Sensor crop question

[musicshootr]

I promise I'm not starting this thread to start arguments about equivalency, I just really want to understand something.

First of all, are the below assumptions correct?

Larger sensors can gather more light, and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

Large full frame lenses passing light through large mounts are efficient from a light transmission perspective, allegedly leading to greater low light performance in instances like Nikon's Z mount.

High-megapixel sensors take a bit of a hit when it comes to low light performance, and the effect is more pronounced the smaller the sensor is (i.e. the move from 24 to 40 mp on APSC is going to introduce noise, where you can get a clean image at 40mp on FF and only have a noise penalty moving up to, say, 60mp?)

If the above is true, then talk me through something.

Fujifilm is instituting digital cropping on their 40mp APSC sensors to give the user more available focal lengths. When you crop in on that high-mp-count APSC sensor, are you getting the light-gathering capability of APSC just digitally cropped in, or does the light gathering take a hit because you're reading out a smaller portion of the sensor?

On the GFX100RF, if you use it in FF 35mm crop mode, do you get better light gathering performance than putting a 35mm lens on a FF camera (assuming equal aperture on the lens) because the overall sensor is better, or is that 35mm FF mode equivalent to putting a 35mm lens on an equivalent megapixel count, equivalent aperture FF system?

Do Sony FF cameras that have an APSC crop mode perform better in low light in that mode than Sony APSC cameras using the same lens at the same aperture?

Are FF cameras giving up some of the low light and optical benefits of the FF format by opting for higher megapixel sensor designs so folks can crop in to smaller sensor areas?

I am just sort of hazy on whether you incur the penalties of a smaller sensor format by cropping into that sensor area on a larger overall sensor and system.

Thank you in advance for your replies.

 

[Kenneth Morris Lee]

[emphasis added]:

Larger sensors can gather more light, and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

Large full frame lenses passing light through large mounts are efficient from a light transmission perspective, allegedly leading to greater low light performance in instances like Nikon's Z mount.

High-megapixel sensors take a bit of a hit when it comes to low light performance, and the effect is more pronounced the smaller the sensor is (i.e. the move from 24 to 40 mp on APSC is going to introduce noise, where you can get a clean image at 40mp on FF and only have a noise penalty moving up to, say, 60mp?)

For those assumptions that are true... to what extent ? And how is it measured ?

 

[musicshootr]

Dynamic range and shadow noise

 

[Kenneth Morris Lee]

Sorry, I was unclear.

If those assumptions are correct: how much penalty or benefit result from the different designs ? To what extent do these factors matter in actual practice ? Under what conditions will any differences be observed ?

 

[Richard Butler]

I promise I'm not starting this thread to start arguments about equivalency, I just really want to understand something.

I doubt you'd get an argument in this forum (there's not much to argue about)

First of all, are the below assumptions correct?

Larger sensors can gather more light, and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

I wouldn't go that far. Large sensors gather more light (when used with a lens of the same f-number). Pixel size can start to make a difference when you have very, very low levels of signal, but for most commonly-used ISOs, pixel size doesn't make much of a difference. And even then, which generation of sensors are we talking about? Generally it's sensor size that's the biggest factor.

Large full frame lenses passing light through large mounts are efficient from a light transmission perspective, allegedly leading to greater low light performance in instances like Nikon's Z mount.

A large lens mount gives you more design flexibility. Maybe you can design a lens that hits the edges/corners of the sensor at more easily accepted angles, but I wouldn't go so far as to assume a mount size / low light performance connection.

High-megapixel sensors take a bit of a hit when it comes to low light performance, and the effect is more pronounced the smaller the sensor is (i.e. the move from 24 to 40 mp on APSC is going to introduce noise, where you can get a clean image at 40mp on FF and only have a noise penalty moving up to, say, 60mp?)

The move from 24 to 40MP doesn't necessarily introduce noise (it might add a tiny bit of additional read noise, a difference that you might see at super high ISOs). At the pixel level it'll tend to look noisier because each pixel has received less light. But if you view the two images at the same size, you'll end up seeing the results of similar overall light capture, so the two images will look very similar.

These three short articles cover all of this in very simple terms .

If the above is true, then talk me through something.

Fujifilm is instituting digital cropping on their 40mp APSC sensors to give the user more available focal lengths. When you crop in on that high-mp-count APSC sensor, are you getting the light-gathering capability of APSC just digitally cropped in, or does the light gathering take a hit because you're reading out a smaller portion of the sensor?

