[falconeye]

Is it possible that the D7000 is less sensitive to oblique incident angles?...If the K-3's sensor does a better job of maintaining efficiency towards wide f-stops that would explain why it can pick up on aberrations...

D7100. I think we can assume that both lenses are identical, rear element included. The K- /F- mounts are made for f/1.2 lenses anyway. I now had a closer look at DxO curves (cf. below) and I believe it is no issue of oblique incident angles.

In fact, ALL tested Pentax lenses look really bad on DX0 Mark for some reason.

Can't say for other lenses, but here we have a direct 1:1 comparison with almost identical cameras and lenses. It is worth to examine this particular case, I believe.

I now had a closer look at DxO curves, esp. the "acutance profile for use case in %". I think the measure is the same as they coincide at F/5.6 center. To bring numbers into perspective, the Zeiss Otus (D7100 f/1.4) scores at (center, corners): 79.1% and 73.4%. Best value is 80.8% (f2.8 center), so 80% is the true ceiling as DxO weights with the human eye's contrast curve and doesn't really look at detail resolution.

The differences are most visible at F/1.4 (even in the center, 67.5% vs. 72.6%), towards the corners and for the (solid line) horizontal resolution (53.5% vs. 65.5%).

Currently, I believe that DxO fetched a particularly outstanding performer for Nikon from Sigma and that what we see, may indeed be sample variation.

---

Last edited by falconeye; 01-15-2014 at 05:51 AM.

[Class A]

I think we can assume that both lenses are identical, rear element included. The K- /F- mounts are made for f/1.2 lenses anyway. I now had a closer look at DxO curves (cf. below) and I believe it is no issue of oblique incident angles.

I also assume that the lenses will perform identically, independently of the mount.

What I meant was that the (Toshiba and Sony) sensors may have different sensitivities towards oblique incident levels. The micro lens designs are probably different and may account for the D7100 sensor being less sensitive to oblique incident levels. If so, it may be partially blind to the extra rays that become available near f/1.4.

Even within Sony, sensors perform differently with respect to oblique incident levels so I wouldn't be surprised if the Toshiba sensor showed a different characteristic.

[savoche]

A bit off-topic, but has it been confirmed that the K-3 uses a Toshiba sensor? Or more relevant, that it doesn't use the same sensor as the D7100?

[Rondec]

A bit off-topic, but has it been confirmed that the K-3 uses a Toshiba sensor? Or more relevant, that it doesn't use the same sensor as the D7100?

K3 uses a Sony sensor.

[jon404]

Looked at the DXO lens results this morning. Well! My DA 35mm f/2.4, 16-45mm zoom, and 55-300mm zoom all did very well. So, if I buy a K-3 in future, I'm all set. Problem is, for a DSLR, I like my K-30 well enough ... and, being retired now, can't honestly find one reason where a new K-3 would improve me as a photographer. But maybe next Christmas, when the next outbreak of the shopping virus goes around...

[normhead]

Looked at the DXO lens results this morning. Well! My DA 35mm f/2.4, 16-45mm zoom, and 55-300mm zoom all did very well. So, if I buy a K-3 in future, I'm all set. Problem is, for a DSLR, I like my K-30 well enough ... and, being retired now, can't honestly find one reason where a new K-3 would improve me as a photographer. But maybe next Christmas, when the next outbreak of the shopping virus goes around...

The K-3 is a cure for the "My K-5 might not be good enough" virus.

[savoche]

The K-3 is a cure for the "My K-5 might not be good enough" virus.

Ah, well, I've got the "my K-5 is too good to be replaced" virus. The K-3 is almost a cure for that too, but not quite. At least not yet

[normhead]

Ah, well, I've got the "my K-5 is too good to be replaced" virus. The K-3 is almost a cure for that too, but not quite. At least not yet

My wife still has the "my K-5 is too good to be replaced" virus, and she's seen what I do with my K-3.

[Rondec]

My wife still has the "my K-5 is too good to be replaced" virus, and she's seen what I do with my K-3.

They are both really good. The K3 has a little better resolution, of course, but otherwise the biggest differences are non-sensor in nature.

(K3 photo).




(K5 II photo)

[mattb123]

Wow, my favorite lens, the DA 15 was last! My favorite zooms (60-250, 18-135) weren't on the list.

Maybe I should just shoot and not read this stuff.

[Rondec]

Wow, my favorite lens, the DA 15 was last! My favorite zooms (60-250, 18-135) weren't on the list.

Maybe I should just shoot and not read this stuff.

The problem is that these sorts of evaluations are all about light transmission and sharpness, not about real life rendering. The DA 15 is slow, but it is a pretty mighty piece of glass when it comes to taking real life photos.

[dadipentak]

Maybe I should just shoot...

That seems to be working pretty well.

[Belnan]

According to DxO sharpness is measured over all apatures. In the case of pentax this can go anywhere from f/1.4 to f/22. Many nikon and sigmas only go to f/16. f/16 to f/22 is generally where sharpness will be at its lowest.

I suspect if they only scored for peak sharpness that pentax would look quite nice compared to the competition.

[falconeye]

According to DxO sharpness is measured over all apatures.

DxO scores are bad and confusing. But not as bad as some comments here suggest.

The DA15 scores last for a reason and I'll explain why. I personally love that lens, it is a unique offering in the Pentax line up noone else has.

DxO has 3 easily accessible scores:

- A: The DxO score which is a score of how much human visible information a camera/lens can capture in very low light. Here, a fast lens always wins if it is reasonably fast wide open. Here, the DA15 is rated low already because it is a slow lens.

- B: The P-MP score, which "only" averages some sharpness at the best aperture, weighting over the entire image field.

- C: Often overlooked but my preferred score: Mid light use case score, same as DxO score but for very good light.

While the DA15 scores only "14" at A, it scores a solid "30" at C, both scores taken at F/5.6 btw ...

The DA15 would score somewhat better if it wouldn't have extremely soft corners (I'd call them unusable) fully open. Clearly visible at the DxO field map. I am surprised actually that the mid light score didn't peak at F/8, an fstop I always try to use with that lens. The DA15 is great, but for it to be this small, it is only really useful at daylight. The DxO score reflects that.

So, in summary, the DxO scores do make sense but are incomprehensible to normal photographers.

My #1 complaint: the DxO scores violate the principle of divide and conquer: First, divide a problem into manageable subproblems, then solve each of it individually. We already know what a slow aperture means. We do not need no stinking DxO physicist to factor this well known fact into the question how sharp a lens it, making the resulting answer a mess.

[Cynog Ap Brychan]

We already know what a slow aperture means. We do not need no stinking DxO physicist to factor this well known fact into the question how sharp a lens it, making the resulting answer a mess.

That has always puzzled me about the DxO tests: we know that an f1.4 lens is going to transmit a lot more light than an f4 version. So why factor this in to the score in such a fashion? Surely a more meaningful measure would be a score based on how much the actual transmission varies from the stated maximum aperture? Or is it just me?