Pentax K3 Field Test Part II

Time for some more exciting glass!

Posted: 02/06/2014

In the first part of my Pentax K-3 Field Test, I shot exclusively with the 18-135mm kit lens that can be purchased in a bundle with the camera body. In part, I did so because that's the one lens K-3 shooters are most likely to own, and we always try to ensure we cover kit lenses for that reason -- even if they're not the most exciting optics available. It also didn't hurt that I happened to own a copy of the same lens myself, meaning I could shoot side-by-side with my K-5, with no fussing and changing lenses back and forth.

With a consumer camera, the kit lens might very well be the only lens you'll shoot with, but that's almost certainly not the case with an enthusiast SLR like the Pentax K-3, though. For that reason, I was keen to get out and shoot with some of my nicer lenses. Sadly, I didn't have duplicates of these, so shooting side by side with these meant an exceptional amount of switching lenses between bodies.

Matching setups. My primary goal for the second Field Test was to get an idea for just how much more detail could be extracted from the 24.35 megapixel Pentax K-3 than my own 16.3 megapixel K-5, which shares the same sensor as the later K-5 II and IIs models. I also wanted to see how noise levels compared in typical daylight shooting at lower sensitivities, and how the K-3's rendering compares to that of the previous generation flagships. And lastly, I wanted to try out a couple of the K-3's new features -- the antialiasing filter simulation function, and the high dynamic range shooting mode.

As in part one of the shooter's report, I shot with the cameras on default settings -- the fairest way to compare them -- and then made changes to file formats, drive modes, autofocus modes, antialiasing modes, HDR modes, and so forth to both cameras simultaneously. I also mostly kept the aperture and sensitivity the same or at least near to each other, while letting the shutter speed fall where it may to get a similar exposure between cameras.

The lenses. Although I do own an smc PENTAX-DA 18-135mm F3.5-5.6 ED AL [IF] DC WR lens, as I noted, I seldom shoot with it -- and for this shoot I consciously left it at home. (It's my travel lens for when I don't have enough space to bring a selection of lenses with me.)

Instead, I took some of my favorite lenses out of my camera bag and brought them along. I had on hand three primes -- the DA 21mm F3.2 AL Limited, DA 40mm F2.8 Limited, and an older FA 100mm F2.8 Macro. I also grabbed a zoom which I've been resolving to try and use a bit more, my DA 10-17mm F3.5-4.5 ED (IF) Fish-Eye. The FA 100mm is a full-frame optic that predates Pentax's digital SLRs; the others are digital-specific, APS-C oriented lenses.

Note that my two Limited primes are the earlier generation and lack the flare-busting HD coating of the newer optics. Pentax also says their apertures aren't quite as round, so the bokeh should be more pleasing on newer lenses. Still, they're not a long way away from optically identical, so if you're looking at those lenses alongside a K-3 body, performance will likely be similar.

(Update: Since writing the above, I've tested the HD variants. While I didn't see any change in bokeh, I saw a significant improvement in flare. See some side-by-side comparisons here with the 21mm and 40mm Limited lenses against their HD variants, and look in the gallery to see samples from every other HD Limited lens model.)

Detail. It didn't take long to come up with the answer to my first question. In fact, the only thing I noticed even sooner is that the Pentax K-3's LCD monitor is noticeably less glare-prone, and easier to see in bright, outdoor light.

With a nice, sharp prime lens on the Pentax K-3, it's a worthwhile amount sharper than my K-5. Good lenses clearly yield quite a bit more detail than the earlier 16.3 megapixel sensor is capable of resolving. Part of the difference doubtless comes thanks to the lack of an anti-aliasing filter on the K-3, and I wish I still had a K-5 IIs with which to make an even fairer comparison, but the difference in detail between the two cameras is a lot greater than the subtle difference between K-5 and K-5 IIs.

