[rdenney]

FINAL EDIT: Turns out, this lens was damaged. It has been returned, and a replacement purchased. A new thread will be started. Do not consider these results to represent this lens when in good condition.

Series Contents

Yes, the BEAST is now mine. I was chasing one on ebay (surreptitiously), but the auction disappeared and reappeared at a somewhat higher minimum bid, suggesting that the minimum bid I was hoping to snag it at was unacceptable to the seller. Fine. But Adorama had a used example for $100 more than the new minimum bid, and for a hundred bucks I'll take the ability to return it. Good luck, seller. Adorama called it excellent, but it really is a bit marked up in unimportant ways, so I would have called it good. But they had it priced appropriately for its outward condition, so that's fine. I'm happy to let someone else cry over the first bit of brassing on the edge of the filter ring. But Adorama did something that they often do that annoys me. KEH does it, too, often enough. They scavenged the hood off of it to sell separately and so I have a hood on order from a Japanese seller. That unsweetened the deal by about $70, but it was still several hundred less than offerings at the other standard suppliers (KEH, UsedPhotoPro, Beards and Hats, etc.).

Lots of stuff to look at with this lens. I had better lighting and far fewer leaves on my test scene, so the look will be different that with the other tests in this series. But I did something I did not do with my previous tests: I left the tripod in the house. I was fully expecting to have to go back and get it, but it was cold, and the wind was blowing annoyingly, and so I just stood there and made the photos. Thus, this is a test of shake reduction, too.

All photos were made at ISO 200, and given the quick default that DXO-Photolab provides to the DNG raw files, which does NOT include sharpening. I rather think the conversion to sRGB left them rather darker than it should have, and I went back and corrected it only when it mattered.

Here's the test scene at 28mm:



Compare this with the test scene from the 645 FA 35mm test:



It's a little misleading because I wasn't in exactly the same position by a couple of feet, but the 28's sweep is definitely wider than the 35's. I was about two feet to the right and maybe a foot closer to the scene in the top photo.

Don't draw conclusions from the rendering and color. The light was too different. The former test was a cloudy, humid day in the Spring when green was hypersaturated and abundant. And the top photo was made on November 10, on a crystal clear, blustery, dessicated, post-cold-front late-Fall day.

So, let's look at the center crops. These are 1:1, and if your monitor has a pixel resolution of 100 pixels/inch, as most approximately do, these are pieces of an image that would be seven feet wide.

28mm, f/4.5, center crop:


f/5.6:


f/8:


f/16:


In the center, at least, stopping down provides no benefit. This lens is as sharp at f/4.5 as at f/5.6, and then diffraction starts to become visible at f/8 and especially at f/16. The strategy with this lens is to used the widest aperture that provides the needed depth of field.

So, what about at the edges? Here's a crop from the very top of the f/4.5 image:



The thin branches are in the focus plane, and the trunk bark at the left is maybe half that distance. The pines in the background are much farther away than the focus plane, and they are not sharp because of lack of depth of field. But at 28mm, there is a lot of native depth of field, even at f/4.5. But I see no loss of performance at the top edge.

Let's look at 35mm, which provides an opportunity for comparison with the FA35. I'm comparing f/8 on the 28-45 with f/11 on the 35. I'm also comparing a hand-held photo at 1/25 shutter speed with a more carefully made photo using a tripod. And it makes a difference.

Here's the 28-45, hand-held at 1/25, at 35mm and f/8:



And here's the FA35, at f/11:



I'm blown away by the overall performance of the 28-45 in fast-moving circumstances, but it is NOT better than a fixed prime used with careful technique. I will go back and make more careful photos with the zoom lens, and update this post.

The shutter speed at 45mm was a bit slower, at 1/15, aperture at f/8. Let's see how that one went:



And here is the 45mm end of the 45-85, which is the end of the longer zoom most favored.



I'm reasonably impressed that it's as sharp as it is at 1/15 shutter speed, but, again, technique matters, and shake reduction is no substitute for a tripod when expectations are high. I'm not done with this part of the test.

If it turns out that the above is what I get even wtih a tripod, the lens may--I said may--go back.

