[jsherman999]

24Mp is better for landscape photography. 36mp is better if you shoot mostly with your lens wide open (assuming that your lens is also sharp wide open). The cropping potential of a 36mp sensor is immaterial when you stop down to f11 or lower because of diffraction. You could just blow up your 24mp image in photoshop to a 36mp equivalent and then crop. It's the "same".

The point on the curve where diffraction starts to affect the image is exactly the same with both a 24MP or 36MP sensor - it's lens-dependant. You never can see more resolution with 24MP, and with most apertures - even when you start to get diffraction-limited - you will still see more detail with more MP.

All reducing pixel size (increasing MP) does here is increase the resolution available, it doesn't change at all the point where diffraction takes over from aberrations as the main cause of blur. Consider this graph, made from DxO measurement data:



That's the same lens, on two different sensors (D3X 24MP vs. D3 12MP) - note how the top of the curve is at the same point, and note how you have more resolution throughout the aperture range, even at f/16, with more MP.

In other words, unless you're shooting landscapes at smaller than f/16 all the time, you will always be able to see at least a little advantage to 36 vs. 24MP MP, depending of course on the lens too.

---

Last edited by jsherman999; 02-23-2014 at 04:57 PM.

[dtmateojr]

Yes, diffraction kicks in at the same aperture. Therefore at f11 for example, your 24mp resolves the same details as 36mp. Your 36mp though is just a waste of space, processing and pixel level quality is worse by virtue of the smaller dot pitch. The 36mp sensor has nothing to gain but much to lose at f11 vs any other sensor in production

[osv]

Great. I guess Zeiss wasted their R&D money on the Otus when its apparently unnecessary. They will resolve more, but my question is how justifiable is it. The increase in sharpness may be marginal at best (if even). Besides a lot of Pentax users want full frame so they can use their old "legacy" glass. Those lenses certainly would not resolve the sensor. To get it to be "sharp" might have to down sample anyways.

today i put a $20 junk plastic ix-nikkor zoom lens on my $2300 a7r... ix-nikkor lenses only came on nikon pronea aps-c cameras, so that lens was never designed for a full-frame sensor.

i shot it at 24mm, f8 or slightly bigger, and it was pretty sharp across the center third of the frame, until the field curvature losses started showing up.

in fact, it was comparably sharp to an olympus 24mm f2.8 lens, that i just paid about $240 for at KEH... which might actually have more field curvature than the nikon zooms do :-/

that level of incredible sharpness indicates to me that sensors are not outresolving lenses, not even close.

read more about this at:

LensRentals.com - Sony A7R: A Rising Tide Lifts All the Boats?

[jsherman999]

Yes, diffraction kicks in at the same aperture. Therefore at f11 for example, your 24mp resolves the same details as 36mp. Your 36mp though is just a waste of space....

No. For one thing, it's lens + sensor dependent, so "f/11" can't be a general rule that applies for every lens. I'd be willing to bet that if you took any lens you have, shot it at your f/11 on 24 & 36MP, then displayed/printed at the same size, you'd be able to see a little more detail with the 36MP image - even though F/11 is probably about a stop or two past the point where diffraction takes over as the cause of blur (which depends on the lens and is different for each lens.) Even if you do find a lens that has it's resolution curve drop off so quickly after f/11 that 36MP and 24MP are essentially the same, 36MP would only be a 'waste' with that particular (probably bad) lens, and then only beyond f/11. For most other lens/aperture combos, you're reaping the benefits of more MP.

Ask the pro landscape shooters on the Nikon fora who swear by the D800, if you don't believe me

Here - below is linked a very enlightening thread on MP and diffraction - read the whole thing, but pay close attention to the posts by 'bobn2' (creator of 'sensorgen' site and a very good engineer with training in this field,) and "Great Bustard" (Joseph James, who coined the term 'equivalence' as it applies to sensors/systems.)

Link: Nikon D800/E Diffraction Limits


.

---

Last edited by jsherman999; 02-23-2014 at 05:33 PM.

[osv]

Yes, diffraction kicks in at the same aperture.

there are some unsubstantiated and/or illogical claims in that link: "To fully utilise a 36Mp full frame sensor, one has to shoot at f5.6 or wider. At f5.6, a perfect lens is capable of fully utilising a 60Mp full frame sensor. At f8, around 30Mp. We therefore say, that the Nikon D800 36Mp sensor is diffraction limited at f8."

let me correct that statement... "to fully utilize *any* sensor, one has to shoot at f5.6 or wider" ...the only test results that i've seen that show higher resolution at f4 are due to lens design, not sensor size... that is proved by the fact that another lens will have a higher resolution at a f5.6 or smaller, on the same sensor.

fa35 f2 showing more resolution at f4 than at f5.6... it's not just all about the sensor:
http://www.photozone.de/pentax/123-pentax-smc-fa-35mm-f2-al-photozone-review...report?start=1

f5.6, f8, etc., are not physical sizes, like inches, and their actual measurement is dependent on focal length.

other issues there revolve around pixel size and spacing on the chip(aka pixel density), and whether or not the lens itself is diffraction limited.

look at the lensrental link i posted, 36mp a7r sensor vs. canon 21.1mp sensor, exact same lens on both, and both are full frame sensors:

canon @f4: 1045 lines
a7r @f4: 1400 lines

clearly, more pixels can equal higher resolution.

