Small sensors gather less light? Wrong!

[djensen]

"If you are not concerned about DOF, and if technology was so good that pixel pitch didn't matter, and if you only used lenses optimised for your smaller sensor, then there wouldn't be any drawbacks to small sensors. "

I think that sums it up nicely...eliminate the design restraints of making sensors smaller (the space taken up by stuff that does not catch light, pixel size, etc...)

Focus the same image on a 3mm sensor or a 10mm sensor, the 3mm will have the same amount of light and information, just in a smaller space. It will be the same number of photons from the same lens just sent slightly further away to a smaller surface- the whole point of a lens.

...or am I totally missing the point of the original post?

-Dan

 

[Chippy99]

It's true that the f-number relates to a given brightness. But the second part of your sentence is incorrect. The brightness behind the lens can vary depending on the image size, even with the same external brightness and the same f number. Now, granted, making a smaller image size with involve higher glass curvature, and more use of parabolic lens elements which might be an issue.

The brightness (f-number) of a lens tells about the light level over the whole image circle, no matter if the image circle is is covering a narrow or wide area.

You can't gather the light outside the sensor and put this light into the area that is covered by the sensor. I think this is where your idea breaks down.

The more I think about this, I think I am wrong (and you are right) in that what I am really suggesting is lower f-stop lenses.

That said, there is nothing wrong in asserting that with different glass, the small sensor can gather just as much light as a larger one. And in respect of your sentence:

You can't gather the light outside the sensor and put this light into the area that is covered by the sensor. I think this is where your idea breaks down.

Yes you can. Why not.

 

[Chippy99]

"If you are not concerned about DOF, and if technology was so good that pixel pitch didn't matter, and if you only used lenses optimised for your smaller sensor, then there wouldn't be any drawbacks to small sensors. "

I think that sums it up nicely...eliminate the design restraints of making sensors smaller (the space taken up by stuff that does not catch light, pixel size, etc...)

Focus the same image on a 3mm sensor or a 10mm sensor, the 3mm will have the same amount of light and information, just in a smaller space. It will be the same number of photons from the same lens just sent slightly further away to a smaller surface- the whole point of a lens.

...or am I totally missing the point of the original post?

No, Dan, you have perfectly captured the whole point of my post!

We assume a 1/2.3" sensor in a p&s is "worse" than a larger 1/1.7" one. But this assumption is incorrect.

 

[DUSTY LENS]

I don't think you understand bro....you can't as much information in the same amount of time in a smaller circle of light...speed of light doesn't change here. Think of the reason you must raise ISO when you close down the aperture on a camera or use a lower shutter speed.

an 8mm circle of light will provide more than the 4mm circle. Same way a FF lens and FF sensor is better than corresponding APS-C or m4/3

Sorry mate, but you really need to open your mind and think. Thinking is so important. What the hell has the speed of light got to do with anything.

An 8mm circle of light does not "provide more" than a 4mm one. That is only true if the lens is the same. With different lenses, both the 8mm circle and the 4mm circle can receive EXACTLY the same amount of light.

So the sensor size, does NOT determine how much light is captured, if you design the lens for the sensor.

Sorry if you are finding this hard to grasp.

Chippy99

You make an interesting case here , but some of your parameters and presumptions are not valid .

1

People don't prefer the Full Frame Sensor because it gathers more light as you assume .

2

It isn't the entire sensor area that determines how much light intensity is needed or how much light intensity is gathered .

3

The lens focal length ( long for FF versus short for APS-C ) is not what determines how much light is gathered by the lens ) or delivered to the sensor .

The intensity of light is what matters for noise , shutter speed , ISO setting , and exposure level , not how much light is gathered by the sensor .

A small sensor will as you stated gather less light , but it does not gather more light intensity over it's area .

The pixel size , ( the area of each pixel ) is important in that a small pixel area will receive only 1/4 the number of photons as a pixel which is double the length and the width of the small pixel provided the light intensity and the time of gathering light is the same for both .

In the case of the small pixel gathering only 10 photons in a darker area of an image could produce noise variation of photons above 10 or below 10 , more than a larger pixel with 4 times the area gathering 40 photons in the same light level of the image area . If the large Pixel receives +1 or -1 photon the noise level is 1/4 the level of noise produced by the small (1/4 the area ) pixel when it receives +1 or -1 photons above 10 or below 10 photons . ( compared to +1 or -1 photons above or below 40 photons .