But, while I wait for tomorrow to pull out the tripod, here are some other shots to test other features--shake reduction and the new coatings. First, a general shot of the front yard in a different direction, with a few minutes of fiddling in Photolab. This is at f/8 and a shutter speed of 1/20. Shake Reduction might not replace a tripod, but it can sure replace a faster shutter speed hand-held. I would normally expect to need a shutter speed of 1/125 to get an image this crisp, even at 28mm. But, also, the lens contributes to a rendering that reminds me of good transparency film. There is a abundance of tonal information to work with, in addition to resolution. And that's with the colors a bit washed out by the sRGB gamut conversion.



Another review of this lens complained about shake reduction. Shake reduction works for rotations about axes normal to the centerline of the lens, which pilots recognize as yaw and pitch, but it does not work for rotations around the centerline, which pilots call roll. So, at marginal shutter speeds the image will be sharp in the middle but may show some movement at the edges. Shake reduction also doesn't correct translational vibration--the camera moving side to side instead of rotating. The image can't tell between the two at distance, and lateral motions at distance are usuallty negligible, because a tiny motion at the camera equals a tiny motion of the scene--too tiny to have an effect. But at close focus distances, it matters. So, I made a photo at close distances just to see what happened. Here's the trunk of that river birch from about three feet away, with the lens zoomed to 45mm, at f/8 and a slow shutter speed of 1/20:



And here's a 1:1 crop from the center at the focus point:



And, to see if shake reduction wasn't enough for the corner at this distances, here's the lower left corner:



I think shake reduction works pretty good in maginal conditions when using a tripod isn't feasible.

Finally, we need a sunset shot for the Grand Lens Test to ride into. I made this photo because who can resist a sunset photo? But I also made it to test for flare. And there isn't much with this lens. 28mm at f/8, with a shutter speed fast enough (1/160) not to be about shake reduction:



Here's a crop from the sun's part of the image to show that flare is minimal under extreme test, both lens reflections and veiling flare.



Rick "somewhat mixed results, and more testing is required" Denney

---

Last edited by rdenney; 12-01-2018 at 02:17 PM.

[rdenney]

Rather than try to edit the above, I would like to supplement the test with a second part, to provide a more complete test and to try to figure out this rather Quixotic lens.

I need advice.

First, let's go back and look at the test scene. I photographed them again, on the tripod and as closely as possible from the original position. The day was much brighter, but I can't help that.

Here is the full image at 28mm, and here we can see how wide it really is at 28.



Compare this with the 30mm Arsat Fisheye:



So, let's look at the 1:1 crops from the center, which on a 100 pixel/inch monitor would be a piece of an image seven feet wide. Here it is at 28mm and f/8 (I'll only look at f/8 for Part Deux). I'm trying to figure out why this lens, for all its vaunted reputation, isn't as good as lenses I already tested. But I need to make sure I have applied sufficient technique to make a fair comparison.

28mm, f/8, 1:1 center crop:


Not super sharp.

35mm, f/8, 1:1 center crop:


Compare to the FA 35mm lens at f/11:


Depressing.

45mm, f/8:


Compare to the 45 end of the 45-85:


I think I want to cry.

So, I checked all details. SR was turned off--it was on a tripod. The tripod is the same one I used before, and it's a thousand-dollar Gitzo. I checked focus 36 ways from Sunday (that old saying will confuse non-Americans--read it as I checked focus using every possible tool, including live view magnified to full with focus peaking). I turned off every feature in DxO that could possible affect sharpness and resolution. But so many other casual images I've made with this lens since it arrived seemed so sharp. What the heck is going on?

I was frustrated, so I wandered around the yard making images. Here's a bit of what I got. All these were at f/8:


And here's a 1:1 crop from that image:


Well, that looks pretty good. And at 16x on the camera's screen, I thought it looked good then, too.

Here's another one:




A slight touch of lateral color, whose correction I had turned off.

And another one:


Look at the tips of the feather.


And it's even sharper before being uploaded to the forum server. Nothing to complain about here.

And this:


I wanted to see how it did with all that detail at a long distance from the camera. Here's a crop:


Nothing to complain about there.

So, I went to a wall so I could get back to approximately two dimensions.