In fact, it was clear enough that I didn't even have to wait until I got home to make that confirmation -- just comparing images side-by-side on the LCD was enough to make the call, although I certainly spent some time in side-by-side comparisons on the PC too. Giving the K-5 some unsharp mask gives the illusion of closing the gap, at least if there aren't areas of extremely fine detail such as text to tip your eye off to the trickery -- but there's no question the K-3 does a better job in the resolution department.

Pentax K-3 (100% crop of image downsampled to K-5 resolution)

Pentax K-5 (100% crop)

A comparison of output from the Pentax K-3 (top) with the K-5. Here, the K-3's output has been downsampled to match the resolution of the K-5's image, using bicubic interpolation in Photoshop 12.1 (Creative Suite 5.5). Neither image has had sharpening applied. Both were shot handheld with the exact same lens, and focused using phase-detection with the AF point in the same location. Click either image to see the full-resolution original.

Pentax K-3 (100% crop)

Pentax K-5 (100% crop of image upsampled to K-3 resolution; unsharp mask applied)

Here, I've instead upsampled the same K-5 image to match the native resolution of that shot with the Pentax K-3. Again, I used bicubic interpolation, but this time I've applied unsharp mask (75%, radius 1.1 pixels, threshold 0) to the K-5 image, to account for the softness you'd expect in an upsampled image. Again, click either image for the full-res original.

Above and below are a couple of examples, with both cameras tripod-mounted and focused in the same area. Out of curiosity, I first downsampled the K-3 shot to match the resolution of that from my K-5 -- and significantly more detail remains visible even after reducing the resolutuion. Then I tried upsampling from the K-5 to match resolution of the K-3. Either way, there's no question in my mind: There's a very worthwhile increase in detail here, which could come in handy for very large prints, or for when I want to perform significant cropping post-capture.

Another example of the K-3's greater detail capture, this time shot on a tripod and focused with contrast detection. Click here to see a shot with the exact same exposure variables and focus point on the Pentax K-5, and see 100% crops below.

Pentax K-3 (100% crop of image downsampled to K-5 resolution)

Pentax K-5 (100% crop)

A comparison of output from the Pentax K-3 (top) with the K-5. Here, the K-3's output has been downsampled to match the resolution of the K-5's image, using bicubic interpolation in Photoshop 12.1 (Creative Suite 5.5). Neither image has had sharpening applied. Both were shot tripod-mounted with the exact same lens, and focused using contrast-detection in live view mode with the AF point in the same location. Click either image to see the full-resolution original.

Pentax K-3 (100% crop)

Pentax K-5 (100% crop of image upsampled to K-3 resolution; unsharp mask applied)

Here, I've instead upsampled the same K-5 image to match the native resolution of that shot with the Pentax K-3. Again, I used bicubic interpolation, but this time I've applied unsharp mask (75%, radius 1.1 pixels, threshold 0) to the K-5 image, to account for the softness you'd expect in an upsampled image. Again, click either image for the full-res original.

Another nice UI touch. Speaking of the LCD monitor, incidentally, there's one new UI feature on the K-3 that I really appreciate. When you use the playback zoom, there's not only an indication of the zoom level, but also a separate indication when you reach the point where the pixels in the image match those on-screen precisely, at 1:1 scale.

It's perhaps slightly odd that there's a very similar zoom level that doesn't offer a 1:1 match -- as you step through the zoom levels you suddenly notice a much smaller step that, to my mind, could be done away with and I wouldn't miss it -- but having a zoom level that is clearly marked "100%" when the overlay is enabled is great. I wish my K-5 had the same indication!

Knoxville's historic Andrew Johnson building, now given over to office use, is one target that almost always gives moiré and false color, thanks to the many venetian blinds in its windows. It didn't disappoint with the K-3 and K-5, giving me a good test of the former's anti-aliasing filter simulation function. Click here to see a shot with the exact same aperture, ISO sensitivity, and focus point on the Pentax K-5, and see 100% crops below.

