[rdenney]

Series Contents

And now for something completely different.

I had intended to look at both the Hartblei 45 and the Arsat 55 in one post, but it got too long. So, here we'll cover the Hartblei 45mm f/3.5 PCS lens only.

First, a quick historical review: As part of reparations, the Soviet Union cleaned out some of the goodies at the Carl Zeiss factory in Jena, in East Germany, after WWI, and again after the Berlin Wall. Much of the brain trust of Zeiss, with help from the USA, came to the west, and set up Carl Zeiss AG in Oberkochen, West Germany. But much of the tooling went to Arsenal, a big "defense" contractor in Kiev, Ukraine. What was left became VEB Carl Zeiss in Jena, which was much later absorbed into VEB Pentacon based in Dresden (and the progeny of the old Exakta Dresden works). The Soviet Union was always quite competent in optical design and production, with its other factorys in, for example, Krasnogorsk.

Thus, Arsenal in Kiev began producing cameras and lenses. They produced mostly medium-format stuff. Their first camera was a Salyut, which was the forerunner of the Kiev 88, a knockoff of the Hasselblad 1600f with the focal plane shutter. Hasselblad couldn't make the 1600f reliable, and soon transitioned to the 500C, which didn't have a focal plane shutter. The Kiev 88 used a semi-screw mount similar to the 'blad 1600f mount, and Arsenal made a line of lenses in that mount. These were mostly designed in the late 60's and early 70's, so they benefitted from computer support in ways the designers of the medium-format lenses back at Carl Zeiss Jena did not attempt. Thus, their lenses are relatively modern, multicoated lense of good design, at least for the 70's. They made a retrofocus fisheye (already tested), originally called the Zodiac. They made a couple of retrofocus rectilinear wides at 45mm and 65mm, the Mir 26 and the Mir 38, respectively. They made several double-gauss normal lenses, including the 80/2.8 Volna, the 90/2.8 Vega, and the 120/2.8 Vega. And there were longer lenses that became, in my view, less remarkable.The 45 was actually a bit wider than 45, and was one of the widest lenses available for 6x6 camera until more modern Japanese designs came out (including the 35mm 645 A and the contemporary P67 45mm lens). The 45mm Mir 26 was never designed to be a shift lens, and truth to tell it already had a reputation for pretty strong field curvature. But it was pretty sharp in the center, which was the standard of expectation for strong wides in those days.

After the Soviet Union opened up and the satellite states emerged from Soviet control, Arsenal tried to compete on the world market with its cameras. That was the golden era if there ever was one for Soviet medium-format cameras--the 90's and early 2000's. During that period Arsenal came out with some new designs. But what also emerged were the ex-factory enhancers, who took the stuff Arsenal slapped together and fixed it up to something closer to western standards. One of those groups was Harblei, which formed in the 90's from current and former Arsenal employees to sell modified and ex-factory serviced Kiev cameras and lenses. Given the scarcity of wide shift lenses in medium format, they sought to fill that void using the Mir 26, and Harblei designed and manufactured a quality barrel with a ring-controlled 12mm shift on a rotating mount. This became the PCS--a shift lens. later, they added tilt, and then they added independently rotating tilt, which became the Super Rotator.

The barrel design is great. Too bad they didn't have better glass to work with. The Mir is friendly enough, but the advantage one gets moving up to that lens from 35mm is entirely the result of the larger format, not the capabilities of that lens.

But in practical terms, it onlly means that photos from that lens won't be able to be enlarged as much, without the edges going visibly soft.

So, let's start with the Hartblei PCS 45mm f/3.5, which used (supposedly) hand-selected Mir-26 optical components, which were given much better coatings and a perspective shifting barrel with 12mm of shift. The shift scale used red paint for the shift graduations beyond 10mm, which they thought was the limit of shifting while maintaining the best image quality. More shift also requires a smaller aperture to avoid vignetting from the lens mount when shifted. In a later post, we'll look at the Arsat 55m PCS, which was purpose-designed to be a shift lens.

Here it is on a very odd Harblei tilting adapter to adapt it to 35mm (in this case, the Canon EF mount)--the lens is everything that is tilted:


As with most shift lenses, there are no aperture controls through the mount, and the metering is done stopped down. I put the camera on Av and it selected the shutter speed given the manual aperture.

Here is the unshifted image from the 45 (at f/11):


Even at this small size and aperture, we can see that the edges go soft except for what's in the foreground. That's the story of this lens.

