[Kunzite]

I do get where Alex is coming from here, though. He's not a test chart junkie... Nor am I... MTF numbers really don't bother me, I'm all about the rendering and, as you say, the final image. But in future, when someone levels criticism at Pentax for the 70-210/4 (or any other lens) being less sharp in tests than its Tamron cousin - which, so far as we know (though never 100% confirmed, as I recall), is optically identical save for the coatings - it would be nice to know why that's the case and have an explanation for it. Not a defence... just an explanation...

Well put.
And yes, this is partly the result of much criticism at various Pentax lenses, and partly toward my own professional quirks that makes me want an explanation when a test doesn't seems to match reality.
It has little to do with enjoying seeing the images; we're not only allowed to discuss about the most important things, right?

[Kunzite]

The first thing that has to be addressed in this thread is that higher MTF and other test scores are better. Using the term "worse" in the title complicates that enormously. You're starting from a position of bias, and then have to fight your way uphill.

I always forget people aren't as literal in their reading as me in my writing
Pentax lenses measured worse, that's strictly about numerical results in various tests.
I'm not saying, "Pentax lenses are worse", or even "less sharp". I know that isn't true. So how come even the best Pentax lenses - IMHO as good as everything else - don't measure as they should? Even against themselves?

[normhead]

Norm, I paid for the upgrade after pixel peeping until my eyes bled (and deciding there's no detail loss to talk about). I totally get it. I'm also a happy owner of the D FA 28-105, and on APS-C I used the DA Limiteds.

But this is about charts... not photography, but this thread and my question.

Maybe we should have a section for those who wish to discuss optical physics.
Not that I've seen much evidence many here could hold their own in such a discussion.
The math and formulas by themselves would be way more than most of us could handle.
50 years after I studied them I doubt I cold even follow a discussion of optical physics.
But as has been pointed out, it's the bodies. The lenses are capable of the same resolution on any system, but as noted by the Pentax jpeg engineer, Pentax focuses on the final outcome, and the final outcome has little (as in not more than 25%) to do with test charts. Despite the adamant refusal of many to even consider that possibility.

I'm having flashbacks.. they forced me to study optical physics in my photography course. If it's not your passion, it's absolutely brutal stuff.
My response at the time was "you don't have to know how to make a lens to know how to use a lens."
I'm sure someone could give you an answer to your optics question, and I'm also pretty certain you wouldn't understand it. I know I wouldn't.

I know just enough to know what I don't know.

But the answer to your question about why one lens tests better on another system than the same lens does on Pentax can only be answered by noting there are different ways of processing the raw information that may affect the test chart reproduction in different ways.

If you can leave it at that, that's all you need to know. If you want to better understand why that is, now you're into the getting into philosophies of the engineers in the design departments at the various companies.

It's not an optics question at all.
It's an industrial design question.
To answer precisely would require intimate knowledge of proprietary information.

The Pentax engineer in the link I posted describe the issue indirectly, if you know what you're looking for.

[BigMackCam]

The explanation would be Pentax values image quality as defined by humans rather than test charts interpreted by machines. Or is that a defence? What's the difference?

That's a comforting catch-all, but lots of folks won't buy that without further detail and/or evidence

Maybe "why" is the wrong question. Given the Pentax and Tamron versions of the 70-210/4 are supposed to be optically identical aside from coatings, perhaps the question should be "how" is it that the Tamron + Nikon combo test sharper than the Pentax combo... from more than one review source? It would be interesting (as well as useful in discussions) to definitively know the reason - whether it really is just sample variation and poor luck, or differences in one or more of the glass, coatings, camera sensor assembly, signal processing, whatever...

If the Pentax 70-210/4 wasn't a tweaked Tamron, this would be of little interest to anyone, save for other-brand shooters looking for bragging rights. Folks might just say, "Hey, Tamron has a lens that's even sharper... but since you can't get that in K-mount, the Pentax lens is an awesome choice anyhow". But it is a tweaked Tamron... and whilst there's nothing wrong with that (modern Tamrons are generally excellent... I love 'em), given the price premium for the Pentax version, some folks might be disappointed to learn they'll be paying more for less resolving capability.

Again, please understand, in practical terms MTFs mean little to me. I'm all about rendering, contrast, flare resistance, etc. So long as a lens is reasonably sharp (the level achievable by almost any decent modern lens, frankly), it's those other factors I'm more interested in. If indeed there really is a difference in performance between the Pentax and Tamron 70-210/4 lenses when used on their respective combos, I don't care much about that in practical terms. But I'd like to be able to knowledgeably address folks who use it as a criticism of the Pentax platform and glass...

---

Last edited by BigMackCam; 03-30-2020 at 01:31 PM.