Aliasing (and Pentax's unique tool to fight it). One of the more significant changes between my aging Pentax K-5 and the K-3 is the latter's lack of an anti-aliasing filter. As I noted a moment ago, removing that filter does make a noticeable -- albeit not huge -- difference in sharpness. You really do need to be viewing shots 1:1 or printing them at extremely large sizes to notice the difference, but for many shots it's a nice thing to have. My concern with cameras that forgo the optical low-pass filter in the quest for detail, though, is that it increases the incidence of moiré. Once moiré and false color are in your shot, they can be a monumental pain to fix, if you can manage to do it at all.

Ever since I heard about Pentax's unique anti-aliasing filter simulation function in the K-3, described in great detail by Dave Etchells further down this review, I've been keen to give it a test and see how it does. It's not a panacea -- there are certainly limitations to the function. Most importantly, flash exposures often have such a short duration that the system won't have time to subtly blur detail like a real low-pass filter would. It also operates with decreased efficacy at shutter speeds faster than 1/1,000 second, and doesn't function at all for HDR captures or for movie recording.

Pentax K-3 (100% crops, no AA simulation)
Pentax K-5 (100% crops)
Side-by-side, the K-3 with anti-aliasing filter simulator disabled shows fairly strong banding and false color effects. By contrast, with its built-in optical low-pass filter, the K-5 image looks much cleaner -- although it's not completely free from artifacts itself. Look at the original image by clicking the links, and you'll see some more subtle banding from the K-5 in the window at right -- they just appear further up, likely due to the different sensor resolution requiring detail at a different frequency to trigger the issue.

Still, it's absolutely unique, unquestionably clever since it operates using a system the camera already has, and it largely solves the issue of having to decide whether to buy a camera with or without a low-pass filter -- if that option is even available. (For many cameras these days, it's not -- you live without the OLPF, or you don't buy the camera.)

With all that said, it can be a bit of a chore to find a real-world scene that provokes moiré and false color in the first place, but one common culprit -- venetian blinds in the windows of an office building -- did the trick, giving me both moiré and false color in shots with both the K-3 and K-5. (Which goes to show that even with an OLPF in place, you're still not immune to this annoying image defect.) The moiré and false color were definitely worse in shots from the OLPF, but would the AA Filter Simulation function fix the issue?

Pentax K-3 (100% crops, no AA simulation)
Pentax K-3 (100% crops, Type-1 AA simulation)
Pentax K-3 (100% crops, Type-2 AA simulation)
If moiré or false color strike, the K-3 has an anti-aliasing filter simulation function that aims to squash them. It does a surprisingly good job, eliminating the overwhelming majority of the unwanted artifacts in this shot with only a slight reduction in sharpness..

The answer was a resounding yes -- in fact, the shots from the K-3 actually seemed ever so slightly better to me than did those from my OLPF-equipped K-5, although that may just have been the luck of the draw with regards to the frequency of detail in the blinds versus the resolution of the sensor.

I didn't see a huge difference between the Type-1 and Type-2 AA filter simulations, and I do think Pentax needs to do a better job of explaining the two modes -- and the distinction between them -- in the user manual, which provides not one word of explanation. Type-2 seemed to have a slightly stronger blur, though, and it's our understanding from the pre-launch press briefing that it describes a circular motion, rather than a linear one.

AA Filter Simulator bracketing. One feature just added to the Pentax K-3 courtesy of a firmware update -- and I actually delayed this shooter's report just slightly to test it -- is the ability to bracket the AA filter effect, shooting several shots in a row with the effect disabled in one, enabled using Type-I simulation in another, and Type-2 in the third. I'd actually been intending to suggest just this behavior in my shooter's report, as you often won't have the time to check for moiré when you're out shooting in the field, and indeed our technical editor had said much the same when the K-3 first launched late last year. Well, Ricoh apparently has mind-reading capabilities, and they got right down to adding the feature.

I gave it a test, and I think it's a great improvement. Instead of having to fiddle with the menu to shoot all three modes, they're captured in a single press of the shutter button, and within less than half a second -- likely fast enough that you'll get much the same framing and subject pose for all three images, unless your subject is particularly active.

