[Kunzite]

I hope you're all well and COV-19 free.

Reading various Pentax lens reviews, and particularly those where there is an other brand version - be it Tamron or Tokina - I've noticed how Pentax lenses tend to measure worse.
For example, the recent Pentax D FA 70-210's resolution falls quite badly at the longer end, and the CA is significantly worse than the Tamron tested on the D810. Or the Tokina Opera, having much less CA than the original Pentax. (That's according to ephotozine, one of the few remaining sites performing in-depth Pentax lens reviews)

The maximum resolution difference could be explained by MTF50 being very dependent on sharpening, although no sharpening should be applied. I cannot accept that basically the same 36MP sensor with no AA filter is somehow less sharp on a Pentax camera; it doesn't make sense.
And what about the CA?

Is it something reviewers consistently are getting wrong? (how could that be?) Is this Canikonyetc cherry picking the lenses sent for review, and Pentax just sending the same beat up sample they're sending everywhere? Is it Canikon's version of the Accelerator Unit applying sharpening and CA reduction? Is the Pentax badge cursed or something?

I could speculate and guess possible reasons; but I'd rather like to know - so I hope the knowledgeable (and hopefully bored enough ) people here would chime in.
Thanks!

[biz-engineer]

Is it something reviewers consistently are getting wrong?

Reviewers don't get it wrong. The Pentax K1 sensor + CFA is the one of the D800, not the one of the D810. And there is more noise reduction baked into the RAW of the K1, hence use of the accelerator chip to help the processing on the K1 II. Imagine comparison from imaging-resources show that the Pentax K1 and Pentax K1 II deliver less details than the D810, and even less details when compared to the D850 and others. Pentax K1 maxes out @ 3500 LP/PH, while other cameras achieve even more per pixel, Z7 deliver up to 5000 LP/PH. No wonder why photographic DR (Photo-to-photo) of the K1 is slightly better than others (old sensor, less noise, and less details). Pentax K1 / II images are smooth, even using an old sensor, that obviously comes at the cost of detail and processing time, we really need to fully understand what the accelerator is about to finally realize the why of magenta cast at base ISO, the noise-sharpness trade-offs involved and higher ISO settings.

[tibbitts]

I find it somewhat amusing that people are throwing out numbers like 5000 and 3500 lines, when I have so much trouble buying lenses for my 16mp K5, with AA filter no less, that at least appear at a glance to provide identical performance in all four sides/corners.

[cport]

Do you have some statistics confirming that Pentax lenses are measured wrong?

[Kunzite]

Reviewers don't get it wrong. The Pentax K1 sensor + CFA is the one of the D800, not the one of the D810. And there is more noise reduction baked into the RAW of the K1, hence use of the accelerator chip to help the processing on the K1 II. Imagine comparison from imaging-resources show that the Pentax K1 and Pentax K1 II deliver less details than the D810, and even less details when compared to the D850 and others. Pentax K1 maxes out @ 3500 LP/PH, while other cameras achieve even more per pixel, Z7 deliver up to 5000 LP/PH. No wonder why photographic DR (Photo-to-photo) of the K1 is slightly better than others (old sensor, less noise, and less details). Pentax K1 / II images are smooth, even using an old sensor, that obviously comes at the cost of detail and processing time, we really need to fully understand what the accelerator is about to finally realize the why of magenta cast at base ISO, the noise-sharpness trade-offs involved and higher ISO settings.

There is no NR "baked in" at base ISO in the K-1 II, and I sure hope reviewers aren't incompetent enough to test at high ISO... and there's no accelerator unit on the K-1, so find the cause elsewhere if you can.
Anti-Pentax rants aren't helpful. Just sayin'.

[Kunzite]

Do you have some statistics confirming that Pentax lenses are measured wrong?

My claim is that they're measuring worse, and I'd like to find out why.
Being measured wrong is just one possible cause; and it could be as simple as using a slightly different processing (which Photozone/Optical Limits actually does).

[normhead]

The issue here is the assumption that MTF resolution translates to image quality.

See this thread for relevant information.
Pentax 100 - Relentless Commitment to Image Quality - PentaxForums.com

Remember that even though the D7000 had better MTF numbers with the same lenses on a K-5, on sites like DxO that try and measure IQ, not MTF, the K-5 was rated slightly higher.
MTF is a small fraction of what constitutes IQ.

The suggestion of the Pentax engineer would be that Pentax looks at all factors that contribute to IQ, not just MTF charts.

[Kunzite]

Imatest doesn't even recommend using MTF50 for testing.

A good thread, thanks, but it doesn't explain why e.g. the Pentax D FA* 50mm has higher CA than the Tokina Opera. And I'm 100% sure the Image Accelerator doesn't add CA...