[c.a.m]

Pentax lenses measured worse, that's strictly about numerical results in various tests.

I think one of the keys is to understand the methodology used by the various review sites. For example, in the ePhotozine review of the new Pentax 70-210, many of the test conditions are not obvious to me. For example, do they feed their Imatest application with JPEG images or RAW data? If JPEG, then camera settings such as sharpness will affect the measurement results.

Following from post 19 above by @Wheatfield. At the ePhotozine article that I referenced above, they state that "The images obtained of the lens chart are then processed through Imatest 3.7 Master, which gives resolution figures..." To me, this suggests that they use JPEG images, not RAW data, because the term 'RAW image' would not make any sense. If they in fact use RAW data, then they have made a serious mis-statement in calling it an image.

The Imatest folks provide a lengthy guide on performing tests. I note the following statement:

"You cannot measure a lens in isolation. It is a part of an imaging system that includes the camera’s image sensor, RAW converter (which may sharpen the image), and signal processing pipeline. Hence,

• Measurements are relative. It’s difficult to determine an absolute number for the lens alone. But you can accurately compare lenses on similar cameras.
• Camera and RAW converter settings, especially those that affect sharpening, noise reduction, and gamma, strongly affect the results. Record them, and be consistent. RAW files often produce the best results. The dcraw RAW conversion software used by Imatest employs no sharpening or noise reduction and has an ideal transfer curve with a (default) gamma of 0.45."

To understand the measurement differences among various lenses, it is crucial to understand the test methodology of the testers.

@Kunzite, would be able to point to other review sites that have 'measured' similar lenses from Pentax and other brands? I'd like to look into their methodologies.

- Craig

[house]

I think one of the keys is to understand the methodology used by the various review sites. For example, in the ePhotozine review of the new Pentax 70-210, many of the test conditions are not obvious to me. For example, do they feed their Imatest application with JPEG images or RAW data? If JPEG, then camera settings such as sharpness will affect the measurement results.

At the ePhotozine article that I referenced above, they state that "The images obtained of the lens chart are then processed through Imatest 3.7 Master, which gives resolution figures..." To me, this suggests that they use JPEG images, not RAW data, because the term 'RAW image' would not make any sense. If they in fact use RAW data, then they have made a serious mis-statement in calling it an image.

The Imatest folks provide a lengthy guide on performing tests. I note the following statement:

You cannot measure a lens in isolation. It is a part of an imaging system that includes the camera’s image sensor, RAW converter (which may sharpen the image), and signal processing pipeline. Hence,

• Measurements are relative. It’s difficult to determine an absolute number for the lens alone. But you can accurately compare lenses on similar cameras.
• Camera and RAW converter settings, especially those that affect sharpening, noise reduction, and gamma, strongly affect the results. Record them, and be consistent. RAW files often produce the best results. The dcraw RAW conversion software used by Imatest employs no sharpening or noise reduction and has an ideal transfer curve with a (default) gamma of 0.45.

To understand the measurement differences among various lenses, it is crucial to understand the test methodology of the testers.

@Kunzite, would be able to point to other review sites that have 'measured' similar lenses from Pentax and other brands? I'd like to look into their methodologies.

- Craig

Using in camera processed jpeg's will of course make comparing across systems useless. The difference in jpeg engine settings will make more difference than the lens. I don't know if ephotozine does but most do state somewhere that you can't compare numbers across systems. Lenstip uses standardised unsharpenend dcraw commands to generate the images from raw files. This should help eliminate jpeg engine differences but perhaps there are other issues preventing cross system comparisons.

[normhead]

That's a comforting catch-all, but lots of folks won't buy that without further detail and/or evidence

What, the paragraphs I posted weren't enough. What you want to know is proprietary, no one knows it, about any system.

ou cannot measure a lens in isolation. It is a part of an imaging system that includes the camera’s image sensor, RAW converter (which may sharpen the image), and signal processing pipeline. Hence,
Measurements are relative. It’s difficult to determine an absolute number for the lens alone. But you can accurately compare lenses on similar cameras.
Camera and RAW converter settings, especially those that affect sharpening, noise reduction, and gamma, strongly affect the results. Record them, and be consistent. RAW files often produce the best results. The dcraw RAW conversion software used by Imatest employs no sharpening or noise reduction and has an ideal transfer curve with a (default) gamma of 0.45.

Exactly. On optical limts Klaus posted "Don't try and compare the same lens on different systems and draw conclusions." Yet that's what this thread as posted is all about.
The reason being, the results don't mean what you think they mean.

