Sensor size and MP vs lens limits?

[MarkMyWords]

Out of curiosity, what is the effective MP resolution of current high-end FF lenses (or any format actually)? I.e. can they fully resolve say a 100MP sensor?

Also, is there a sensor size sweet spot to minimise the impact of lens imperfections? What I mean is, an absolute glass imperfection of say 0.001mm with FF creates only 50% the image error compared with MFT, all else being equal. Not all else is equal, e.g. absolute lens precision is presumably more difficult as size increases? Corollary: a FF lens on a crop sensor camera with the same MP will give comparably lower resolution performance if sensors are beyond the MP limit. I think.

In practice, I’ve had great results using a FF lens with APS-C, so I’m guessing this MP limit is fairly high.

Also, does MF allow higher MP because of sensor pixel density limits, or the lens limit factor described, or simply higher price bracket?

 

[Tom Axford]

It is very difficult to make meaningful comparisons between lens resolution and sensor resolution without going into lots of messy details.

For instance, nearly all lenses show significantly more resolution near the centre of the image than towards the edges. Sensors have the same resolution everywhere.

Etc.

Probably the best information you are going to get is the MTF (Modulation Transfer Function) charts from those manufacturers who publish them. And, even then, in most cases, these are theoretical MTF charts for the lens design. They typically do not show the actual MTF for a particular lens, which may be much worse than the theoretical MTF (because of slight errors in the manufacturing process).

It's a jungle!

 

[John Crowe]

I have progressed over 15 years from 6 MP APS-C to 50 MP full frame, with 8 digital cameras. My impression is that the best lenses will continue to improve to about 100 MP on full frame.

Medium format, 33mm x 44mm, gets to 100 MP primarily based on sensor size. A 50 MP full frame sensor equates to a 33x44mm 84 MP sensor. Having said that many GFX 100 users are very happy with many full frame lenses, even compared to medium format lenses, so again it follows that the best full frame lenses can easily go beyond 60 MP on full frame. Sony does have a 60 MP full frame body and I have not heard about their typical lenses not being able to deliver on that body.

Using full frame lenses on APS-C gets a bit more complicated depending on the specific sensor. While APS-C has essentially been stuck at 24 MP for a decade, many in use today are less than that. There are also now two available at 32 MP, but in a smaller Canon APS-C sensor, so even tighter pixels.

Cropping a 50 MP full frame sensor to typical APS-C delivers 21.5 MP, so really not much different from the typical 24 MP. Cropped to Canon APS-C delivers only 19 MP compared to 32 MP of that sensor. This is a much wider difference. Once again I assume that users of these 32 MP cameras are getting the improved results from their full frame lenses. This suggests 84 MP full frame can easily be supported by current full frame lenses.

We seem to have proof that current full frame lenses can support an 84 MP full frame sensor. I am comfortable thinking that 100 MP would not be a problem.

For my typical print sizes and my shifting/stitching habits I may never go beyond my 50 MP full frame. For telephoto photography I would, however, jump at a 65 MP Canon APS-C (equivalent to 100 MP on Canon APS-C).

 

[D Cox]

A lens doesn't resolve a sensor: it's the other way around. The flow of information is from the outside world, through the lens (which limits the information), through the sensor (which further limits the information), the ADC, the JPG engine (if used), and the display or printer. Information is lost at each stage.

I doubt if there is as yet any sensor for the 24x26mm format that can record all the detail in the image from a sharp lens. You can of course use a soft lens, or set the maximum aperture so that most of the image is out of focus. A low resolution sensor may well be able to capture all the detail (such as it is) from such a setup.

Don Cox

 

[D Cox]

It is very difficult to make meaningful comparisons between lens resolution and sensor resolution without going into lots of messy details.

For instance, nearly all lenses show significantly more resolution near the centre of the image than towards the edges. Sensors have the same resolution everywhere.

Etc.