[jsherman999]

read more about this at:

LensRentals.com - Sony A7R: A Rising Tide Lifts All the Boats?

This also tells us that one of the best landscape combos available to anyone right now might be the A7R + Zeiss 35mm f/2.8 FE. Pretty impressive.

[oxidized]

This also tells us that one of the best landscape combos available to anyone right now might be the A7R + Zeiss 35mm f/2.8 FE. Pretty impressive.

not so sure about that

The problem with Sony's Zeiss FE Sonnar T* FE 35mm f/2.8 ZA | Leica BOSS

[osv]

This also tells us that one of the best landscape combos available to anyone right now might be the A7R + Zeiss 35mm f/2.8 FE. Pretty impressive.

resolution-wise, and distortion-wise, yes, it's a stunning piece of glass, i wish that i owned one! as oxidized pointed out, tho, there is some competition.

what nearly everyone failed to notice with the fe 35 was how dark it is... here is a thread with the photozone review, that finally broke the info, with imatest measurements, all camera correction turned off... with all of the comments in the blogosphere about these cameras, you'd think that someone would have done that testing before?

more relevant to pentax people... i made a couple of posts in that thread, wrt the pentax fa35 f2, which is pretty bright in the corners on the a7r... it's much, much brighter than the fe 35 is... i also put an fa35 a7r sample landscape shot at full resolution in that thread; can't do that on this forum very easily, or i'd post more examples here.

Photozone.de Fe 35 Review is up: Sony Alpha/NEX E-mount Talk Forum: Digital Photography Review

i also put the cheap pentax 35mm 3.5 tak on the a7r, and it looks very similar to the fa35, at f8 and such... so hang on to all of that old full-frame pentax glass!

[dtmateojr]

there are some unsubstantiated and/or illogical claims in that link: "To fully utilise a 36Mp full frame sensor, one has to shoot at f5.6 or wider. At f5.6, a perfect lens is capable of fully utilising a 60Mp full frame sensor. At f8, around 30Mp. We therefore say, that the Nikon D800 36Mp sensor is diffraction limited at f8."

let me correct that statement... "to fully utilize *any* sensor, one has to shoot at f5.6 or wider" ...the only test results that i've seen that show higher resolution at f4 are due to lens design, not sensor size... that is proved by the fact that another lens will have a higher resolution at a f5.6 or smaller, on the same sensor.

fa35 f2 showing more resolution at f4 than at f5.6... it's not just all about the sensor:
http://www.photozone.de/pentax/123-pentax-smc-fa-35mm-f2-al-photozone-review...report?start=1

f5.6, f8, etc., are not physical sizes, like inches, and their actual measurement is dependent on focal length.

other issues there revolve around pixel size and spacing on the chip(aka pixel density), and whether or not the lens itself is diffraction limited.

look at the lensrental link i posted, 36mp a7r sensor vs. canon 21.1mp sensor, exact same lens on both, and both are full frame sensors:

canon @f4: 1045 lines
a7r @f4: 1400 lines

clearly, more pixels can equal higher resolution.

That Canon vs Sony test makes sense because NONE of them are diffraction limited at that aperture. Stop down to f11 and they will resolve the same number of lines even with a perfect lens

An f-stop is an f-stop. f5.6 for aps-c is the same for ff and has the same diffraction spread.

---------- Post added 02-23-14 at 06:43 PM ----------

No. For one thing, it's lens + sensor dependent, so "f/11" can't be a general rule that applies for every lens. I'd be willing to bet that if you took any lens you have, shot it at your f/11 on 24 & 36MP, then displayed/printed at the same size, you'd be able to see a little more detail with the 36MP image - even though F/11 is probably about a stop or two past the point where diffraction takes over as the cause of blur (which depends on the lens and is different for each lens.) Even if you do find a lens that has it's resolution curve drop off so quickly after f/11 that 36MP and 24MP are essentially the same, 36MP would only be a 'waste' with that particular (probably bad) lens, and then only beyond f/11. For most other lens/aperture combos, you're reaping the benefits of more MP.