Cropping into a small region of the sensor is exactly the same as using a sensor of that smaller size. If you use the (26MP) APS-C crop mode on the 61MP Sony a7 models, you get essentially identical images to those shot with the 26MP Sony a6700 (they use essentially the same pixel design), and very similar to the 24MP APS-C cameras. And, similar to most modern APS-C sensors.

On the GFX100RF, if you use it in FF 35mm crop mode, do you get better light gathering performance than putting a 35mm lens on a FF camera (assuming equal aperture on the lens) because the overall sensor is better, or is that 35mm FF mode equivalent to putting a 35mm lens on an equivalent megapixel count, equivalent aperture FF system?

If you crop to use a 36 x 24mm region of a large sensor, you're effectively using a full-frame camera. There's nowhere for any benefit to come from. The overall sensor isn't better (again, it's the same 3.76um pixels as the 61MP Sony cameras and the 26MP APS-C cameras).

Do Sony FF cameras that have an APSC crop mode perform better in low light in that mode than Sony APSC cameras using the same lens at the same aperture?

No

Are FF cameras giving up some of the low light and optical benefits of the FF format by opting for higher megapixel sensor designs so folks can crop in to smaller sensor areas?

Yes

I am just sort of hazy on whether you incur the penalties of a smaller sensor format by cropping into that sensor area on a larger overall sensor and system.

You do. Using a crop is the same as using a smaller sensor.

Richard - DPReview.com

 

[musicshootr]

I don't know - I was hoping the forum could discuss it and help me understand the main issue at hand, that being whether crop modes in high MP count cameras incur the same penalties as a smaller sensor size.

 

[musicshootr]

This is super helpful. Thank you!

 

[JimKasson]

I promise I'm not starting this thread to start arguments about equivalency, I just really want to understand something.

Richard gave a good answer. Let me take a crack at it and see if another perspective helps.

First of all, are the below assumptions correct?

Larger sensors can gather more light,

At the same exposure in lux-seconds. But you need more light to take full advantage of that.

and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

Highly questionable. In most photography, the most important source of noise is photon noise. And full well capacities seem to run about 3000 electrons per square micrometer, regardless of pixel ptich.

Large full frame lenses passing light through large mounts are efficient from a light transmission perspective, allegedly leading to greater low light performance in instances like Nikon's Z mount.

Sort of. There is less likelihood of mechanical vignetting if the throat diameter is large.

High-megapixel sensors take a bit of a hit when it comes to low light performance,

That's not my experience.

and the effect is more pronounced the smaller the sensor is (i.e. the move from 24 to 40 mp on APSC is going to introduce noise, where you can get a clean image at 40mp on FF and only have a noise penalty moving up to, say, 60mp?)

If the above is true, then talk me through something.

Fujifilm is instituting digital cropping on their 40mp APSC sensors to give the user more available focal lengths. When you crop in on that high-mp-count APSC sensor, are you getting the light-gathering capability of APSC just digitally cropped in, or does the light gathering take a hit because you're reading out a smaller portion of the sensor?

If you're using a smaller portion of the sensor, it's as if you were using a camera with a smaller sensor.

On the GFX100RF, if you use it in FF 35mm crop mode, do you get better light gathering performance than putting a 35mm lens on a FF camera (assuming equal aperture on the lens)

No.

 

[musicshootr]

This is so interesting to me. I think of a lens and lens mount kind of like a window letting light shine in on the sensor, so a big window (FF lens and mount) not letting in more light to the sensor in a FF system cropped in digitally to APSC makes little intuitive sense to me, but I believe you. I suppose it makes sense once I think about lens diagrams, though. The angles the light is entering the body at and hitting the sensor are designed for the light to hit the sensor all the way out to the corners so it makes sense that that light would be lost if you crop that sensor area out.

If you care a lot about low light performance, are you further ahead shooting a photo at the full size and MP count and cropping in post? Or is that not any different than using the digital teleconverter? (Obviously the better solution would be to use the desired focal length from the start, but if one has to crop, is there any advantage to waiting to do so in post?)


I'm asking this fully hypothetically, by the way. I don't own any cameras that have this functionality.

 

[Richard Butler]

This is so interesting to me. I think of a lens and lens mount kind of like a window letting light shine in on the sensor, so a big window (FF lens and mount) not letting in more light to the sensor in a FF system cropped in digitally to APSC makes little intuitive sense to me, but I believe you. I suppose it makes sense once I think about lens diagrams, though. The angles the light is entering the body at and hitting the sensor are designed for the light to hit the sensor all the way out to the corners so it makes sense that that light would be lost if you crop that sensor area out.