So it is not the size of the sensor , but it is the size of the pixel which determines sensitivity to noise .

The intensity of the light on any part of the sensor is determined only by the true F stop of the lens and the light intensity arriving at the lens from the source .

This intensity is the same with a small sensor and with a large sensor .

The advantages of the large sensor are in several parameters other than light gathering power and noise .

Dusty

 

[Nordstjernen]

So you start with a much larger lens than you need and put on a reduces, which adds bulk and weight, to make a brighter lens with much shorter focal length?

I know this principle from astronomical telescopes. But using a tele compressor (the elements that reduces the focal length) doesn't make the telescope smaller. The benefit is a brighter lens at the cost of shorter focal length and smaller useable image circle.

 

[OldScotch]

It's like saying a garden hose can gather as much water as a barrel, you just have to use a funnel.

That's true, but if it gets to be too much water, the funnel will overflow. And if you take your reasoning too far, the lens will burn a hole through the sensor meaning it's not capturing anything.

You can concentrate a lot of light on a very small area or a very large area, but there are limits as if it gets too small it can not be contained - therefore sensor size does matter with consideration to how much light can be captured.

 

[Nordstjernen]

You can't gather the light outside the sensor and put this light into the area that is covered by the sensor. I think this is where your idea breaks down.

Yes you can. Why not.

This is true. You can use a tele compressor (the opposite of a tele convertert). But this is not a good solution -- look at my answer further down in this thread. A way better solution would simply be a brighter lens with shorter focal length.

 

[Henrik Herranen]

Sounds like you know what you are taliking about, and I might be getting confused, so please can I pick your brain?

You're welcome.

Can we have 2 lenses with the same entrance pupil and the same focal length, but that with different glass curvatures have different magnification and therefore different image sizes?

By changing what you call magnification you have just changed focal length.

Take the 50mm f/1.4 lens in question. Magnify the image by two and what you have is a 100mm f/2.8: your physical aperture stays the same: 50mm/1.4 = 100mm/2.8 = 36 mm. This is a physical fact and cannot be changed. This is actually exactly what DSLR teleconverters do: they add magnification. A 2X magnified image is four times, or two stops less bright, and equally the f-number is two stops slower: f/2.8 vs f/1.4.

If we can, then the both have the same f-stop, but the produce different concentrations of light on the sensor. i.e. you can have a small sensor that receives the same amount of light as a large sensor, with two diffent lenses both with the same f-stop.

Unfortunately, this is the incorrect part. Lenses with the same f-stop (or, to be pedantic, lenses with the same t-stop, which is slightly different) always have the same light intensity (same amount of light per square millimeter). Adding "magnification" changes the apparent focal length of the lens and also increases the f number. If this wasn't the case you'd just invented the free lunch.

Hope this helps.

Kind regards,

• Henrik

--
And if a million more agree there ain't no great society
My obligatory gallery: http://www.iki.fi/leopold/Photo/Galleria/

 

[TrojMacReady]

didn't you?

Yes, I singled out pixel pitch because it isn't an issue at all. So my question back at you, why mention it?

 

[Henrik Herranen]

The brightness (f-number) of a lens tells about the light level over the whole image circle, no matter if the image circle is is covering a narrow or wide area.

No, this is incorrent.

The f number (or, to be pedantic, the t number) tells the intensity of the light, i.e. the brightness of the image per area unit, e.g. square centimeter.

Fact: an f/2.8 lens on a P&S will create an image that is equally bright as an image on a FF sensor. However, because of its size, the larger sensor collects more total light. This is exactly the reason of the FF advantage: the sensor, when used with larger optics, quite simply has more photons to work with.

Kind regards,

• Henrik

--
And if a million more agree there ain't no great society
My obligatory gallery: http://www.iki.fi/leopold/Photo/Galleria/

 

[qZar]

Assuming, we want to get similar photographic conditions, i.e. view angle and DOF, we would need to reduce lens focal length and to increase aperture.