I specifically aimed at a focused on a nail head on the wall. Here it is at 1:1:


I looked at this image at 400% for a while (part of an image 28 feet wide). The upper nail is a bit sharper than the middle nail, but the nails are sharp enough. Here it is again with the normal sharpening applied:


I just can't figure this out. This lens makes stunning images--sharp, contrasty, and clear. The strands of that feather are 1-2 pixels wide. Looking at the forest detail at 400%, I see that the image isn't sharp to the exact pixel, but it seems so sharp at 1:1 (which is already a huge enlargement). Is it that the contrast is so good at such high spatial frequency that it seems hyper-sharp when it is only really sharp?

Does this lens have a pixel-peeper filter? Test images aren't as sharp as photographs made with intent?

Or do I have a bad example at the sharp end? I need to know--my return privileges will run out soon.

But the real images of real things just look so dang good!

A colleague on another forum tried one of these, coupled to a 645z package he was considering, and found that the image seemed to have a weird wavy sharpness, such that it was just impossible to predict what would be sharp to the pixel. But the images all seemed so richly detailed. He sent it back and didn't purchase the Pentax.

Rick "who would be crying in pleasure but for the reputation of this lens, and that damn lamp post" Denney

[kaseki]

I don't know what is going on either, but here is a suggestion for a tedious experiment.

Hand focus at some distance, maybe half way from porch to lamp post. Do not change focus. Do not change aperture. Using a resolution target or similar fine structure subject, take photos at different distances, both along the optical axis and half way to the edge of the field. Repeat for a much different lighting condition.

Evaluate the images for whether the depth of field that you see agrees with theoretical values. Evaluate the images for whether the depth of field is different on-axis versus off-axis. Evaluate when the sun is shining on the target versus when the target is shaded. It looks to me in your exhibit above that dimmer objects are less resolved. I'm not sure how that can happen in physical optics when way above the quantum limit, but maybe an internal camera algorithm is meddling. For example, it could take less bright pixel data and smear it a bit to ensure no noise in any of the computed color channels. However, why wouldn't that be seen using other lenses?

[rdenney]

I don't know what is going on either, but here is a suggestion for a tedious experiment.

Hand focus at some distance, maybe half way from porch to lamp post. Do not change focus. Do not change aperture. Using a resolution target or similar fine structure subject, take photos at different distances, both along the optical axis and half way to the edge of the field. Repeat for a much different lighting condition.

Evaluate the images for whether the depth of field that you see agrees with theoretical values. Evaluate the images for whether the depth of field is different on-axis versus off-axis. Evaluate when the sun is shining on the target versus when the target is shaded. It looks to me in your exhibit above that dimmer objects are less resolved. I'm not sure how that can happen in physical optics when way above the quantum limit, but maybe an internal camera algorithm is meddling. For example, it could take less bright pixel data and smear it a bit to ensure no noise in any of the computed color channels. However, why wouldn't that be seen using other lenses?

I wondered if the difference in lighting could account for what I'm seeing, but I rather doubt it. Notice that the lack of sharpness is uniform for the lamppost, the hedge behind it, and the pine trees behind that. The lamppost is about 50 feet away, the hedge another 30 feet behind that, and the trees maybe 60 feet beyond the hedge. At f/8, the 45-85 was sharper at all those distances, and the 35 was much sharper at all those distances--too much to be accounted for by one stop. I looked for details that have about the same exposure on both, too.

But I have also looked at images online made with the 28-45, including full-resolution images. And they are sharp to the pixel in ways this is not.

When I unpacked the lens, I noticed there was a bit of rattling. I thought it was the SR mechanism that needed power to be fully stabilized. And indeed when the lens is attached to the camera there is no such noise. But now I wonder if there is a loose element or group in the lens that falls into correct position in some attitudes and out of it in others.

Roberts Camera, from whom I have bought with confidence in the past, has one for sale that is only a little pricier than this one, but they have a 15% restocking fee on the lens and now my confidence is shaken. I doubt any used-lens purveyor will test as carefully as I do, but usually it has not been a problem. Adorama has also been good to me, but it's not reasonable to expect them to perform Roger Cicala-like lens testing (and that's why Lens Rentals' used examples of this lens are priced much higher).

I have tested the autofocus calibration by autofocusing, switching to manual focus, starting live view, and magnifying the image. Focus has been perfect every time. This is also confirmed by my microprism screen.

