[MJSfoto1956]

Diffraction kicks in later with a larger sensor:

• FF sensor, 24mp, diffraction limit f/11.3
  
• 1.5X crop sensor, 24mp, diffraction limit f/7.4
  
• 1.5X crop sensor, 16mp, diffraction limit f/9.1

And it also "kicks in" earlier in the center than at the corners.
Basically add another stop to the corners before diffraction affects corner sharpness.
So taken as a whole, the diffraction limit "depends"...

YMMV

Michael

[npc]

Diffraction kicks in later with a larger sensor:

• FF sensor, 24mp, diffraction limit f/11.3
  
• 1.5X crop sensor, 24mp, diffraction limit f/7.4
  
• 1.5X crop sensor, 16mp, diffraction limit f/9.1

For the same size sensor with more pixels (finer pixel pitch), diffraction presents itself at a larger aperture. But that doesn't mean that suddenly the resolution of the 24mp sensor plummets at f7.4. It simply means that the sensor's resolution is being gradually compromised. The 24mp sensor will still out-resolve the 16mp sensor from f/7.4 to f/9.1. At apertures smaller than f/9.1, both sensors will resolve the same amount of detail.

Check this Diffraction Limited Aperture Calculator for verification:
Digital Camera Diffraction – Resolution, Color & Micro-Contrast

Check this Diffraction Limited Aperture Calculator for verification:
Digital Camera Diffraction – Resolution, Color & Micro-Contrast

Thanks! That's a great article (also part 1). The calculators let you easily check if you should expect any resolution gain going from 16mipx -> 24mpix if you know how the lens perfom on the lower resolution sensor at certain f-stop.
Another good one that I've stumbled upon recently is this one:
Do Sensors â??Outresolveâ?� Lenses? - Luminous Landscape
which might be little more "scientific" but basically describes the same thing (how diffraction,aberations and sensor resoliton are related to final image resolution)

---------- Post added 02-27-15 at 06:28 PM ----------

That is not quite how it works. I would refer you to the lens resolution figures on photozone.de where values are available for different resolution sensors. For the lower resolution sensors, the lenses perform more poorly, both center and edges. In other words, the sensor limits the lens proportionally across the frame. My statement is a little tongue-in-cheek, but it is based on the observation that as better films and sensors are developed, the measured resolutions of many of the better primes (and many of the not so good) tends to improve.

As noted in several comments above, the lens is what it is. A higher resolution sensor will always result in the same or better lens performance than with a lower resolution sensor.


Steve

OK, that's true, it only will work close to that on a "perfect" sensor - with no AA filter, demosaicing, etc. Both K-5 IIs and K-3 lack AA filter so for my naked eye results there will be very close to a "perfect" sensor.
When there's AA filter, its purpose is to blur the image - so no matter how sharp are the lens, the resolution is somewhat lower than the sensor resolution before the image is sampled by the photosites - to avoid moire. The sharper parts of the image are still sharper as result but the difference is minimized by the blurring. Also the less megapixels on the sensor - the stronger the AA filter needs to be, lowering the resolution even more. That can easily be seen on the photozone.de charts say for DA70 since it is tested on both 10 and 16 mpx. You can see on 10pix all the results are proportionally lower than the upper limit (which is the sensor resolution) comparted to the results on the 16 mpix sensor (both models have AA filter so they should never reach the limit). And at the optimal f/5.6 and f/8 on 10 mpix resolution is is almost even across the frame (so to me that is - the lens totally outresolves the sensor there) while on 16 mpix central resolution is noticably higher, almost hitting the maximum (so for me that reads: center certainly outresolves the sensor, k-5 AA filter is weak and rest of the frame may not really outresolve the sensor).

On a side note one some other lenses on photozone.de they have the central resolution hit or even exceed (i think the 43 ltd on k10d was one ) the maximum possible resolution which ... should not be possible unless some sharpening was applied to the image before analyzing (like in the raw conversion/demosaic process).