But... there are many limiting factors in lens construction. Eg.:
glass refraction index - we are not able to reduce the focal length too much.
Distortions (aberrations)

lens focusing distance - when we want the lens to have adjustable focus we need to give it a distance from the sensor.
photo cells Signal/noise ratio - the smaller the cell the worse the ratio gets.
photo cell lens construction - the angle of incidence cannot be too small.
Zoom capabilities - requires even more distance from lens.

Actually, what you say is: for smaller sensor we can use greater apertures and we will get the same.

No, we will not. we are limited. to improve the quality it's easier to make the photo cell bigger.

 

[TrojMacReady]

didn't you?

Be fair - he just commented on the bit he didn't agree with and didn't see the need to say yes to everything he did agree with.

At least that's the way I read it.

I listed pixel pitch as a problem, but admitted further on in the same post that advances in technology would remove that particular drawback.

He replied by stating that pixel pitch isn't a problem due to the fact that technology is advancing.

But yes, I could be nicer.

Today it isn't a problem. And a few years ago it wasn't either (before cameras were equipped with small CMOS sensors). I remember the FZ50 being more efficient per area than the top of the bill Canon full frame camera, despite the first having a much smaller pixel pitch. It's an overstated issue and most of the time has been.

 

[Chippy99]

This intensity is the same with a small sensor and with a large sensor .

You completely miss the point.

 

[Chippy99]

Hope this helps.

Kind regards,

Indeed it does. Thanks!

So more concentrated light means - necessarily - a shorter focal length and therefore a higher f-stop. I was beginning to think as much.

Having said that, it begs the question what makes lenses expensive. Is it principally the focal length, or the maximum aperture? Of course I am sure it's both.

But a low f-number (f/1.2, f/0.8) etc might not be prohibitively expensive with a very small lens with a short focal length.

 

[Chippy99]

Hope this helps.

Kind regards,

Indeed it does. Thanks!

So more concentrated light means - necessarily - a shorter focal length and therefore a higher f-stop. I was beginning to think as much.

Having said that, it begs the question what makes lenses expensive. Is it principally the focal length, or the maximum aperture? Of course I am sure it's both.

But a low f-number (f/1.2, f/0.8) etc might not be prohibitively expensive with a very small lens with a short focal length.

Typo alert, I meant:

So more concentrated light means - necessarily - a shorter focal length and therefore a lower f-stop. I was beginning to think as much.

 

[Henrik Herranen]

1

People don't prefer the Full Frame Sensor because it gathers more light as you assume .

People prefer FF sensors for many reasons, but the main reason for their high-ISO capacity is that they capture more light, as counted in "photons per sensor area".

The intensity of light is what matters for noise , shutter speed , ISO setting , and exposure level , not how much light is gathered by the sensor .

This is incorrect and just the opposite. Light is a discrene phenomenon, and at the end of the day the total number of photons captured determines the noise level of the final image - particularly at high ISO.

The pixel size , ( the area of each pixel ) is important in that a small pixel area will receive only 1/4 the number of photons as a pixel which is double the length and the width of the small pixel provided the light intensity and the time of gathering light is the same for both .

Yes, but there are four times as many of them!

While pixel size certainly affects pixel quality, if you intend to print / watch the image at a given size (as everyone except pixel peepers do), it is irrelevant. It is only the quality of the final image that matters. Smaller pixels may be more noisy (and usually they are) but this is offset by there being more of them. The noise cancels out when scaling to a given size, or at least becomes very finely grained.

I explained an analogy of this in some detail recently at:
http://forums.dpreview.com/forums/readflat.asp?forum=1041&message=35472137

The article link shows how audio reproduction was revolutionized in the late 1980's when some guys invented that you need only one-bit digital-to-analog converters to represent 16-bit data if you use enough oversampling (equivalent in the imaging world to having "enough" pixels).

Hope this helps.

Kind regards,

• Henrik

--
And if a million more agree there ain't no great society
My obligatory gallery: http://www.iki.fi/leopold/Photo/Galleria/

 

[kilobytemachine]

Forgive me posting this here - it's not Sony specific, but I am a Sony user and it will get read more here than in some off-topic forum.

But, this is something I only thought about when waking up this morning. It's a universally accepted truth that small sensors gather less light than larger sensors and the former are therefore not as good in low light conditions, producing more noise in high ISO shots.