But I think it has to go back. I can't wrap my mind around this lens, which is quite expensive even on the secondary market, not being as sharp as the FA 45-85, or as sharp as examples that have been reviewed online. If I pay $200 for a lens, and it's not perfect but it is entirely usable for my purposes, I'll probably keep and enjoy it. But if I pay 15 times that, I expect it to be one of the best lenses I have ever experienced.

Rick "bummed" Denney

[kaseki]

...

But I think it has to go back. I can't wrap my mind around this lens, which is quite expensive even on the secondary market, not being as sharp as the FA 45-85, or as sharp as examples that have been reviewed online. If I pay $200 for a lens, and it's not perfect but it is entirely usable for my purposes, I'll probably keep and enjoy it. But if I pay 15 times that, I expect it to be one of the best lenses I have ever experienced.

Rick "bummed" Denney

I agree. If the price only incorporates the depreciation of having been pre-owned, then it has to be essentially perfect otherwise. And in any case, even a lower priced example with significant cosmetic blemishes fails the photographer if it fails its optical design intent.

[Sarnian]

Rick,

I've owned this lens and you definitely have a bad copy - probably a loose element, like you summise.

[rdenney]

Rick,

I've owned this lens and you definitely have a bad copy - probably a loose element, like you summise.

Thank you--that's reassuring.

I have it boxed up to send back to Adorama. Adorama is a good company--not a bit of concern about accepting the return.

Rick "the search continues" Denney

[Chapel]

I found with my 90mm macro if I don't use the 12 second delay on the tripod that the pictures are soft with the SR off or on. With the 12 second delay they are amazingly sharp. Maybe that's something to try.

Greg

[rdenney]

I thought the same thing at first, and took measures to prevent it. I used mirror lockup with this lens when on the tripod, but frankly my walk-around shots were sharper using SR and hand holding. My tripod is extremely stable. Example: I got sharp images using the 300 A*, which has no tripod ring, but would show double images when using a Canon tripod ring with what felt like an absolutely rigid shim. I replaced it with a different ring and get no motion effects. The 400/5.6 images were also sharp, even without mirror lockup. That Gitzo 3-series carbon-fiber tripod with an Arca Swiss ball head is amazingly rigid.

To be honest, I would replace a tripod or head that required a 12-second delay for a lens as short as 28mm.

Rick "owner of six tripods, ranging up to Manfrotto's heaviest--used for many years with large format--which is no more rigid than this Gitzo" Denney

[Chapel]

I was trying to calibrate the focus of my lens and nothing was sharp with a 2 second delay. My hand held was way better. I put it on live view fully magnified with focus peaking and it takes 9-10 seconds before all the movement stops.
Greg

[MFullana]

Rick,
I have the same lens, and I am a stickler for details in my large prints and this lens is one of the sharpest lens that I have used. You definitely got a bad copy..
Michael

[clackers]

I was trying to calibrate the focus of my lens and nothing was sharp with a 2 second delay. My hand held was way better. I put it on live view fully magnified with focus peaking and it takes 9-10 seconds before all the movement stops.
Greg

The phenomenon is 'shutter shock'.

[Chapel]

The phenomenon is 'shutter shock'.

The movement is before the shutter goes after I press the release. It's better when I tether since I release the shutter through Lightroom.
Greg

[texandrews]

Thank you--that's reassuring.

I have it boxed up to send back to Adorama. Adorama is a good company--not a bit of concern about accepting the return.

Rick "the search continues" Denney

Both Adorama and B&H are good companies....but I am very underwhelmed by their used gear and their ratings. KEH, Roberts/UsedPhotoPro, and Lens Authority are far better choices. I have never had an issue with KEH---in fact, I have bought "bargain" from them and thought it "very good", and "excellent" as pristine, and my 25 came from Lens Authority, which is the sales outlet for Lens Rentals, where Roger Cicala, maybe the preeminent lens tester in the U.S., works. FYI, Roberts is great to sell to.

[clackers]

The movement is before the shutter goes after I press the release. It's better when I tether since I release the shutter through Lightroom.
Greg

?

Even in Mirror Up, to see in Live View the shutter's open. So when you hit the shutter must first close, then open again for whatever shutter duration you chose.

The K-1 has an electronic shutter option, BTW, for this very reason.