There is one unfortunate catch, however. With firmware v1.02 applied, enabling the new AA filter bracketing function through the menu system immediately disables exposure bracketing, if that was active. (And vice versa.) Personally, I'd like to see the two able to be combined -- shooting off a burst of nine images is still easily within the K-3's buffer capacity, and I can't think of a good reason one function should override the other. (Nor is it the most intuitive behavior for a change of one setting to affect another without any warning whatsoever.)

Still, that aside I'm very happy to see the new AA filter bracketing function arrive before I could even ask for it. Hopefully Ricoh will see fit to provide the ability to couple it with exposure bracketing in a future update.

Noise levels. Any concerns I had about noise at lower sensitivities and in good lighting quickly evaporated once I got home and took a look at my images. Near base ISO, the Pentax K-3 yields images that are indistinguishable from my K-5 in terms of noise -- there's just all that extra detail to play with. Great news!

Punchy color. Overall, rendering is very similar between the two cameras. At defaults, they have much the same "Pentax look", which is quite punchy and saturated, and especially tends to try and add some pep to blue skies and green foliage. Personally, I find it a little too punchy for my tastes, but that's easily resolved. When I'm shooting JPEG, I typically opt for the Natural custom image setting, which gives a more realistic rendering. I can easily understand why it's not the default, though, because consumers do tend to gravitate towards that punchier look.

View the IR Lab's in-depth Pentax K-3 image quality test
results by clicking here. Be sure to read further on to
see side-by-side comparisons of the K-3 and competitors.

Purple fringing. One not-so-desirable attribute I noticed while shooting with the K-3 under bright sunlight is that it seems to show somewhat more pronounced purple fringing on strong highlights than does my K-5. The crops below are from shots taken moments apart, with the exact same lens and subject. This is especially noticeable where you have a very bright highlight adjacent to a high-contrast edge.

Note in the left-hand crops that there's a purple cast at the bottom of the glare on the lamp from the K-3, not present in the K-5 version. Looking back at my night shots, it's noticeable with strong point light sources against a dark background there, as well.

Don't confuse this highlight fringing with the fringing shown in the crops at right, which is more typical and lens-dependent. (We're pretty close to the corner of the frame here.) Both K-3 and K-5 suffer about equally in this regard, although the K-3 shows more blue fringing on the right side, and the K-5 has more purple on the left, plus a little cyan fringing at right.

Pentax K-3 (100% crops, no AA simulation)
Pentax K-5 (100% crops)
Color fringing rears its head in this shot. In part, it's because we're very close to the corner of the frame, but the K-3 also shows an unattractive purple fringe beneath the sharp highlight on both lamps.

These fringes aren't terribly pretty seen 1:1 onscreen, but realistically you'd need a very large print size for them to be noticeable, and the defringing tool in apps like Adobe's Photoshop Lightroom should clean them up without too much fuss. Not a showstopper, by any means, but if you shoot scenes with lots of sharply defined point light sources, it might be something to bear in mind.

Nice rich color from the Pentax K-3, and very similar to that from the earlier K-5. Both images were taken just moments apart. Click here to see that from the K-5, shot with similar exposure variables, including auto white balance and the default "Bright" custom image mode. The "Natural" mode yields less saturated results.

More consistent metering. Compared to that in the K-5, the Pentax K-3's new metering system with its finer-grained metering sensor does seem to be doing a better job. I've taken a look back over my images, and fewer of them seem to need exposure compensation in the first place. Those which do, other than unusual scenes that would confuse pretty much any metering system, invariably seem to ask for just a touch of negative exposure compensation. Doubtless that's partly down to my personal tastes: I don't like to lose highlight detail when I'm shooting JPEG-only. Shooting raw I think the exposures are close enough that I'll need to dial in exposure compensation only for particularly difficult subjects -- the overwhelming majority of shots are near enough to the center of the ballpark that the exposure latitude of raw will get me the rest of the way to a perfect exposure without issue.