Bingo.

If you look at the diagram at the top of this page and then mouse-over the word APS-C, you can see that using an APS-C sensor means you only capture the central section of the light being projected by the lens.

It doesn't matter whether you interpret the diagram as being a full-frame lens being placed on an APS-C sensor or you cropping-in and only using the APS-C region of a full-frame sensor: the extra light beyond that region isn't going into your final image.

If you care a lot about low light performance, are you further ahead shooting a photo at the full size and MP count and cropping in post? Or is that not any different than using the digital teleconverter? (Obviously the better solution would be to use the desired focal length from the start, but if one has to crop, is there any advantage to waiting to do so in post?)

Again, as soon as you crop away pixels, you're no longer using the light that was captured by the parts of the sensor whose capture you've cropped out.

So the best way to make full use of any sensor is to use a lens that projects the image you want to take onto the full area of the sensor.

Richard - DPReview.com

 

[Erik Kaffehr]

I promise I'm not starting this thread to start arguments about equivalency, I just really want to understand something.

First of all, are the below assumptions correct?

Larger sensors can gather more light, and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

Yes larger sensors can collect more light. But, the size of the pixels does not matter a lot.

A sensor of a given size can collect a given amount of light, regardless of pixel size, as long as the fill factor is near 100% on the pixels.

Large full frame lenses passing light through large mounts are efficient from a light transmission perspective, allegedly leading to greater low light performance in instances like Nikon's Z mount.

The size of the lens matter little, it is the size of the inlet pupil that matters. The inlet pupil determines the amount of light projected on the sensor. A lens having a wider angle of view will see more light than a lens with a narrower angle of view.

High-megapixel sensors take a bit of a hit when it comes to low light performance, and the effect is more pronounced the smaller the sensor is (i.e. the move from 24 to 40 mp on APSC is going to introduce noise, where you can get a clean image at 40mp on FF and only have a noise penalty moving up to, say, 60mp?)

No, that is not the case, as long as the images are viewed at the same size.

If the above is true, then talk me through something.

Fujifilm is instituting digital cropping on their 40mp APSC sensors to give the user more available focal lengths. When you crop in on that high-mp-count APSC sensor, are you getting the light-gathering capability of APSC just digitally cropped in, or does the light gathering take a hit because you're reading out a smaller portion of the sensor?

The capability of the sensor to collect light mostly dependent on sensor size.

On the GFX100RF, if you use it in FF 35mm crop mode, do you get better light gathering performance than putting a 35mm lens on a FF camera (assuming equal aperture on the lens) because the overall sensor is better, or is that 35mm FF mode equivalent to putting a 35mm lens on an equivalent megapixel count, equivalent aperture FF system?

If you put a 35 mm lens on a 33x44 mm sensor cropped to 24x36 mm it will be a 35 mm lens on a 24x36 mm sensor, as simple as that.

Do Sony FF cameras that have an APSC crop mode perform better in low light in that mode than Sony APSC cameras using the same lens at the same aperture?

No. Better performance in low light needs a faster lens.

Are FF cameras giving up some of the low light and optical benefits of the FF format by opting for higher megapixel sensor designs so folks can crop in to smaller sensor areas?

No.

I am just sort of hazy on whether you incur the penalties of a smaller sensor format by cropping into that sensor area on a larger overall sensor and system.

Cropping an image throws away both detail and photon statistics. It is possible to compensate for that by optimizing the system design around the smaller sensor, essentially by increasing the aperture number.

What happens when going from 24x36 mm to APS-C may be (using the same lens design):

• The lens will have shorter focal length, so the image will be projected on a shorter distance.
• Angular resolution will be the same.
• Resolution per millimeter will go up.
• Downscaling the lens will make the inlet pupil smaller, while keeping field of view constant, thus reducing light collected within a given exposure time.
• If inlet pupil size is kept, the aperture will be faster, but because of the faster aperture more optical correction would be needed.

In the end, consider the inlet pupil diameter with field of view the limiting factor.

Best regards

Erik

Thank you in advance for your replies.

 

[chrisfisheye]

I promise I'm not starting this thread to start arguments about equivalency, I just really want to understand something.

First of all, are the below assumptions correct?

Larger sensors can gather more light, and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

There are other debates than equivalency.. There is also the debate about: It is the lens that collects the light.

In practice, you can ignore this debate and jump directly to the conclusions.

But putting at the same level the lens and the sensor and considering that both contribute to light gathering is a nonsense (imho).

