The Pentax K1: True Color per Pixel

[Michael Erlewine]

Some years ago I was doing my best to point out that in my search for “sharpness,” it was not just a matter of resolution and acutance, but results were also dependent on how well the particular lens was corrected, thus APO (Apochromatic) lenses produced, to my eyes, a better (“sharper”) image because the various chromatic, etc. aberrations had been removed.  I got very little response from photographers on this issue, other than Lloyd Chambers.

Because of this apparent improvement in color and sharpness to my eyes, I gradually found myself using APO lenses, and the better the correction, the better I liked the photo results. The takeaway was that instead of just looking for greater resolution, etc., we might rather look for better color representation. Of course, this casts aspersion on the Bayer matrix sensor.

Now comes the Pentax K3II (24 Mpx) and the new, and larger, Pentax K1 (36 Mpx), with their Pixel Shift Resolution, sometimes called “Super-Resolution, which is perhaps a misleading term. What is “super” is not that it has greater resolution (more megapixels), but rather that it has more accurate color, much more accurate color.

Pixel-Shift Resolution not only effectively removes the possibility of any moiré, but without the softening effect of an optical AA filter. The Pixel-Shift mode also makes use of an EFC (Electronic Front Curtain) to avoid shutter-shake and a method whereby four exposures are taken at high speeds, while shifting the sensor by 1 pixel (in different directions) for each one. These separate images are then combined to produce a single photo with much greater appearing resolution, still at the same overall dimensions. Effectively, by taking four separate images and combining them, more light is captured, and a greater dynamic range is possible.

Using this method, no demosaicing interpolation is needed. It provides true color for each pixel and, as mentioned, it avoids the issues of the standard Bayer matrix sensor. Four complete sensor maps are recorded and combined, Red, Blue, and two Greens. Each sensor map is 14 bits, so every pixel has 56 bits of data recorded.

This Pixel-Shift feature works for still photography in the Pentax K3II, and the new Pentax K1 claims it also now works for photos of moving subjects, but I have not been able to try that.

In essence, this new camera, unlike our conventional Bayer systems, where each pixel has but a single color data-unit, the Pentax K1 in Pixel-Shift Mode captures all the color data for each pixel, with the effect of providing much clearer color and definition.

I would think that this approach would interest a lot of folks on this forum. As a still photographer, this is right up my alley. I sent my Pentax K3II camera back, and am waiting for a copy of the new Pentax K1. In the meantime, I am collecting Pentax-able lenses, and managed to pick up a Voigtlander 125mm F/2.5 APO-Lanthar in Pentax mount, the two versions of the Voigtlander 90mm (and macro), and the Voigtlander 40mm Ultron.

[Andrea B.]

The K1 does interest me. But I'm simply not sure about what lenses I would use on the K1. And then I already have way too many cameras currently. Anyway, it will be quite interesting to see your tests with your soon-arriving K1 !!!

I have a full spectrum converted K5 and never thought the Pentax colours were particularly accurate which is not to say the Pentax colours weren't beautiful - they were. Of course colour profiles can cure that if one puts in the work.

The pixel-shift idea is good and in its infancy. It will be interesting to see how it progresses. I wonder if Nikon will get into the game too?

BTW, you mentioned APO lenses.....did you see Jim Kasson's write-ups about LOCA and focus shift in four macro lenses? He worked with the Leica 100/2.8 Apo Macro-Elmarit, Zeiss 100/2.0 Makro-Planar, Nikon 105/2.8 Micro-Nikkor VR, Sony 90/2.8 FE Macro and the Coastal 60/4.0. I recalled your painstaking experiments with some of these lenses when I was reading Jim's write-up.
Experiment begins here and continues over several days: http://blog.kasson.com/?p=14704

[Erik Lund]

Some years ago I was doing my best to point out that in my search for “sharpness,” it was not just a matter of resolution and acutance, but results were also dependent on how well the particular lens was corrected, thus APO (Apochromatic) lenses produced, to my eyes, a better (“sharper”) image because the various chromatic, etc. aberrations had been removed.  I got very little response from photographers on this issue, other than Lloyd Chambers.

You state this almost as if you have invented the term APO,,, Not very humble - Considering all the help you have received here on NikonGear,,,

[Michael Erlewine]

You state this almost as if you have invented the term APO,,, Not very humble - Considering all the help you have received here on NikonGear,,,

Where did that come from? I have always and often pointed out how grateful I am for all of the guidance I have received from Bjørn Rørslett and NikonGear. I have thanked you as well, and others.  On this APO issue and "sharpness," I did not initially get much support. It took years. If I have offended you or anyone here, I apologize. I am just telling it as I understand it.

[Bjørn Rørslett]

Something must have been lost in translation, I surmise. Let's leave it at that.

Like any other lens feature, APO may or may not be the solution to a given problem or contribute to a desirable 'drawing' of a lens.

One should also realise APO rendition might not work equally well across the entire focusing range of a lens, as there are inevitable optical compromises to be made by the designers. For example, the APO capability of the the legendary 125 mm f/2.5 APO-Lanthar starts to break down towards 1:1.

[Michael Erlewine]

Erik, I totally appreciate all that you contribute, and am a big fan of your work! I love what happens here on NikonGear and always have. I was pointing at one single issue, not all the wonderful stuff we all have shared here.

What I found interesting in the Pentax K1, enough to post on, is this technique to get a more pure pixel color. I can see that it is at the heart of what I have been trying to figure out for a long time. I understand that this approach is what makes some Hasselblad cameras so great, but I have never been able to afford one of those. The Pentax K1 is affordable, and for close-up, still, tripod photography may be very helpful. 