[Nicolas06]

Diffraction kicks in later with a larger sensor:

• FF sensor, 24mp, diffraction limit f/11.3
• 1.5X crop sensor, 24mp, diffraction limit f/7.4
• 1.5X crop sensor, 16mp, diffraction limit f/9.1

For the same size sensor with more pixels (finer pixel pitch), diffraction presents itself at a larger aperture. But that doesn't mean that suddenly the resolution of the 24mp sensor plummets at f7.4. It simply means that the sensor's resolution is being gradually compromised. The 24mp sensor will still out-resolve the 16mp sensor from f/7.4 to f/9.1. At apertures smaller than f/9.1, both sensors will resolve the same amount of detail.

Check this Diffraction Limited Aperture Calculator for verification:
Digital Camera Diffraction – Resolution, Color & Micro-Contrast

What funny for theses number is how precise because the effect would be of course gradually impact the picture, not like a hard hit. So it is the begining of diffraction taking effect, middle on the end? I would say this is the very begining limit because my K3 at f/9.1 outresolve without any problem my K5 at f/9.1 or even f/7.4 on most lenses anyway.

[Nicolas06]

I have a K-5 IIs and I am definately ok with the performance of my 15/21/35 limiteds on it. But the 21 is not the sharpest lens ever and the 15 is known for its corner issues.
The 35 is fine on K-5 IIs but recently i saw some pictures taken with it mounted on sony a5100 via adapter and I was somewhat dissapointed with the results (the sony has a 24mpix sensor but I'm not sure if it has AA filter or not) ...

So I was wondering - what is your experience with these lenses on K-3 ? Are they up to the 24 mpix sensor?

The more I take some time to rest and analyse the K3 overall better performance, the more I think that overall:

- the biggest gain comes from the AF that is much faster, much more accurate in low light, much more accurate with fast lenses and much more accurate using off center AF point. This alone can have really more visible impact on sharpness/rendering on most shallow deph of field photos than any added resolution. Put in simply, the AF is accurate with fast apperture even on lenses not known for their good AF support. That's a game changer.

- Compared to K5 (not K5-IIs) there very visible gain in resolution on most lenses except a few on the worst one (18-55 kit lense or 18-250...). The more I see review and reports everywhere, most of it comes from the low pass filter removal that is just a blur filter whose goal is to reduce sharpness/resolution. The 16 => 24MP part alone give a bit more, but no at much as the low pass filter removal.

- The 24 MP sensor is great at high iso, keeping more sharpness than the K5, in particular because starting 3200 isos, there a software blur filter applyed to K5/K5-II/K5-IIs/K30/K50 familly of sensors to improve the high isos numbers repported. This filtering is done on the RAW so you can't get ride of it. On the oppositve, because the pixel are bigger you'll get less noise to start with and this is start to show at very high iso 3200isos+ in shadows.


Overall the K3 is a better general purpose camera, even than K5-IIs. The less processed RAWs out of K3 are more logical but also mean you need to master more denoising software with K3 than K5 familly. For example I discovered that DxO prime algorithm improve much more K3 images than K5 and that it is needed in more occasion. But the better AF alone together with more possibility for crop and increased performances from most lenses make it a great buy. I spend much more time shooting sub iso800 anyway (iso 100 really) than iso1600+.

From the K5-IIs the one reason to upgrade is the autofocus, not the sensor. The sensor difference is much more visible between K5 and K3 because of the lack of low pass filter.

Here for example a 3200iso shoot from K3 and there no issue (click on it for bigger version if wanted):




Here are 2 100% crop of K3 images processed with DxO, but with all setting disabled. They would look far better with some post processing but this show you that this is really quite sharp already but it is because it use apperture where the lenses shine. But this give you an idea of how much margin you have for cropping in ideal conditions.

DA50-135, 80mm, f/5.6, 1/250, iso 100.



DA50, f/8, iso 100, 1/400