[Rondec]

Okay so 2160 is 33% more than 1440. So are crop camera images 33% less sharp than full frame cameras?

https://www.youtube.com/watch?v=Z4BDi3tph4M

Although this might be considered an unfair comparison, because a. the K-3 has no AA filter b. K-3 uses pixel shift c. the Tamron lens Ed uses is not as sharp as a DA* lens. And its not like Ed buys crappy gear.

Back in the day crop had very heavy AA filters, now thery're real light or if any. Then why does it appear that crop is just as sharp as full frame, and in many cases sharper ? I mean, they're not putting on AA filters so blurry that they're taking away 33% sharpness from a FF image.

Crop versus full frame may or may not make a difference in the final print or image, depending on the size of the print, or the monitor it is viewed on. But certainly, it would make sense that if you print a D810 image and a K5 IIs image at 36 inches on the long side, the D810 image will have more detail visible and will appear sharper in comparison. On the other hand, compare 4 by 6 prints and you couldn't see a difference.

The pixel shift is a specific application which cameras with image stabilization can do to add detail and dynamic range to images. It is not usable in many situations due to movement within the image, but when usable, it certainly improves K3 II images to being better than D610/6D images, but still not quite as good as D810 images.

[Glorfindelrb]

Although this might be considered an unfair comparison

It is unfair and naive. Ofc stacking 4 shots with 1.5x crop (372mm² x4 = 1488mm²) vs 1 shot of a ff (864mm²) sensor should give an advantage to the stacked image (is the conditions are ok with stacking). A pano of 4 aps-c images vs a single FF shot should also.

Pixel shift is just an in camera stacking with the optimisation of countering the effect of the bayer filter to get the best possible result with only 4 shots and without moving the camera.

[Nicolas06]

It is unfair and naive. Ofc stacking 4 shots with 1.5x crop (372mm² x4 = 1488mm²) vs 1 shot of a ff (864mm²) sensor should give an advantage to the stacked image (is the conditions are ok with stacking). A pano of 4 aps-c images vs a single FF shot should also.

Pixel shift is just an in camera stacking with the optimisation of countering the effect of the bayer filter to get the best possible result with only 4 shots and without moving the camera.

Reading that idea it is also unfair to compare an APSC with a 35mm sensor as they don't share the same surface area. After all a larger sensor is just an in camera optimization that allow to get less noise and less deph of field at same apperture...

For me it is unfair because well, it is quite likely the Pentax FF will have the same feature anyway and it apply to a very restrictive use case: perfectly still scene, subject on a tripod.

To evaluate the interrest of a pixel shift, it would make more sense to compare pixel shift on K3-II and no pixel shift on K3, K3-II or D7200 with 4 pictures stacked in photoshop or something like that.

As if a K3, D7100, D7200 or K3-II achieve similar sharpness as a D600/D610/D750, the response is basically yes, very similar sharpness and better sharpness than FF with less pixels. The high iso avantage is not so present on a landscape because you need to be closed down, and to be even more closed down on an FF than on an APSC. There the dynamic range of course but when you need lot of dynamic range, it is not the small difference between the 2 sensors that will make the difference but more if you decide for an HDR shoot or not where you can add 4-8EV of dynamic range.

D810 is sharper when the lens follow because it has more MP. Maybe we will get a 36MP APSC one day that will do just as well. If it coupled with a few low iso setting (like iso50 or better iso25) then it should perform quite competitively against a D810.

[fs999]

Does Sony have paied DXOMark to remove the 645Z test ?
I can't find it ! If you click on OP's link, the highest ranked camera is Sony A7R II...

[Zygonyx]

It hasn't been published yet, you only can find the main results as a comparative sheet in Sony RX1-II test.

[fs999]

It hasn't been published yet, you only can find the main results as a comparative sheet in Sony RX1-II test.

No, only in the text above the table :

In terms of rankings, only the Pentax 645Z with its physically larger 43.8 x 32.8mm CMOS sensor (nearly 1.7x the surface area of a full-frame 35mm sensor) is ahead by just over a half-stop in noise levels.

The first camera in the table is Sony A7R II...

[Deimos]

They removed it....

[fs999]

It would be interesting to know why...

[davidsladek]

something tells me that they removed it so the new Phase One could replace it... I think they are clearly confusing their vested interest with a credibility of testing website. And, unfortunately, a lot of people and other review sites take that credibility for granted.

[Zygonyx]

Incredible, yes.

[zoolander]

It is unfair and naive. Ofc stacking 4 shots with 1.5x crop (372mm² x4 = 1488mm²) vs 1 shot of a ff (864mm²) sensor should give an advantage to the stacked image (is the conditions are ok with stacking). A pano of 4 aps-c images vs a single FF shot should also.

Pixel shift is just an in camera stacking with the optimisation of countering the effect of the bayer filter to get the best possible result with only 4 shots and without moving the camera.

No its not unfair or naive to compare outcomes between sensor sizes, people do this on a regular basis.

The K-3ii came out, and from memory the improvement in sharpness over the K-3 was 12% or 14%. The 6D has fewer pixels at 21pm, so I think its 11% mega pixels.

So from 33% we subtract 11% (for fewer pixels) = 22%

The K-3ii has no AA filter which is an 8% improvement.

Plus 14% for pixel shift = 22%

That Canon Lens Ed was using is better then the Tamron for the K=3ii.

This above is just a rough calculation. So as far as I'm concerned they should be similar in terms of detail, but why does Ed from Photouniverse observe something greater, greater that he got rid of the Canon gear. He also had a Nikon D810 and ditched that too because of sharpness issues with his older lenses.

What I'm saying here is you cannot unequivocally declare full frame sharper. There are unique factors that come into to play. But yeah okay a 24mp FF versus 24mp apsc sensor, the FF will resolve more.

This whole argument is just like cars. What is better, a V8, a V6 or a 4 cylinder turbo. Yeah a V8 has lots of grunt, but the cars usually weigh a lot and chew gasoline and can achieve a high top speed. 6's are good if they're turbo charged, but lose in top speed to V8's. 4 cylinder turbo charged all wheel drive cars built to meet world rally homologation rules, are very quick and light with lots of grip, and also chew gasoline, but lose in top speed compared to V8's. People will say, oh you can't compare 8 cylinder cars to 6's and 4's and turbo charged cars .........

Either way, they'll all go very quick and do what they're designed to do. Everybody has their preference, much like cameras. The K-3ii is turbo charged with pixel shift, I know its not perfect, but its useful.

As far as I understand pixel shift, it is not a stack of 4 images.

[Edgar_in_Indy]

Well, it's now been just over two weeks since "Sophie" from DXoMark told me that the 645Z results "should be published within the next couple of weeks."

[Fogel70]

Okay so 2160 is 33% more than 1440. So are crop camera images 33% less sharp than full frame cameras?

No, 2160 is 50% more than 1440.
But 1440 is 33% less than 2160.

[Glorfindelrb]

They removed it....

Ok, DxO cannot invoke any excuses now. I feel sorry for Zygonyx who got its stuff borrowed for weeks for this comedy.

[Edgar_in_Indy]

No, 2160 is 50% more than 1440.
But 1440 is 33% less than 2160.

After so much speculation and conjecture in this thread, it's nice to have at least one thing that can be conclusively proved.