Probably the best information you are going to get is the MTF (Modulation Transfer Function) charts from those manufacturers who publish them. And, even then, in most cases, these are theoretical MTF charts for the lens design. They typically do not show the actual MTF for a particular lens, which may be much worse than the theoretical MTF (because of slight errors in the manufacturing process).

It's a jungle!

The lens tests that Roger Cicala used to publish on the Lens Rentals blog tell you quite a lot. They show measured MTFs for a number of lenses. He also posted some intelligent comments on such topics.

Don

 

[Tom Axford]

It is very difficult to make meaningful comparisons between lens resolution and sensor resolution without going into lots of messy details.

For instance, nearly all lenses show significantly more resolution near the centre of the image than towards the edges. Sensors have the same resolution everywhere.

Etc.

Probably the best information you are going to get is the MTF (Modulation Transfer Function) charts from those manufacturers who publish them. And, even then, in most cases, these are theoretical MTF charts for the lens design. They typically do not show the actual MTF for a particular lens, which may be much worse than the theoretical MTF (because of slight errors in the manufacturing process).

It's a jungle!

The lens tests that Roger Cicala used to publish on the Lens Rentals blog tell you quite a lot. They show measured MTFs for a number of lenses. He also posted some intelligent comments on such topics.

Don

Yes, that blog is one of the best sources of detailed measurements on real lenses. Unfortunately, they apply to only a small selection of lenses.

 

[57even]

Out of curiosity, what is the effective MP resolution of current high-end FF lenses (or any format actually)? I.e. can they fully resolve say a 100MP sensor?

Also, is there a sensor size sweet spot to minimise the impact of lens imperfections? What I mean is, an absolute glass imperfection of say 0.001mm with FF creates only 50% the image error compared with MFT, all else being equal. Not all else is equal, e.g. absolute lens precision is presumably more difficult as size increases? Corollary: a FF lens on a crop sensor camera with the same MP will give comparably lower resolution performance if sensors are beyond the MP limit. I think.

In practice, I’ve had great results using a FF lens with APS-C, so I’m guessing this MP limit is fairly high.

Also, does MF allow higher MP because of sensor pixel density limits, or the lens limit factor described, or simply higher price bracket?

There isn't a limit exactly.

If you take a lens at it's best aperture you could resolve far more than any current sensor, but at quite low contrast. You wouldn't necessarily see any more detail, just less aliasing.

As you stop down, the lens will be increasingly diffraction limited, and at large apertures, aberration limited. This will always be more noticeable on a high resolution sensor.

And because smaller sensors have higher spatial resolution for the same MP (more pixels per mm) the diffraction limit is lower - but at the same time, you don't need to stop down as far for the same DoF.

But 100 MP is perfectly practical for FF, and 50 MP for APSC. Arguably, the more you have the better to prevent colour moire even if you can't see much extra detail.

 

[sybersitizen]

... I’ve had great results using a FF lens with APS-C, so I’m guessing this MP limit is fairly high.

It is fairly high. I and others have tested this using mainstream full frame lenses on high density small sensors that have pixel pitches equivalent to more than 100MP on full frame. The higher pixel density can record more detail with the same lens.

Also, does MF allow higher MP because of sensor pixel density limits, or the lens limit factor described, or simply higher price bracket?

Medium format can offer more pixels from the same sensor technology simply by virtue of its greater area. Of course costs will be greater for the larger chip.

 

[DylanPH]

Out of curiosity, what is the effective MP resolution of current high-end FF lenses (or any format actually)? I.e. can they fully resolve say a 100MP sensor?

Some can, others cannot.

It doesn't matter either way, high res sensor has ZERO down sides other than cost and file size. Cost will get lower, and file size can be resolved simply by having small RAW options.

Also, is there a sensor size sweet spot to minimise the impact of lens imperfections?

yes, size of the universe, bigger is better.

What I mean is, an absolute glass imperfection of say 0.001mm with FF creates only 50% the image error compared with MFT, all else being equal. Not all else is equal, e.g. absolute lens precision is presumably more difficult as size increases? Corollary: a FF lens on a crop sensor camera with the same MP will give comparably lower resolution performance if sensors are beyond the MP limit. I think.

