[Arbalist]

I guess this has been mentioned before but how many Megapixels on an APS-C size sensor before current lenses become the limiting factor?

[boriscleto]

Current lenses will never be the limiting factor. The limiting factor in how many pixels you can squeeze onto a sensor is going to be physics.

[Arbalist]

OK, I found an answer somewhere else "The resolution of Nikon's latest pro lenses (14-24, 16-35, 70-200) is designed to support at least 60 MP on FX and 24 on DX"

[dms]

I believe that is too simplistic--in principle there is always an effect of both.

[JinDesu]

OK, I found an answer somewhere else "The resolution of Nikon's latest pro lenses (14-24, 16-35, 70-200) is designed to support at least 60 MP on FX and 24 on DX"

If that's true, then the sample pics of the D800 should be better than what they are.

[Laurentiu Cristofor]

I expect diffraction to be a bigger issue than lens resolution.

[falconeye]

To answer this question, one has to understand MTF first. There is no hard barrier except diffraction. And diffraction limits are far beyond 100 MP.

The D800 samples are bad because all early samples from any camera are bad. Have always been. This is because marketing departments ask big names to do sample shots and they typically lack the technical depth to understand what it means to exploit the added resolving power.

Actually, calibration, shake and AF accuracy are a much more limiting factor for effective resolution than lens blur. This means that both APSC and FF have their sweet spot. For APSC, it is probably closer to 12MP than 24MP, but there is no hard limit.

[DAZ]

What I have seen as I went over 10MP is technique and attention to detail become paramount. To use the 16MP of the K-5 I have to be on my “A” game and that is hard to do all the time. Most of the time I don't use more then 12MP. Going to 24MP just seems like a waste as to get there I would almost always have to use a tripod or fast shutter speed and I still don't really use 16MP most of the time. I moved to the K-5 for the DR not the MPs. As for the DR of the K-5 I can almost always get what I want from it's DR so even more is not that much of a draw for me to get a new camera.


There are those that do need more MPs or DR but most probably just want it not need it.


DAZ

[Laurentiu Cristofor]

And diffraction limits are far beyond 100 MP.

I am confused by this statement. On a 100MP APS-C sensor, diffraction would take effect from around f/4 (I used an online diffraction calculator). So how is diffraction not an issue up to 100MP? We already get diffraction above f/8 with 15MP sensors.

[falconeye]

I am confused by this statement. On a 100MP APS-C sensor, diffraction would take effect from around f/4 (I used an online diffraction calculator). So how is diffraction not an issue up to 100MP? We already get diffraction above f/8 with 15MP sensors.

I talked about diffraction limits, not diffraction effects.

The full treatment of the issue is complicated. Because it must consider lens blur due to aberration and lens blur due to diffraction.

For the very best lenses along the optical axis, the resolution limit (~MTF10) is about 150 lp/mm times F-stop (according to my own research) and the latter is about 1500 lp/mm divided by F-stop (Rayleigh criterion). The combined resolution maxes at F-stop 3.2 sqrt(1500/150) with a combined resolution of ~333 lp/mm.

As you can see, the diffraction limit for FF is around 7.8 billion pixels (3.3 billion for APSC) and "far beyond 100 MP" was a mild way to say just this. Moreover, the very best lenses can resolve a 380 million pixel sensor in the center (160 APSC). I have actually seen this with my own eyes, using a special high resolution B&W film with a DA70 and inspecting the result under a microscope. But don't be mistaken ... there is so much blur from other sources you will almost never touch into this high resolution regime with your photography, even if lens and sensor would allow so.

Eventually, there are only very few lenses with so little aberrations. Measurements in the photozone style can only indirectly identify them, by measuring the F-stop of highest resolution. These very few lenses have it at f/2.8 rather than f/4 (f/5.6 is a more typical point of max. resolution). One such lens is the Nikkor AF-S 200mm f/2 G ED VR (FX) photozone test which resolves best at f/2.8 with 4076 LW/PH in the center on a 4000 LW sensor. This is spectacular as all other such lenses are shorter than 50mm. But this 200/2 is seven times the price of a DA*200/2.8 ...

But the Nikkor 200/2G + Pentax Q may be the most powerful combination wrt resolution reach below astronomical scopes ... because it even outresolves pixels as small as the ones in the Q.


...

Below this point, the answer will depend:
- Which lens?
- Which F-stop?
- How far off center?
- How exact focus?
- How much shake?
- What ISO?

If you always want full sensor resolution, I'd say the initial *istD (6MP) was already a bit over the top

[Laurentiu Cristofor]

I talked about diffraction limits, not diffraction effects.

Let's start slower - what is the difference? The diffraction calculator I used also uses the term "diffraction limited" in conjunction with the choice of aperture. So what do you mean exactly by "diffraction limits"? You seem to mean something different than what I found used elsewhere.

[falconeye]

The diffraction calculator I used also uses the term "diffraction limited" in conjunction with the choice of aperture.

This diffraction calculator is a bit ill-defined because it uses a corresponding circle of confusion which normally doesn't exist in this context. Maybe, it is in this article. I use the well-defined Rayleigh criterion which the article mentions too. And the formula is simple enough, no need for an online calculator (line pair separation > 1.22 * 0.55µm * FStop). Online calculators don't help to understand.

Moreover, the diffraction limit is typically defined for N=1 for air/glass surfaces (an f/1.0 lens, line pair separation > 0.67µm, pixel > 0.34µm). It may be that no diffraction-limited lenses at f/1.0 exist, but that doesn't alter the limit I discussed it in my post above. Microscopes fully exploit the limit.

[Laurentiu Cristofor]

This diffraction calculator is a bit ill-defined because it uses a corresponding circle of confusion which normally doesn't exist in this context. Maybe, it is in this article. I use the well-defined Rayleigh criterion which the article mentions too. And the formula is simple enough, no need for an online calculator (line pair separation > 1.22 * 0.55µm * FStop). Online calculators don't help to understand.

Moreover, the diffraction limit is typically defined for N=1 for air/glass surfaces (an f/1.0 lens, line pair separation > 0.67µm, pixel > 0.34µm). It may be that no diffraction-limited lenses at f/1.0 exist, but that doesn't alter the limit I discussed it in my post above. Microscopes fully exploit the limit.

Can we just start from your definition of "diffraction limit" - what is it?

[falconeye]

Can we just start from your definition of "diffraction limit" - what is it?

It is not "my" definition and it is all said above, so I can only repeat: It is the Raleigh criterion with an as wide as possible a lens which is typically defined to mean f/1.0. The Raleigh criterion says that two points separated by the Airy disk radius can be (barely) resolved. The Airy disk radius for green light at f/1 is 0.67μm. So, pixels must be larger than 0.33μm. That's the diffraction limit. It's a constant if expressed as a size.

[Laurentiu Cristofor]

It is not "my" definition and it is all said above, so I can only repeat: It is the Raleigh criterion with an as wide as possible a lens which is typically defined to mean f/1.0. The Raleigh criterion says that two points separated by the Airy disk radius can be (barely) resolved. The Airy disk radius for green light at f/1 is 0.67μm. So, pixels must be larger than 0.33μm. That's the diffraction limit. It's a constant if expressed as a size.

By "your definition", I just meant the one you are using. I expect most photographers think of diffraction limit in relation to the maximum f-stop they can use and few have heard of the Raleigh criterion. So it's helpful to define concepts that are not common knowledge. Thanks for doing that!