Ask the pro landscape shooters on the Nikon fora who swear by the D800, if you don't believe me

Here - below is linked a very enlightening thread on MP and diffraction - read the whole thing, but pay close attention to the posts by 'bobn2' (creator of 'sensorgen' site and a very good engineer with training in this field,) and "Great Bustard" (Joseph James, who coined the term 'equivalence' as it applies to sensors/systems.)

Link: Nikon D800/E Diffraction Limits


.

Assuming that all lenses are perfect, f11 is f11. Same diffraction. Same resolution for the same sensor size.

---------- Post added 02-23-14 at 06:55 PM ----------

The whole point of my post is that, say at f11, a D4 with 16mp will resolve exactly the same as a D800 with 36mp with the same lens. The D800 will have a bigger image and that's it. You can blow up the D4 image in PS to arrive at the same 36mp without loss of detail. The advantage of the D4 is that you are not wasting space, your canera can shoot faster, your PC can process the images faster AND on the pixel level, the D4 has an unfair advantage over the D800 because of the larger sensels.

[ElJamoquio]

The whole point of my post is that, say at f11, a D4 with 16mp will resolve exactly the same as a D800 with 36mp with the same lens

It won't.

The 'diffraction limit' is not a hard limit. It's a impediment to a sharp image, sure, like camera shake is, but it's also more-or-less a gaussian distribution. A 36MP sensor will still have more information than a 16 MP sensor, at F/11 by a margin prob. larger than the sharpness of the kit lens compared to the 31mm limited.

The only downsides to the larger files are expense, storage, and processing speed. From a technical standpoint there is basically no 'real world' situation where a 16 MP sensor will have the same info as a 36MP sensor.

[dtmateojr]

It won't.

The 'diffraction limit' is not a hard limit. It's a impediment to a sharp image, sure, like camera shake is, but it's also more-or-less a gaussian distribution. A 36MP sensor will still have more information than a 16 MP sensor, at F/11 by a margin prob. larger than the sharpness of the kit lens compared to the 31mm limited.

The only downsides to the larger files are expense, storage, and processing speed. From a technical standpoint there is basically no 'real world' situation where a 16 MP sensor will have the same info as a 36MP sensor.

Read/re-read my blog post again so you will understand the very basic physics concept behind diffraction. It IS a hard physical limit.

Edit:
Diffraction changes the input data. The sensor remains the same.

[osv]

That Canon vs Sony test makes sense because NONE of them are diffraction limited at that aperture. Stop down to f11 and they will resolve the same number of lines even with a perfect lens

umm... o.k., show us the exact math that proves what the specific number of possible resolvable lines at f11 really are.

meanwhile, here are real-world test results from three full-frame sensors at f11, using the exact same lens: 24mp vs. 12mp vs. 36mp:

d3x: ~850 lines
d700: ~680 lines
d800: ~940 lines

"Even on the D800 resolution is as high, or higher, at f/16 than it was at f/2.8. At f/11 the resolution is as good, or better, than at f/4. And at both f/11 and f/16 resolution is clearly higher than it was wide open. Perhaps the diffraction monster’s teeth aren’t as long and wicked as I thought."
LensRentals.com - Overcoming My f / Entekaphobia

[pinholecam]

This also tells us that one of the best landscape combos available to anyone right now might be the A7R + Zeiss 35mm f/2.8 FE. Pretty impressive.

The lens is good to a point but its also flawed for some types of landscape/cityscape work (and not worth what it is imo)
Carl Zeiss Sonnar T* FE 35mm f/2.8 ZA (Sony SEL35F28Z) - Review / Test Report - Analysis

It has too much vignetting that has to be corrected via in-camera firmware or post. (-2.6ev @f2.8 and 0.96 @f8)
Correcting for that leaves the corners 'brittle' for further post processing.


I'd rather just op for one of the other 35mm out there which is not at all expensive and often faster (unless its a Zeiss or Leica).
Unless one needs AF.

Theres also that 'colored rings' problem, found at Leica Boss as linked by oxidized, which I think is a variation of the same issue.

[ElJamoquio]

Read/re-read my blog post again so you will understand the very basic physics concept behind diffraction. It IS a hard physical limit.

Edit:
Diffraction changes the input data. The sensor remains the same.

It is a physical limit, but it is not a 'hard' limit. I understand the concepts behind diffraction; do you understand a gaussian distribution?

[dtmateojr]

It is a physical limit, but it is not a 'hard' limit. I understand the concepts behind diffraction; do you understand a gaussian distribution?

I'm a physics major. I eat gaussian curves. I major in optics and dsp. I know what I'm talking about I don't think you understand diffraction.