-shot version of this image here

And yet some things I expected to throw the algorithms completely haywire, such as traffic on the roads or flags waving in the breeze were relatively artifact-free. I can only assume that there's at least some intelligence to the algorithm, and that if there are significant areas of change between the four frames captured by the camera for each shot, only one of the frames is used for some areas.

A tripod is an absolute must for Pixel Shift photos

But even on this breezy day, I saw quite a significant improvement in per-pixel sharpness for many of my photos. It made me yearn for the technology to be hand-holdably, or able to provide the same detail with moving subjects, but that's definitely not the case.

Even if only useful for static scenes, though, there's no question that Pixel Shift Resolution is a very worthwhile addition to the Pentax K-3 II's toolkit. Unless Ricoh decides to gift the K-3 with a firmware update adding the technology -- and I don't know if it's even technically possible, because my understanding is that the K-3 II's Shake Reduction mechanism is physically improved -- then this alone would be enough for me to consider buying the K-3 II over its sibling.

That little extra bit of sharpness isn't going to make a huge difference for typical print sizes, but it could make just enough difference to go from a really big print to one that's even a little bigger, or could lend itself to a little more cropping after the fact if I decided to second-guess my composition.

Pixel Shift Resolution is the best argument yet for shooting raw

It was well after I'd finished shooting for the evening that I suddenly had a "Eureka!" moment. On the way home, I had to stop for shelter in a gas station, after a sudden pop-up thunderstorm started moving my car around rather alarmingly with 60 mile per hour wind gusts. While keeping safe indoors, I went back over the day's shooting in my mind -- and suddenly a realization came to me. Pixel Shift Resolution is by far the best argument I've seen yet for always shooting raw, even if it means you have to bring a huge flash card or two!

Pixel Shift Resolution processed as single shot in-camera, 31mm-equivalent, 1/100 sec. @ f/5, ISO 200
See the Pixel Shift Resolution version of this image here

You can always turn a Pixel Shifted raw into a single-shot JPEG...

The reasons for that assessment are twofold. Firstly, by shooting in raw format, you don't feel that you're risking anything by using Pixel Shift Resolution. Since the mode creates one gigantic, 100MB+ file that's basically a container for all four original raw files that made up the Pixel Shifted shot, you can always go back to a single-shot raw after the fact if Pixel Shifting didn't provide a good result.

In fact, you can even process a Pixel Shifted raw file in camera, disabling Pixel Shifting after the fact and bake yourself a fresh single-shot JPEG that's identical to what you'd have gotten in single-shot mode. Instead of feeling like you're taking a risk, shooting with Pixel Shift Resolution in raw mode you quickly develop a "nothing ventured, nothing gained" mentality.

100% crop comparison: At left is the image as captured by Pixel Shift Resolution. Note the presence of significant artifacts in the sign and spider grass, both of which moved in the light breeze between shots. At center is the same image, but the raw file has been processed in-camera to disable Pixel Shift Resolution.

Finally, at right is a hybrid of the two images. I erased the areas that were filled with artifacts, but kept those where there wasn't noticeable motion, such as in the hedge, lawn and bricks. The result is an image which is noticeably sharper than the single-shot version, but also relatively artifact-free for just a couple of minutes work.

31mm-equivalent, 1/100 sec. @ f/5, ISO 200
See the full Pixel Shift Resolution image here, a version of the same raw file processed in-camera to disable pixel-shifting here, and a separately-captured single-shot image here.

...and you can also create a hybrid of Pixel Shifted and single-shot imagery

The real reason you want to shoot in raw is even cooler, though. Standing in that gas station, an idea formed in my head. What if I was to take a Pixel Shifted raw file and process it twice -- once to create a Pixel Shifted JPEG, and a second time to create a single-shot one? In theory, that would allow me to layer the pair in Photoshop and select which areas I wanted to use from which version.

In effect, the Eraser tool in Photoshop would become a magical paintbrush, able to paint in more detail where I wanted it if I stacked the single-shot image on top and then selectively erased certain areas. Or alternatively, I could stack the Pixel Shifted JPEG on top and erase portions of that instead, erasing any areas with artifacts and replacing them with a slightly less sharp (but artifact-free) version of the same scene.

Pixel Shift Resolution, 90mm-equivalent, 1/125 sec. @ f/7.1, ISO 400
See the single-shot version of this image here

And guess what? It works absolutely brilliantly. That's not too surprising when one considers that it's effectively what's being done by some in-camera processing techniques already, but until that brief layover in the gas station it simply hadn't occurred to me to try the same thing myself in Photoshop.

It strikes me that Ricoh could make the Pentax K-3 II's Pixel Shift Resolution function even more useful by performing the same technique automatically when processing a Pixel Shifted raw file to create an in-camera JPEG. (Or for that matter, in its Digital Camera Utility app as an automatic tool.) It likely wouldn't be that terribly difficult to have an algorithm look for areas that appear pixelated, and then seamlessly fall back to one of the single-image version for those areas.

Pixel Shift Resolution, 60mm-equivalent, 1/800 sec. @ f/5, ISO 200
See the single-shot version of this image here

Using this trick yourself does, of course, rely on your having two renderings that differ only in their usage of Pixel Shift Resolution, but are otherwise identical. That means you can't process in Photoshop or DxO Optics Pro yet, because neither can handle Pixel Shifted raw files properly. For the time being, you have to either process the raws in-camera as I did, or using Ricoh's supplied Digital Camera Utility software to get two JPEGs, then bring those into Photoshop or your image editing app of choice.

With this simple trick, Pixel Shift Resolution becomes infinitely more useful

"This one simple trick" is an expression one hears rather too often on the internet, and invariably it disappoints. But not this time! Even though Pixel Shift Resolution is really, really sensitive to subtle movement, once I realized that it could be fixed so (relatively) painlessly, it struck me as an infinitely more useful tool.

Pixel Shift Resolution, 31mm-equivalent, 1/250 sec. @ f/7.1, ISO 200
See the single-shot version of this image here

Without this technique, I think I'd probably seldom if ever used Pixel Shift Resolution outdoors, except on a completely still, breeze-free day. With the ability to paint more detailed, Pixel Shifted imagery in or out where I need it, though, it becomes radically more useful.

Hopefully that will be even more true if and when third-parties support the spectacularly large 100MB+ files it outputs. (I'd love to be able to pull a single Pixel Shifted raw file into Adobe Photoshop or Lightroom, or DxO's Optics Pro, do all my processing in the one app to get the image just so, and then select where to use Pixel Shifted imagery as the final step before baking myself a JPEG.)

[Edit 2016/05/10: Current versions of Adobe Photoshop and Lightroom can now handle Pixel Shift Resolution raws without any third-party software, although neither can handle HDR raws yet. To process PSR raws, upgrade to Adobe Camera Raw 9.1.1, Lightroom CC 2015.1.1 or Lightroom 6.1.1 respectively.]

Pixel Shift Resolution, 132mm-equivalent, 1/160 sec. @ f/7.1, ISO 200
See the single-shot version of this image here

Much more to come in my second field test!

And that brings us to the end of my first field test with the Pentax K-3 II. There's still plenty to test in my second field test, so watch this space! In particular, I want to take a closer look at the AstroTracer function, and of course I'll be shooting in low light at higher sensitivities, or with longer exposure times. I also want to try out the claims of better autofocus tracking and image stabilization.

Watch this space for more, coming soon!

Pixel Shift Resolution, 90mm-equivalent, 1/125 sec. @ f/7.1, ISO 200
See the single-shot version of this image here


Buy the Pentax K-3 II

Editor's Picks