Everyone knows this, right? I mean it's obvious. The photocells are smaller and they can capture less photos in a given time compared to larger ones. Obviously.

But this is WRONG.

It would be true if the sensor was waved around in the air naked, but in a camera, it is not. The flaw in the paragraph above is that it should read "The photocells are smaller and they can capture less photos in a given time compared to larger ones, given the same level of illumination " And therein lies the flaw.

The amount of light gathered and available for the sensor is not governed by the sensor, it's governed by the lens . If you focus the light from the lens onto a smaller area, the intensity of light increases as the area gets smaller and smaller - ultimately like a magnifying glass in the sun with a spot that is so bright and hot that it burns paper. But all the light is still there, in the tiny spot. That tiny spot is "seeing" all the photons just as the larger magnifying glass lens front is "seeing" them.

Now, clearly if you put a Canon 50mm f/1.4 on the front of an APS-C sensor, in the focal plane, the APS-C sensor takes up less area than an full frame sensor, and so it is true in that case that the smaller sensor gets less of the light.

So my thoughts maybe true, but of only academic interest in a DSLR where things like the distance from the lens rear element to the focal plane are fixed and determined by the system. The smaller sensor sees less of the lens' light and there's not much can be done about it.

But in a proprietary system like a p&s, everything is up for grabs by the designers. It is perfectly possible to have an APS-C sensor gather exactly the same amount of light as a FF one, but focusing all the available light from the lens onto the sensor. The sensor size does not determine how much light the lens captures, the lens does! Putting a bigger sensor behind the glass would not capture any more light if the smaller sensor was already capturing all the light!

So, given the above is true (and thinking about it, clearly it is), I wonder why the small-sensors-gather-less-light myth persists?

Sure, small sensors enable the use of smaller and cheaper lenses, which then gather less light. But that does not need to be the case. Someone could quite easily design a p&s camera witha 1/1.7" sensor that performed just as well as a A900 in terms of light gathering. Practically speaking, you'd have a massive camera though, because the lens would have to be as large as a FF system lens. So the benefit of having a very small sensor is largely eliminated and there would be little point in the design.

And there maybe other practicle issues that influence sensor performance. I am not an expert, but I would not be surprised if heat in the sensor also produces noise and a smaller sensor will be less able to dissipate heat than a larger one.

But consider this. Canon (or whoever) come out with a new p&s and everyone immediately looks to see if the sensor is 1/2.3" or 1/1.7" in size, because they think the larger sensor must gather more light. But this is just plain wrong. Maybe the smaller sensor gathers more? Who knows. It depends on how the lens has been designed, not on the sensor size.

Bizarre, but true!

You've just described the effect of putting a f/1.4 lense on a crop sensor camera vs a f/2.8 lense on a FF sensor camera. (approximately speaking, depending on crop factor) The exposure would be the same. However, that same 1.4 lense can also be mounted on the FF sensor camera. To receive the same amount of light, the cropped camera would need a f/0.7 lense. Kinda hard to come by those...

No, I haven't, but I can understand the confusion. The f number is the ratio of the entrance pupil diameter over the focal length.

Envisage two lenses with the same entrance pupil and the same focal length. But one focuses all the light perfectly onto an APS-C sensor; the other focuses all the light perfectly onto an FF sensor.

Both sensors receive the same amount of light and both lenses have the same f-stop.

I am not talking about faster lenses to achieve the same amount of light capture.

So what you're saying is it's possible to have two lenses with the same f number and focal length but producing different illumination per unit area?

You know, I am not 100% sure about this. I wrote it, and was 100% sure, and now I am thinking perhaps that's wrong. I need to think about that a bit more!

(And by the way, my f-stop definition above is the wrong way up - it's focal length divided by aperture, not the other way around. Silly mistake.)

Actually, I dearly wish you were correct. That way, I'd potentially be able to get D3S noise performance on my budget 4/3 camera.

But seriously, if you were correct, then every light meter ever produced would be of somewhat limited use.

 

[Chippy99]

Forgive me posting this here - it's not Sony specific, but I am a Sony user and it will get read more here than in some off-topic forum.

But, this is something I only thought about when waking up this morning. It's a universally accepted truth that small sensors gather less light than larger sensors and the former are therefore not as good in low light conditions, producing more noise in high ISO shots.