And as we all recognize, APO has no standard definition and serves almost as a catch-all for everything we feel is lacking in a photo, at least for me. This concept of pixel-shifting to produce more pure color is wonderful. I can more plainly see that much of what I hunger for has always been about color, not just resolution, acutance, or lens correction.

Andrea: I am looking into that link, but it is not something I can just scan through, right?

[Michael Erlewine]

Some more interesting results from Lloyd Chambers on his free blog:

http://diglloyd.com/

[Anthony]

Thanks, Michael, the K-1 is not for me, but it is fascinating to read of a technological development like this which may come into general use one day.

[Andrea B.]

http://blog.kasson.com/?p=14871

Michael, I think that link has a summary of the LoCA results for the 5 lenses that Jim Kasson tested.

[Michael Erlewine]

http://blog.kasson.com/?p=14871

Michael, I think that link has a summary of the LoCA results for the 5 lenses that Jim Kasson tested.

I saw that. His selection of lenses is eclectic IMO. I don't know the Sony, but I know the rest and have them or have had them. I happen to really like the Leica, but it is a pain to use on a Nikon body. I converted it.

There are so many factors, and this pure color in the new Pentax K1 is another whole factor, and one very much aimed at what I care about. Correction is not enough, if the color is muddy.

[BruceSD]

Michael, one should not expect a Nikon centric blog like NikonGear to be all that welcoming to discussions about the new Pentax K1 camera.  Especially when for subjects that don't move fast, most reviewers are finding that it's IQ when in the Pixel Shift mode greatly exceeds that of the Nikon D810.

You might want to consider posting your K1 finding here  -  http://www.pentaxforums.com/forums/169-pentax-full-frame/  .   

Manufacturer specific blogs tend to be provincial, and you'll get a warmer welcome from folks who are already very familiar with the K1 and Pixel Shifting over at  PentaxForum.com.

No Andrea, Pixel Shifting is not in its infancy.

[Hugh_3170]

Nope, NikonGear is a brand agnostic site.  Many of us also utilise non-Nikon equipment.  For example I also use micro 4/3 and have a few other non-Nikon cameras and lenses as well.

The event of sensor shift cameras such as the Pentax K1 and the Olympus M43 bodies is of general interest, and the "pioneers" such as Michael that test such developments in applied settings add a lot to this site.

FWIW, I welcome the K1 discussion in this thread.  We are all different and I agree that the mileage for others might differ.  Just my 0.02c worth - over & out!

[Tristin]

Bruce, that statement is not true.  The founders of this forum are quite vocal about this forum's inclusion of all photographic tools, manufacturer be damned.

[Andrea B.]

Admin Note:  Really & truly, discussions of any and all gear are welcomed here on NG. Long long ago the site started as Nikon oriented, but that is old history. We are brand agnostic because it is the photographs which are ultimately more important than the gear used to make them.

I myself have a Sony a7R, a Pentax K1, a little Pentax Q, a sturdy old Lumix GH1 that just won't wear out -- and -- a *cellphone* haha -- all in addition to my Nikons.

*********

Michael, I was thinking like you - with per pixel colour from a pixel shift camera combined with a good sharp (resolving) lens having few colour aberrations, the results should be stellar.
BTW, Lloyd reports ACR difficulties for K1 conversion of pixel shifted files. Did you see that? I'm sure it will all get worked out eventually by the various converter developers, but there may be some temporary pain at the beginning from having to use Pentax software unless Silkypix has miraculously improved somehow.

[Michael Erlewine]

Adobe has already stated they will fix the “bluish tinge” in the Pentax K1 files in the next release of ACR. I have been with Nikon Gear for many years and the folks here look at all kinds of cameras and lenses, so I don’t find NG chauvinistic when it comes to brands. A lot of us appreciate Nikon for good reasons, but I try all kinds of new cameras, and usually send them back.

I post on other sites, but mostly here because there actually is an interest in technical considerations of all kinds in this forum. The Bayer interpolation scheme has dominated most DSLRs and while each camera brand has their own approach, they still accomplish the same thing, but Bayer means that there is not a unique RGB value for each pixel location. Instead the three colors interpolate (smear) RGB values to come up with a value for each pixel.

Refining how well a lens is corrected (APO) can only take us so far, it seems. So, when a method comes along such as Sensor-Shift, that uses four complete exposures (R,G,G,B), although it takes up a lot of memory, it promises us unique pixel color, and this is an attention getter, at least for me. It has long been my persuasion that the secret for what I am looking for in my work is color accuracy, and not just acutance, resolution, or degree of correction, although they all are important. However, I had no way to test this.

I found some evidence in the Sony A7s (which I sent back) when it gave more size to each sensor well or whatever we call it. That was instructive, but too small of a resolution for me in pixels.

Looking at some of the pixel-shift examples of the Pentax K1 within Lloyd Chamber’s columns, it is obvious to my eyes that this is exactly what I need for the close-up work I do. I have been thinking that 50 of 100 Mpx is what is important, and it would help, but I can see that purity of color as shown in the same 36 Mpx size as my D810 is more important.

As for earlier comments about my not crediting all the help I have gotten over the years from members of this forum, I am only too aware of that help and so very grateful that this forum has reconstituted itself. In my comments, I was not speaking, in general, but about one specific area where I had to do my own work and find out for myself what works and what does not, and that exercise was good for me.  And is this not what many of the forum members here do all the time, test things and share them with us? I am trying to learn to do the same thing, as best I can.

I am more a photographer than a good tester, so this was one small area (correction of color) I had to figure things out in, rather than have them presented to me. As they say: it is all good.