Acutance and the Pentax K1

[Michael Erlewine]

I am surprised that there is so little interest in the Pentax K1 camera and in particular the SuperRes or Pixel-Shift mode of this camera. The degree of acutance is beyond anything I have experienced and either others don’t see this or there is just generally no interest in this kind of accuracy. I am looking elsewhere to post on this topic, but I don’t see interest in the other forums I frequent either, except, of course, for the Pentax forum itself, which does have interest, but they are already hardcore fans.

So, for what it’s worth, here is a shot with the Pentax K1 in pixel-shift mode and the Voigtlander 90mm APO-Lanthar lens. It is stacked, but lightly (3 layers) and with ISO 100 and f/11.

[Peter Connan]

Michael, to be honest I can see no difference between the quality of the photos you post from the K1 than from the other cameras you use.

I will take your word that there is a difference, but perhaps it is not really obvious at the file sized used for sharing on the web?

Also, as I understand it, the process is one of taking a number of photos and combining them. Unfortunately, almost nothing I take photos of stands still for long enough to do this. Even when shooting night starscapes, I am pushing the limits of shutter speed versus movement just to gather enough light (or rather to minimize ISO noise). In this situation, you generally have some static objects and some moving ones.

Using masks in Photoshop it is possible to seperate these elements, in a batch of photos, stack the batches together and later re-combine these elements. So in other words, doing the same thing the K1 does, but selectively and with a far greater number of photos. Much more work I am sure, but I wonder if these same techniques could not be used before focus stacking?

[Bjørn Rørslett]

The better test material for any evaluation of difference in acutance would result by no stacking at all. Sorry to point out the obvious.

There are plenty of stacking artefacts in the background and the shadows do not look very nice here. The main subject is crisp and fine, though.

[Tristin]

The studio still-life examples of the K1's pixel shift are certainly superior to the other 36+ mp cameras. It competes with.  I don't think anyone would argue against that Michael.  The real question is; Is that extra technical edge worth backing out of an expansive camera system and possibly loosing access to lenses one loves?  That is why the K1 isn't the hotcake it could be.  The three FF mounts it competes with absolutely dwarf it in terms of options.  Had Pentax taken a bolder move and switched to a new mount that was highly adaptable, the K1 would be a major force.  Most people can only afford to own one mount and K mount really is not a mount to envy.

The other issue is to what extent a person is willing to go for that last few percent of technical edge.  In my experience, chasing technicality too much usually means unconsciously leaving the artistry farther behind.  I do not mean in photography solely.  There are plenty of virtuoso musicians cranking out a million empty notes in a set because they spend all their time working on chops instead of pondering what they want the notes to mean.  This isn't to say that people willing to jump through hoops for the K1's IQ are lacking in artistry.  I am just sure that many of the photographers who can afford multiple mounts are hesitant to commit too much of their time and attention to that last one percent of technicality, knowing full well that everything we do is an act of balance and compromise.

[Michael Erlewine]

I hear you and I agree. I tend to see what I can do with a stack, rather than look at one single layer. Here is a single layer from that group of three, if that makes anything clearer.

[Bjørn Rørslett]

Strange - this looks completely different. Quite dull and flat in tonality, in fact? All sparkle and feeling of sharpness ('acutance') has vanished. I wonder whether there is an issue with colour spaces going on here?

If this is for real it won't convince many not committed to the Pentax cause at present ....

[Akira]

In order for the acutance to make more obvious and common difference, we would need either huge prints or high-res monitors (like the yet-to-see 8k ones?).  Your D810 images have already been stunning enough on my humble monitor.

[Michael Erlewine]

There is plenty of color here.

[Tristin]

The differences are quite noticeable, and lovely, on my 4k screen.

[Michael Erlewine]

Another

[Akira]

Michael, the last berry shot easily transcend the acutance of my own retinas!

[Michael Erlewine]

Michael, the last berry shot easily transcend the acutance of my own retinas!

What I am trying to point out here, without having time to prove through a lot of experience with this camera, is that there is something here with the deeper color/luminance gathering that transcends what I have been working with via APO lens correction. This is obviously a missing ingredient to the recipe for a fulfilling image. I am sorry that Nikon did not introduce me to this, and so on. We may be at a turning point, where all kinds of new paths will open up in photography. I don't have time or money to follow them all, but I can't help but follow this path far enough to verify this new technology, new to me. In this case, I don't have to fake interest. All I have to do is look at the results, and I know I have to go further.

[BW]

I'm sorry I lack the enthusiasm for sharp pictures, but for me photography is a result of several pieces coming together. The technical aspect of photography is IMO less important than the artistic. But thanks to the technical knowhow of the people at NG, including you Michael, I can learn more tech stuff and become a better photographer in the end When talking about the K-1, I see no reason why this technology have to be built in to the camera. It might as well be a software based feature? As long as its based on tripod work and more than one exposure there is no reason to get it SOOC?

[Bjørn Rørslett]

The question apparently never asked is how much smearing of the finest detail is sacrificed in order to improve on colour fidelity? Surely the pixel shifting must also act as a low-pass filter as there will be the inevitable positional errors either on the camera or lens side or even more likely, with both. These errors have to act as a spatial convolution of detail vs image co-ordinates. The lens itself must resolve to an insanely high level just to avoid it being the major low-pass component in this scheme.

The spatial autocorrelation of colours tend to be high at short distances and that is the direct reason why the Bayer matrix technology works at all. For many, but perhaps not all, subjects, the potential resolving power needed to fully portray all the detail could well be be higher the what is required to resolve the subject colour variation.

I do not possess a clear answer to the above, but any inputs or insights are more than welcome.

[Michael Erlewine]

Well, Bjørn, I for one wait to see what you come up with. As I am not as technical as you are, my approach is to see if I can get from actually shooting what I am looking for in an image. Certainly, if I understand it right, and you correct me, Bayer Interpolation is nothing by a very sophisticated smear, no? Please explain.