In practice, I’ve had great results using a FF lens with APS-C, so I’m guessing this MP limit is fairly high.

yea, so that you can use them on the A7RIV, which is even higher than most APS-C cameras

Also, does MF allow higher MP because of sensor pixel density limits, or the lens limit factor described, or simply higher price bracket?

mostly lens and price.

 

[MarkMyWords]

Thanks all, appreciate the insights. Sounds like a 100MP FF sensor is supportable by current lenses.

Another sensor comparison comes to mind: a 100MP MF sensor has the same pixel size and density as a 50MP FF sensor. Assuming they are made using the same technology, the only differences between MF and FF, all else being equal, would be…?

- 1 stop shallower DOF

- 1 stop more light

- 40% additional edges enlargement/cropping

Which suggests limited value for MF, given its cost and size?

 

[SrMi]

Thanks all, appreciate the insights. Sounds like a 100MP FF sensor is supportable by current lenses.

Another sensor comparison comes to mind: a 100MP MF sensor has the same pixel size and density as a 50MP FF sensor. Assuming they are made using the same technology, the only differences between MF and FF, all else being equal, would be…?

- 1 stop shallower DOF

- 1 stop more light

- 40% additional linear enlargement/cropping

See appendix in Roger's article:

More Ultra High-Resolution MTF Experiments

I get asked several times a week if this lens or that is ‘capable of resolving’ this number of megapixels. Some people seem to think a lens should be ‘certified’ for a certain number of pixels or something. That’s not how it works. That’s not how any of it works.

A 100MP MF (Fuji) sensor has the same pixel size as a 61MP FF sensor.

Given the same technology, larger sensors have more DR (output level).

 

[MarkMyWords]

Thanks all, appreciate the insights. Sounds like a 100MP FF sensor is supportable by current lenses.

Another sensor comparison comes to mind: a 100MP MF sensor has the same pixel size and density as a 50MP FF sensor. Assuming they are made using the same technology, the only differences between MF and FF, all else being equal, would be…?

- 1 stop shallower DOF

- 1 stop more light

- 40% additional linear enlargement/cropping

See appendix in Roger's article:

More Ultra High-Resolution MTF Experiments

I get asked several times a week if this lens or that is ‘capable of resolving’ this number of megapixels. Some people seem to think a lens should be ‘certified’ for a certain number of pixels or something. That’s not how it works. That’s not how any of it works.

A 100MP MF (Fuji) sensor has the same pixel size as a 61MP FF sensor.

Given the same technology, larger sensors have more DR (output level).

DR is “the ratio between the maximum and minimum signal that is acquired by the sensor.” Not sure how simply doubling the number of identically performing pixels would increase DR? And by how much, theoretically?

 

[SrMi]

Thanks all, appreciate the insights. Sounds like a 100MP FF sensor is supportable by current lenses.

Another sensor comparison comes to mind: a 100MP MF sensor has the same pixel size and density as a 50MP FF sensor. Assuming they are made using the same technology, the only differences between MF and FF, all else being equal, would be…?

- 1 stop shallower DOF

- 1 stop more light

- 40% additional linear enlargement/cropping

See appendix in Roger's article:

More Ultra High-Resolution MTF Experiments

I get asked several times a week if this lens or that is ‘capable of resolving’ this number of megapixels. Some people seem to think a lens should be ‘certified’ for a certain number of pixels or something. That’s not how it works. That’s not how any of it works.

A 100MP MF (Fuji) sensor has the same pixel size as a 61MP FF sensor.

Given the same technology, larger sensors have more DR (output level).

DR is “the ratio between the maximum and minimum signal that is acquired by the sensor.” Not sure how simply doubling the number of identically performing pixels would increase DR? And by how much, theoretically?

Sony a7rIV and Fuji GFX 100 have the same sensor technology and pixel size. Here is a comparison:

https://www.photonstophotos.net/Charts/PDR.htm#FujiFilm GFX 100,Sony ILCE-7RM4

The key is to compare at the same output size, not at pixel size.