There is ample evidence, but if you're looking for a slam dunk you can throw back as a smart retort, just say, "By the time yo've printed your final image, the things tested in the test charts are only a small part of the final result. The Pentax image may still be as good or better. " That might be the best you can do. Understanding exactly why requires way more base understanding of the issues. Face it, most of us here think noise is little artifacts in our image. We can't even understand the origins and reasons for noise. So how do we explain the various techniques that could be used as part of a general process applied to all photographs used to reduce noise effectively. Same with moire. The problem with the K-3 was cross talk in the sensor. How did Sony solve it? How did Nikon solve it? How did Pentax solve it? How effectively and efficiently did they solve it? Did they go for really efficient code that did 90% of the job or did they produce less efficient code that handled 60% of the issue? How much did they clean up, how much did they leave for PP software? It is in the answer to those questions that you understand how you get different results with the same lens and the same sensor on different systems.

But the big thing to understand is that the engineers from each company could probably have achieved the same results as another one of those companies. They chose not to, because of a different design philosophy. It isn't Pentax couldn't match the others, it's Pentax chose not to match the others. They had different priorities and those priorities may be just as crucial to the final image as those characteristics that produce test chart results. I keep saying it over and over, it doesn't seem to sink in. "For the way people take pictures, not for the test charts." Pentax has been very forward about this. So sad, people don't take them at their word.


In the end all that matters is "Do you like the images or don't you?"

---

Last edited by normhead; 03-30-2020 at 02:47 PM.

[BigMackCam]

What, the paragraphs I posted weren't enough. What you want to know is proprietary, no one knows it, about any system.

Exactly. On optical limts Klaus posted "Don't try and compare the same lens on different systems and draw conclusions." Yet that's what this thread as posted is all about.
The reason being, the results don't mean what you think they mean.

OK - fair enough.

But some people are going to keep comparing the original Tamron and Pentax-tweaked versions of the lens. CameraVille guy has already shown his findings, concluding the Pentax combo captures less detail on the same resolution sensor. I don't pay him much heed, but some folks do or will. Then they read the ephotozine review - generally a respected source for Pentax gear - and see that MTF figures, comments and ratings that suggest the Pentax lens performance doesn't quite match the Tamron, despite being excellent overall. Then they look at the price of the Pentax lens. Some people will question what's going on. If we answer them by saying "Pentax values image quality as defined by humans rather than test charts interpreted by machines", we'd better be able to show them evidence that the Pentax version renders differently in one or more favourable ways, or talk of "Kool-Aid" will surely follow

So maybe that's the better angle of attack, here. Rather than using a common body to test two optically-near-identical lenses of different brands and mounts, the testing should comprise side by side captures of the same scenes, one set taken with the K-1/K-1II plus Pentax lens, and the other with a Nikon (D800E? D810?) plus Tamron. We can forget all about MTF data, CA levels etc. We just show the beneficial rendering aspects of the Pentax combo over the other.

My concern with that, of course, is... what if there's no rendering difference, or if the differences are so tiny as to be immaterial?

For my own purposes, all that matters is "do I like the images?"... But there are folks shooting Pentax (including some who might be on the fence about staying) plus others who might be considering Pentax who will care about this stuff...

[mikesbike]

One thing that has been brought up is a key issue- cameras in the digital world are electronic beasts, with different sensors and electronic designs. Things were test-simpler in the film era, for more true results. I have noticed a trend among Pentax DSLR bodies being tested by various labs. That is a consistently-mentioned characteristic of the native sharpness before any is added, being more conservative (lower) with the Pentax than with other brands. That is, more post process being needed. Such is the reason OpticalLimits warns about comparing the same lens between systems.

Also notice, as to actual lines measured, upon looking at the same lens results within the Pentax system, take a look at the numbers of lines with the same lens on the K10D vs the 16mp K-5 and you will see a major difference.

But as to JPEGs out of the camera, Pentax has more in-camera adjustability for the setup- the Custom Image menus to set up sharpening, contrast, color palette, etc.

---------- Post added 03-30-20 at 03:33 PM ----------

See ephotozine's latest Pentax reviews, those for which they're giving numbers rather than "excellent/good/etc" - no lens will get close to 3500lw/ph.
Yet Nikon's best lenses can approach 4000lw/ph, on a D810. Even a zoom like the 70-300 will reach 3500 (@f11)...

Does the above shed light on this issue?

Now getting down to the ephotozine's reviews of the two lenses in question. Taking into consideration of being tested on two different camera systems, it is best to ignore actual lines-count differences, and just look at how well the lens does. Even with the different systems, a rational evaluation emerges. RE: the 50mm f/1.4 comparison.