Of course, I knew about Pentax pursuing actual image quality rather than test results. But now I wonder, what if Pentax wasn't uniformly slightly "worse" than the competition, if there are other factors why the numbers weren't as high (or low)?

[biz-engineer]

There is no NR "baked in" at base ISO in the K-1 II, and I sure hope reviewers aren't incompetent enough to test at high ISO... and there's no accelerator unit on the K-1, so find the cause elsewhere if you can.Anti-Pentax rants aren't helpful. Just sayin'.

Same lenses tested with the D810 give more resolution than when tested on the D800E. The D810 sensor is not the same and presumably not the same microlens/CFA assembly. And the K1 sensor is the one from the D800E. Even the DFA*50 1.4 maxes out at 3000 LP/PH when tested on a Pentax K1. The common denominator is the sensor CFA and microlenses assembly... just look at a number of lens tests on D800E, D810, D850 and do some math, it becomes obvious that the K1 isn't the least resolving of the bunch.

---------- Post added 29-03-20 at 17:09 ----------

MTF is a small fraction of what constitutes IQ.The suggestion of the Pentax engineer would be that Pentax looks at all factors that contribute to IQ, not just MTF charts.

Knowing that most people don't print, I assume it was decided that the K1 got a bit of extra smoothing, also reduces the size of lossless compressed RAWs and JPEG files.

---------- Post added 29-03-20 at 17:17 ----------

A good thread, thanks, but it doesn't explain why e.g. the Pentax D FA* 50mm has higher CA than the Tokina Opera. And I'm 100% sure the Image Accelerator doesn't add CA...

Raw files are processed via camera image processor.

[Kunzite]

Are you saying Pentax' image processor is adding CA?

[normhead]

Of course, I knew about Pentax pursuing actual image quality rather than test results. But now I wonder, what if Pentax wasn't uniformly slightly "worse" than the competition, if there are other factors why the numbers weren't as high (or low)?

We know what we're losing with the high CA numbers, we don't know if there is something else we might be gaining. The last example of CA I posted the CA was only visible in one image of the 8 or so keepers I got that day.

CA image...


No visible CA





Using the same set up positioning etc. I see very different CA, just from changing the angle of sun to subject a few degrees or a slight rotation of the camera on the tripod.

You can obsess about CA, you don't have to. Most of the time I don't even bother correcting it.



The best conditions for producing CA would be high contrast good lighting. Or in other words , a test chart. I'm not sure how relevant that is to everyday shooting.

Even in high contrast situations.


Someone needs to talk to an optical engineer to see how they see it.

What we need to do here is ship the same lens to someone with a Sony and Pentax and Nikon adapters, and test the different incarnations on the same camera, then the Pentax lens on the Sony Camera and k-1. We have way too many variables to know what's going on. Are the sensors used by Pentax more prone to reproducing CA, is it some component of the Pentax implementation of the design? Was the same lens used for both of the tests, and could that lens be defective? We don't know anything.

But based on my experience with my own lenses, sometimes they produce CA , sometimes they don't. There are parameters I've never bothered to define. So, I'm completely unwilling to commit to any kind of explanation, beyond let's keep an eye on it until we get a more thorough understanding of CA, especially with regards to it's prevalence in the shooting conditions most of us shoot in every day.

[Kunzite]

You can obsess about CA, you don't have to. Most of the time I don't even bother correcting it.

Just to be clear, I'm not obsessing either about CA nor the lower resolution numbers. If I did, I'd have jumped ship long ago, to a brand measurably "better" (but which I'd like less to use). On the contrary, I'm happy with all my D FA lenses.
I'm only curious about the discrepancy... "same" lens, different results.

Someone needs to talk to an optical engineer to see how they see it.

Right. Do we have one here?

[normhead]

Just to be clear, I'm not obsessing either about CA nor the lower resolution numbers. If I did, I'd have jumped ship long ago, to a brand measurably "better" (but which I'd like less to use). On the contrary, I'm happy with all my D FA lenses.
I'm only curious about the discrepancy... "same" lens, different results.


Right. Do we have one here?

We have one, bdery, but I'm not sure he's skilled in DSLR lens design parameters.

I'm guessing a lot of this stuff is proprietary. We can look at the lens designs to see what they've done, but not accurately enough to know intimate production details, and we don't know what they mean. At least we know Pentax has done the testing they needed to do to fulfill the requirements of their design philosophy. Others with different design philosophies will probably have slightly different results. Better or worse could conceivably depend entirely on your shooting style, but how would we know?

[Kunzite]

But this can't be the design philosophy if the "same" lens gets different results.

[Wheatfield]

But this can't be the design philosophy if the "same" lens gets different results.

I wonder if the bodies used to test the same lens with different name plates aren't baking in different results.