White balance from the Pentax K-3 and K-5 was mostly pretty similar, but occasionally they diverged pretty radically. Here, the K-3 did a much nicer job than did the K-5, which gave the scene a very cold cast. Both cameras metered pretty equally, needing -0.7EV of exposure compensation to hold onto the highlights. That's not surprising, given the deeply shaded subject. Click here to see the result from the K-5.

HDR -- served raw! Lastly for this section of the Field Test, I wanted to take a look at the K-3's HDR capabilities. There are two upgrades in the raw department, and between them it strikes me that I'd be much more likely to use in-camera HDR.

As things stand, although my K-5 supports HDR -- even handheld -- I simply never use the feature. That's largely because it can only output processed images in JPEG format. Once I head home, if I decide I'm not satisfied with the result once I look at the HDR in Lightroom -- well, it's probably too late to do anything about it. So I shoot my HDRs as separate images in raw, and process them once I get home -- if I remember. More likely, I get too busy and the HDR never happens at all.

With the K-3, you can save HDR images in either .PEF or .DNG raw format -- and it's not just a processed image, either. You can also control the step size between exposures, something that couldn't be done with the K-5, and which gives you quite a bit more control. The raws actually include all three source images in a single file. There's a positive and a negative to that approach, as I quickly discovered.

Pentax K-3 (HDR 2; 1EV steps)	Pentax K-3 (HDR 2; 2EV steps)
Pentax K-5 (HDR 2; 3EV steps)	Pentax K-5 (HDR 2)
The Pentax K-3 gives more control over HDR shooting in-camera, and it also allows you to save the original, source images in a single, combined raw file. You can even shoot raw+JPEG. By contrast, older models like the K-5 allowed JPEG only for HDR capture, and gave no control over step size. The K-5's result also has much more muted color than that from the Pentax K-3's in-camera HDR files.

On the plus side, if you shoot raw+JPEG as I did in this review, you have matching filenames for both types. And if you shoot raw only, you have a single file that makes it clear to you this is an HDR. On the minus side, though, the file sizes are huge. I'm talking close to 100MB per image huge.

And also, every third-party app I tried -- be it Adobe Photoshop Lightroom, Adobe Camera Raw, or DxO Optics Pro -- didn't recognize that there were three separate images, and cleverly layer them for me or even open them as separate files. As far as every program was concerned, this was a raw file containing a single, non-HDR raw image, and they wouldn't be told otherwise. (Want to try your own program? You can download both .PEF and .DNG HDR raws in the gallery.)

The answer, it turned out, was in Pentax's own, bundled software package, the Silkypix-based Digital Camera Utility 5. (And I'm guessing if you own a current, retail copy of Silkypix, it'll likely exhibit the same behavior.)

For one thing, Digital Camera Utility 5 does recognize the K-3's HDR raw files, both in .PEF and .DNG formats. That in itself felt like a breakthrough, after battling the third-party software without success. It wasn't entirely satisfying, though, because I'm a man of habit. I didn't want to learn another software package -- I wanted to use the apps I'm comfortable with.

Pentax K-3 (HDR Auto; 2EV steps; auto-align)	Pentax K-5 (HDR Auto; auto-align)
The Pentax K-3 and K-5 give similar results with handheld, auto-aligned HDRs. Both cameras do apply a focal length crop if auto-align is enabled, so for really wide-angle HDRs you'll want to make use of a tripod and disable auto-align. Auto HDR was used for both of these shots.

And then I found nirvana. A little option in Digital Camera Utility's Tools menu by the name of "Separation of HDR RAW file" made life great once more. With that one little option, suddenly the one unwieldy raw file could become three, sharing the same prefix and then _1, _2, or _3. The original file type is respected -- if you shot DNG, you get three separate DNG raws. It even provides an option to rename the new files, should you choose.

For the first time in my many years of owning Pentax DSLRs, I think Digital Camera Utility 5 will be staying on my PC, solely to give me access to this tool. And I think I'll be shooting more HDRs, as well, safe in the knowledge I'll have easy access to the fuss-free, camera-processed version, but with the originals on hand should I want to dig deeper.

Image shot with Pentax K-3; click here to see the result from the K-5.