I am seeing the Tokina Opera version reaches a very high standard of excellence at f/2.8 on the Nkon camera in the central area. The Pentax DFA* version is still excellent in the central area, though not reaching as high in lines-count on the Pentax camera. However, within its own system, edge performance from f/1.4 through f/4 is in the excellent range as is the central area! Not so with the Tokina Opera on the Nikon system. The most difficult to accomplish for lens design in a 50mm f/1.4 lens has always been to achieve sharpness at large apertures, including wide open. Amazingly, the Pentax DFA* accomplishes this beyond any expectation! Razor sharp at f/1.4!! Edge-to-edge sharpness!

The two lenses appear to be of similar performance at f/5.6, with the Tokina Opera a bit better at smaller apertures, which is easier to accomplish, so I would take that trade-off to get the near-impossible delivered by the Pentax DFA*! No question as to which one I would buy!

As to CAs, although the Tokina Opera on the Nikon camera shows better than the Pentax DFA*, what does that matter when both lenses are well within the excellent performance range as stated by the review?

As to the DFA 70-210mm f/4 lens, it is quite common for even top-notch zoom lenses to dip somewhat in quality level at the longest FL setting. Sometimes, even this is improved quite a bit by zooming just short of the longest FL- in this case zoom just to 200mm to see if performance comes up to excellent again- or at least very good.

I have no interest in either of these two lenses for my own usage, but these would be my observations in terms of making recommendations.

[Kunzite]

I think one of the keys is to understand the methodology used by the various review sites. For example, in the ePhotozine review of the new Pentax 70-210, many of the test conditions are not obvious to me. For example, do they feed their Imatest application with JPEG images or RAW data? If JPEG, then camera settings such as sharpness will affect the measurement results.

Following from post 19 above by @Wheatfield. At the ePhotozine article that I referenced above, they state that "The images obtained of the lens chart are then processed through Imatest 3.7 Master, which gives resolution figures..." To me, this suggests that they use JPEG images, not RAW data, because the term 'RAW image' would not make any sense. If they in fact use RAW data, then they have made a serious mis-statement in calling it an image.

[...]

@Kunzite, would be able to point to other review sites that have 'measured' similar lenses from Pentax and other brands? I'd like to look into their methodologies.

- Craig

Using JPEG would be a serious mistake. But I'll give them the benefit of the doubt, as their wording is vague.

"RAW image" is used a lot.

Hmm... I wanted to say Optical Limits, but the results (with lenses tested on the corresponding 16MP Nikon) are close, and the excellent D FA 50mm macro is doing well.
Klaus even warns against comparing the numbers for different systems.

[c.a.m]

But some people are going to keep comparing the original Tamron and Pentax-tweaked versions of the lens.

@BigMackCam, I continue to appreciate your thoughtful posts.

I guess I'd agree that some folks will read reviews, compare Pentax lenses to similar ones in other brands, and wonder why Pentax is 'not as good'. As discussed above in other posts, we really don't have a complete understanding of the testing conditions and methodologies of these reviews and tests, so I'd propose that the published quantitative test results should be viewed with caution.

In the context of this thread, I'm starting to think strongly that the Pentax lenses might 'measure worse' in certain instances because of the relatively soft default sharpening that is implemented in Pentax cameras for out-of-camera JPEG images. We've seen other reviews in which default settings have either benefited or penalized cameras -- it's not a level playing field.

But, apart from the technical puzzle, I wonder whether it's a bit of an academic exercise. After all, it's not as if there are choices to make in buying off-brand lenses for the K-mount, nor are other folks considering buying a Pentax lens to fit their non-Pentax camera (except for adapted lenses, of course). Pentax lenses only fit Pentax cameras natively, and Tamron, for example, doesn't make their modern 70-210 in a K mount.

- Craig

[c.a.m]

Using JPEG would be a serious mistake. But I'll give them the benefit of the doubt, as their wording is vague."RAW image" is used a lot.

I agree -- ideally, RAW files would be used as the input to the Imatest program(s).


On the other hand, in their January 2019 explanation of their methodology, ePhotozine states that "All lens corrections in camera are switched off and the highest resolution possible chosen. ... No in camera sharpening is allowed."

If they are using RAW data, why be concerned about in-camera lens corrections, "highest resolution" (which is typically selected for JPEG outputs only), and in-camera sharpening?


- Craig

[Sandy Hancock]

I can't use the Tamron 70-210 on my K-1 or KP.

I'm fairly fussy, and I'm happy with the handling and results from my Pentax version. I just can't warm to the DA* 60-250.

That's enough for me.

[mikesbike]

I still think it would be wise for Pentax to update the DA* 60-250mm f/4 with a DC motor and HD coatings. This would be at minimum cost to the company, since the lens is already designed, etc. and is highly regarded.

Ditto with the DA* 50-135mm f/2.8 as well.

[cport]

I would expect that Pentax 60-250/4 will be discontinued soon. The same happened to 12-24/4 ...