A better view is to consider that this the lens that collects the light. The sensor "only" records it. This distinction is important.

Take the same lens, same aperture. The larger sensor will record more light. So you could think that you need it and that this is thanks to the combination lens/sensor. Mostly wrong ! In most cases, you could have simply use a focal reducer. Said differently, it is the lens that does the job, not the sensor.

Tait said, it is true that the sensor size allows to break some limitations. But this is not really a theorical argument because it is currently the medium refractive index of the air which causes this limitation. Other medium could be used, in theory.

And last argument but not least.;

A given aperture size sees the scene identically. Whatever focal length, sensor size, only this parameter matters. Every object in the scene is seen identically, this includes:

• same optical resolution (diffraction will be similar)
• same light gathering for this object

Take this example: get a huge lens designed for a big sensor but combine it with a tiny sensor.. Do you necessarily consider that you collect less light ?. Well maybe not for the star you were able to shoot with this lens. You see also that in astrophotography this is of course the lens which matters. Who cares about the sensor ???

By the way a sensor has nothing to do with optics, it is not part of the optical system.

Sorry if I diminish a bit the role of the sensor

 

[ThrillaMozilla]

First of all, are the below assumptions correct?

Larger sensors can gather more light,

That's the conventional wisdom, but I think it's a little misleading. I would say the lens collects the light and the sensor records it.* If the lens doesn't collect the light, then the sensor doesn't receive it. A large sensor can "collect" more light only because of the lens. For the same field of view a longer lens is used with a larger sensor, which could have a larger aperture. It's the larger aperture that collects more light. (A larger aperture means a larger entrance pupil, or front lens diameter, not a smaller f-number.) You could say that using the same lens the larger sensor collects more total light because the field of view is larger, but this is probably not an appropriate comparison because you would be either cropping or making a very different image.

You could buy a camera with a larger sensor, but you would not record any more light from the subject unless you also bought a different lens. The number of pixels per bird depends only on the lens aperture diameter!** That could be an expensive lesson if you didn't know.

Any additional light collection also entails a narrower depth of field, which may or may not be important. It could of course be an advantage, a penalty, or inconsequential. (Conversely, though, a larger format does not carry a penalty if a wider depth of field is required, because you can simply stop down! If this means nothing to you, ignore it.)

and large pixels on that large sensor (i.e. 24mp FF cameras) have the greatest low-light performance.

Usually not. The usual reasoning is that larger pixels collect more light, but that reasoning is spurious and misleading.

...High-megapixel sensors take a bit of a hit when it comes to low light performance,

No. That is a common but confused point of view. See Richard Butler's reply for more details. More pixels on the subject is good.

...If the above is true, then talk me through something.

Fujifilm is instituting digital cropping on their 40mp APSC sensors to give the user more available focal lengths.

The size of the sensor does not affect the focal length or the size of the image on the sensor!

When you crop in on that high-mp-count APSC sensor, are you getting the light-gathering capability of APSC just digitally cropped in, or does the light gathering take a hit because you're reading out a smaller portion of the sensor?

People usually complicate this and get themselves into all sorts of conundrums. Here's a very simple thought experiment you can do. Take a photographic print in your hand, and with a pair of scissors cut about 1/3 off the width and height.

That's what it means to use an APS-C sensor or crop mode on a full-frame camera. It really is that simple.

If you want the same field of view that you had before using scissors, then you'll have to use a shorter lens. That lens might have a smaller diameter and therefore collect less light, but not necessarily.

Alternatively, if you want the same size of photo as before using scissors, and you use the same lens, you can enlarge the picture, and in so doing you will make imperfections and noise a little more apparent.

On the GFX100RF, if you use it in FF 35mm crop mode, do you get better light gathering performance than putting a 35mm lens on a FF camera (assuming equal aperture on the lens) because the overall sensor is better, or is that 35mm FF mode equivalent to putting a 35mm lens on an equivalent megapixel count, equivalent aperture FF system?

Do Sony FF cameras that have an APSC crop mode perform better in low light in that mode than Sony APSC cameras using the same lens at the same aperture?

Nope. Refer to the print and scissors analogy that I discussed above.

Are FF cameras giving up some of the low light and optical benefits of the FF format by opting for higher megapixel sensor designs so folks can crop in to smaller sensor areas?

I am just sort of hazy on whether you incur the penalties of a smaller sensor format by cropping into that sensor area on a larger overall sensor and system.

Thank you in advance for your replies.