Now, let's look at a 1:1 crop from the center, at f/3.5:

Well, that's not very good!

As it turns out, f/11 is sharper than f/8, but at f/16 diffraction cuts it back slightly. That doesn't mean f/16 won't make a sharper looking print, because it will help control curvature--more on that in a moment. Here's the center at 1:1 and f/11:


That's a lot better--usably good.

As bad as the center focus point was at f/3.5, let's look at the lower left corner:


First, we see the effects of vignetting at the wide aperture, making the corner dark. But what strikes me is that the azaleas, which are only about 8 feet from the camera, are somewhat sharper than the focus point 50 feet away. That is strong field curvature. The lens showed fairly strong lateral chromatic aberration, too, and software correction sometimes helps when the lens is not shifted.

The corner also improves quite a bit at f/11:


But what about a more distant corner? Here's the upper right corner at f/11:


Some of the branches in view are near the focal plane, but are being undermined by field curvature. F11 isn't enough to bring in the edgtes and corners unless it's in the foreground. For typical photos, with sky at the top and stuff to see at the bottom, the curvature might not be so much of a disadvantage. And interior room, particular a one-point perspective, might present a tunnel that this lens would handle better than a landscape.

Let's look at the shifted performance. Here's the full image with 12mm of left shift:


The lamp was still the focus point--I shifted after focusing. The camera is not perfectly level, thus the wall is at an angle, and there is also a bit of barrel distortion.

Looking more closely at the lower left:


I didn't go to the corner because the wall was out of focus. But the flowers were not out of focus despite being much closer than teh focus point. This shows strong field curvature.

Looking just over the top of the azalea bush, the lower leaves are about half the distance to the focal plane, and the grass behind it is behind the focus plane by a lesser amount:


Again, f/11 can't sharpen up the edges unless they are in the foreground. Notice also the lateral chromatic aberration. Software won't fix it on a shifted lens, because the CA is not symmetrical about the center of the image, and software corrections assume that it is.

So, what aperture would it take to keep the focus plane somewhat sharp to at least close to the edge? Here's f/22:


For this lens, f/22 is the aperture to use if the edges are important, and just limit print size to what can tolerate the resulting diffraction. But the above isn't that bad--remember these are snips from a print that would be 7 feet wde if your monitor has 100 pixels/inch. And sharpening would help, too.

This lens will not be going to Alaska with me. I would be much better off using the FA 35mm and repairing perspective convergence in software.

Tomorrow, I'll show the alternative, the Arsat 55mm PCS. It's better.

Rick "sometimes, depth of field does more good than diffraction does harm" Denney

---

Last edited by rdenney; 06-30-2018 at 12:14 AM.

[BigMackCam]

Very interesting test, Rick.

Regarding the lateral CA on that twig in the f/11 shot, I was able to remove the red and blue fringing quite easily using the "defringe" tool in Darktable. Worth knowing if you should find yourself with some photos that are suffering badly from it

[rdenney]

Very interesting test, Rick.

Regarding the lateral CA on that twig in the f/11 shot, I was able to remove the red and blue fringing quite easily using the "defringe" tool in Darktable. Worth knowing if you should find yourself with some photos that are suffering badly from it

It's easy in the clip to remove the effect. But lateral color is usually symmetrical around the center of an image, and all the CA removal software I've seen makes that assumption. The problem is that the center isn't the center when the lens is shifted.

You can remove the color using other tools, but the trick to CA removal is getting the colors to realign by applying the different geometric distortion corrections to each color.That actually makes the image sharper and more resolved, beyond just removing the color fringe. DXO Photolab applys that sort of correction for lens and camera combination in its database. Photolab's removal of purple fringe (which is a focus issue, not a lateral distortion issue) is the just finds purple edges and desaturates the purple. One thing haven't figured out about Photolab, yet, is how to apply a proper lateral color correction even when the lens is not in their database. The tools they have are not that powerful without that. Photoshop's is a little better, but it does not allow one to decenter the effect, and it doesn't work with shifted lenses at all. You can make one corner look good but the others look worse. (I did make a correction once on a photo made using the Canon 24mm TSE Mark I lens by creating an image canvas much larger than the image, and placing the image in the larger canvas according to its shift, so that the lens center was in the center of the canvas. Then, the CA lens correction tool in Photoshop would work. After applying that tool, I cropped off all that extra canvas. PITA.

Rick "who'd rather shift the color than just desaturate it" Denney