 

[57even]

Thanks all, appreciate the insights. Sounds like a 100MP FF sensor is supportable by current lenses.

Another sensor comparison comes to mind: a 100MP MF sensor has the same pixel size and density as a 50MP FF sensor. Assuming they are made using the same technology, the only differences between MF and FF, all else being equal, would be…?

- 1 stop shallower DOF

- 1 stop more light

- 40% additional edges enlargement/cropping

Which suggests limited value for MF, given its cost and size?

There is an incremental improvement in noise and DR for the same image size, and a larger gap between diffraction-limited and aberration limited lens performance. In other words, a larger sharp working range. This is simply because there is less physical enlargement between the sensor and the image.

So the difference, very roughly speaking, is similar to APSC vs FF.

 

[Bob A L]

I find this an interesting and maybe even puzzling question myself. I don't know if anyone else has ever done this, but some years ago, I worked up a spreadsheet that calculates pixel dimensions for various size sensors at various mp values. And as small as a FF 100mp sensor goes, it is eclipsed by some of the small sensor claims especially in phone cameras. If you do the math and then visualize how many pixels would fit across the thickness of a piece of 20lb bond printer paper (approx .0035") for example, it is absolutely amazing how small these pixels will actually be. I just have trouble wrapping my mind around the possibility of making something that small work in the first place. Nowdays when some phone mfgs claim 100mp+ in a sensor less than 1/4" on the long side, I can't comprehend how this works from a physical standpoint. Giving them the luxury of using the largest phone sensor I know of this would relate to a FF sensor of over 1300 mp. I understand electronics tech has far exceeded what we ever considered years ago, but it still boggles my mind. Even a FF 100mp sensor will have pixels that at least 28 of them will fit across the thickness of typical printing paper, not to mention how small they would be from a 100mp 1/4" wide sensor. Really amazing stuff, and almost beyond comprehension for us mortals. Not something you want to dwell on and stay sane. Nevertheless modern lens technology must be totally fantastic to resolve such small detail.

 

[57even]

I find this an interesting and maybe even puzzling question myself. I don't know if anyone else has ever done this, but some years ago, I worked up a spreadsheet that calculates pixel dimensions for various size sensors at various mp values. And as small as a FF 100mp sensor goes, it is eclipsed by some of the small sensor claims especially in phone cameras. If you do the math and then visualize how many pixels would fit across the thickness of a piece of 20lb bond printer paper (approx .0035") for example, it is absolutely amazing how small these pixels will actually be. I just have trouble wrapping my mind around the possibility of making something that small work in the first place. Nowdays when some phone mfgs claim 100mp+ in a sensor less than 1/4" on the long side, I can't comprehend how this works from a physical standpoint. Giving them the luxury of using the largest phone sensor I know of this would relate to a FF sensor of over 1300 mp. I understand electronics tech has far exceeded what we ever considered years ago, but it still boggles my mind. Even a FF 100mp sensor will have pixels that at least 28 of them will fit across the thickness of typical printing paper, not to mention how small they would be from a 100mp 1/4" wide sensor. Really amazing stuff, and almost beyond comprehension for us mortals. Not something you want to dwell on and stay sane. Nevertheless modern lens technology must be totally fantastic to resolve such small detail.

Actually, it isn't.

Small sensors is that lenses have small apertures and are diffraction limited at almost all apertures, and require moulded lenses that cause quite a lot of aberration.

100 MP just allows you to do a lot of oversampling and 'digital zooming'. None of them AFAIK actually produce 100 MP images, or even close. They mostly settled for 12 MP, which is still good enough for 4K.

If you combine blocks of 4 pixels you get a 25 MP image with full colour per pixel, which theoretically looks better than a native 25 MP image.

This is much like shooting 4K video and downsizing it to 2K. Looks sharper.

 

[tbcass]

This can actually can be more of a problem with small sensors because of how densely packed the "pixels" are.

--
Tom