Don't worry about it. If you know what you are doing and have some specific requirement for the best possible image under difficult conditions, and if you can tolerate a depth of field penalty to realize the benefits, then consider the additional expense (and probably size and weight) of full-frame equipment. Otherwise, you might be happier with APS-C.

=============

Footnotes:

*The word "records" is from Chrisfisheye. If I had read his reply, I would have needed to write and edit this!

** Ignoring differences in illumination and lens transmission, obviously.

 

[ThrillaMozilla]

This is so interesting to me. I think of a lens and lens mount kind of like a window letting light shine in on the sensor, so a big window (FF lens and mount)

The lens, yes, the mount, no. Be assured that the lens will funnel all that light through even the narrowest mount. For super-fast, super-expensive lenses you might get a bit more vignetting at the corners with a small mount.

not letting in more light to the sensor in a FF system cropped in digitally to APSC makes little intuitive sense to me, but I believe you. I suppose it makes sense once I think about lens diagrams, though. The angles the light is entering the body at and hitting the sensor are designed for the light to hit the sensor all the way out to the corners so it makes sense that that light would be lost if you crop that sensor area out.

If you care a lot about low light performance, are you further ahead shooting a photo at the full size and MP count and cropping in post? Or is that not any different than using the digital teleconverter?

More pixels per bird is better. I would use a longer lens or a teleconverter.

(Obviously the better solution would be to use the desired focal length from the start, but if one has to crop, is there any advantage to waiting to do so in post?)

Crop mode may have one advantage: it fills the viewfinder, displaying the intended subject with more magnification. And maybe it helps with focusing, or maybe not.

I'm asking this fully hypothetically, by the way. I don't own any cameras that have this functionality.

 

[J A C S]

Which one collects the dust - the broom or the dustpan?

 

[ThrillaMozilla]

Which one collects the dust - the broom or the dustpan?

What can be said unambiguously is that if you use a larger sensor but the same lens and aperture, you will not collect a single additional photon from the subject unless you redefine the subject to include a larger area. On the other hand, if you use the same sensor but a different lens or larger aperture, you can increase the number of photons collected from the same subject.

 

[J A C S]

Which one collects the dust - the broom or the dustpan?

What can be said unambiguously is that if you use a larger sensor but the same lens and aperture, you will not collect a single additional photon from the subject unless you redefine the subject to include a larger area.

Why would you use a larger sensor with the same lens (which is often hard or impossible)?

On the other hand, if you use the same sensor but a different lens or larger aperture, you can increase the number of photons collected from the same subject.

Unless the sensor saturates, assuming that such a lens even exists.

 

[chrisfisheye]

Which one collects the dust - the broom or the dustpan?

It is Hubble which collects light, difficult to justify your analogy... Putting on the same level the lens and the sensor is just a big nonsense (imho).

Besides you do not necessarily need a sensor to collect light, I mean just look in the optical viewfinder and you will see how good a lens performs. A lens by itself collects light without the need to work in conjunction with a sensor.

The sensor is nothing else than a recording device which is not part of the optical system.

 

[ThrillaMozilla]

Which one collects the dust - the broom or the dustpan?

What can be said unambiguously is that if you use a larger sensor but the same lens and aperture, you will not collect a single additional photon from the subject unless you redefine the subject to include a larger area.

Why would you use a larger sensor with the same lens (which is often hard or impossible)?

You would often use a smaller sensor with the same lens, in order to get a smaller pixel pitch. It follows therefore that you can also use a larger sensor with the same lens. Talking about using a larger sensor just made the example easier to describe.

On the other hand, if you use the same sensor but a different lens or larger aperture, you can increase the number of photons collected from the same subject.

Unless the sensor saturates, assuming that such a lens even exists.

 

[JimKasson]

Which one collects the dust - the broom or the dustpan?

What can be said unambiguously is that if you use a larger sensor but the same lens and aperture, you will not collect a single additional photon from the subject unless you redefine the subject to include a larger area.

Why would you use a larger sensor with the same lens (which is often hard or impossible)?

You would often use a smaller sensor with the same lens, in order to get a smaller pixel pitch.

Pixel pitch is not well correlated with sensor size in the MFT to 645ish range these days.

It follows therefore that you can also use a larger sensor with the same lens. Talking about using a larger sensor just made the example easier to describe.

On the other hand, if you use the same sensor but a different lens or larger aperture, you can increase the number of photons collected from the same subject.

Unless the sensor saturates, assuming that such a lens even exists.

--
https://blog.kasson.com