Everyone knows this, right? I mean it's obvious. The photocells are smaller and they can capture less photos in a given time compared to larger ones. Obviously.

But this is WRONG.

It would be true if the sensor was waved around in the air naked, but in a camera, it is not. The flaw in the paragraph above is that it should read "The photocells are smaller and they can capture less photos in a given time compared to larger ones, given the same level of illumination " And therein lies the flaw.

The amount of light gathered and available for the sensor is not governed by the sensor, it's governed by the lens . If you focus the light from the lens onto a smaller area, the intensity of light increases as the area gets smaller and smaller - ultimately like a magnifying glass in the sun with a spot that is so bright and hot that it burns paper. But all the light is still there, in the tiny spot. That tiny spot is "seeing" all the photons just as the larger magnifying glass lens front is "seeing" them.

Now, clearly if you put a Canon 50mm f/1.4 on the front of an APS-C sensor, in the focal plane, the APS-C sensor takes up less area than an full frame sensor, and so it is true in that case that the smaller sensor gets less of the light.

So my thoughts maybe true, but of only academic interest in a DSLR where things like the distance from the lens rear element to the focal plane are fixed and determined by the system. The smaller sensor sees less of the lens' light and there's not much can be done about it.

But in a proprietary system like a p&s, everything is up for grabs by the designers. It is perfectly possible to have an APS-C sensor gather exactly the same amount of light as a FF one, but focusing all the available light from the lens onto the sensor. The sensor size does not determine how much light the lens captures, the lens does! Putting a bigger sensor behind the glass would not capture any more light if the smaller sensor was already capturing all the light!

So, given the above is true (and thinking about it, clearly it is), I wonder why the small-sensors-gather-less-light myth persists?

Sure, small sensors enable the use of smaller and cheaper lenses, which then gather less light. But that does not need to be the case. Someone could quite easily design a p&s camera witha 1/1.7" sensor that performed just as well as a A900 in terms of light gathering. Practically speaking, you'd have a massive camera though, because the lens would have to be as large as a FF system lens. So the benefit of having a very small sensor is largely eliminated and there would be little point in the design.

And there maybe other practicle issues that influence sensor performance. I am not an expert, but I would not be surprised if heat in the sensor also produces noise and a smaller sensor will be less able to dissipate heat than a larger one.

But consider this. Canon (or whoever) come out with a new p&s and everyone immediately looks to see if the sensor is 1/2.3" or 1/1.7" in size, because they think the larger sensor must gather more light. But this is just plain wrong. Maybe the smaller sensor gathers more? Who knows. It depends on how the lens has been designed, not on the sensor size.

Bizarre, but true!

You've just described the effect of putting a f/1.4 lense on a crop sensor camera vs a f/2.8 lense on a FF sensor camera. (approximately speaking, depending on crop factor) The exposure would be the same. However, that same 1.4 lense can also be mounted on the FF sensor camera. To receive the same amount of light, the cropped camera would need a f/0.7 lense. Kinda hard to come by those...

No, I haven't, but I can understand the confusion. The f number is the ratio of the entrance pupil diameter over the focal length.

Envisage two lenses with the same entrance pupil and the same focal length. But one focuses all the light perfectly onto an APS-C sensor; the other focuses all the light perfectly onto an FF sensor.

Both sensors receive the same amount of light and both lenses have the same f-stop.

I am not talking about faster lenses to achieve the same amount of light capture.

So what you're saying is it's possible to have two lenses with the same f number and focal length but producing different illumination per unit area?

My apologies, I am wrong about this. I was not thinking it through properly.

A more concentrated image (focussed onto a smaller rather than a larger sensor) must mean a shorter focal length (all other things being equal). i.e. a lower f-number.

I had mistakenly been thinking you could have the same focal length but different "magnification", but this is incorrect.

Apologies once again.

 

[JanKritzinger]

TrojMacReady wrote:

per area

facepalm

 

[TrojMacReady]

TrojMacReady wrote:

per area

facepalm

That's exactly the point. If you talk about pixel pitch, it only makes sense to compare per area in the first place. So enlighten me why that deserves a facepalm.
Ceteris paribus, pixel pitch is not a problem.