[bxf]

I didn't want to further confuse any of the countless existing threads about FF vs APS-C, so I'm starting a new one, hoping to keep it simple.

For the purpose of this post, I am not interested in equivalence. I am not interested DOF. I have two questions that I believe will be understood by just about anyone who has been reading this forum:

1. Total Light. Since it has been stated quite emphatically (even by DPR) that Total Light has significant impact on IQ, can someone explain how a given fixed* quantity of light can affect the "non-cropped area" of an FF sensor? (I'm not elaborating here because I believe readers will know what I'm asking). If this Total Light argument is correct, then there must be a good explanation, even if it would take a physicist to explain.

Common sense, of course says that this Total Light concept cannot be right. Therefore...

2. Why, for any given level of technology, do FF sensors consistently show better high ISO performance (i.e. less noise)?

In another thread, Bossa suggested that this is merely a result of greater magnification applied to the APS-C image when performing the comparisons. This makes sense to me, and yet nobody has commented on this suggestion, the inevitable inference being that it has no merit.

Common sense does not always provide the correct answer. So can anyone answer the above, without getting into arguments as to whether or not equivalence exists/is an issue/blah blah/etc?

Bill

* Fixed for any given aperture setting on any given lens

[alamo5000]

1. Total Light. Since it has been stated quite emphatically (even by DPR) that Total Light has significant impact on IQ, can someone explain how a given fixed* quantity of light can affect the "non-cropped area" of an FF sensor? (I'm not elaborating here because I believe readers will know what I'm asking). If this Total Light argument is correct, then there must be a good explanation, even if it would take a physicist to explain.

I don't understand your question.

2. Why, for any given level of technology, do FF sensors consistently show better high ISO performance (i.e. less noise)?

My understanding is that this has to do with pixel density and sheer sensor real estate as well as how the light receptors on the sensor are built in.

I am sure that scientists could pack 36mp onto a sensor the size of your thumbnail but what that means is the little light receptors will each individually have to be smaller which degrades their ISO performance. The smaller and more compact and smashed in together the more noise you get because they become less sensitive to light.

So if you have 36mp spread out over a bigger sensor you have bigger more spaced out 'light recievers' than would be if you smooshed them all into a small sensor.

[bxf]

I don't understand your question. ...

I'm questioning the concept of "more total light". People more knowledgeable than I have been saying that a FF sensor yields lower noise because there is a greater amount of light falling on the sensor as a whole. Well of course there is, because the sensor is larger. But if you consider the light hitting only that part of an FF sensor that would be taken by an APS-C sensor, one wouldn't expect that to change just because there is "more sensor" beyond this APS-C area.

[alamo5000]

The same answer I gave above about pixel density and light receptors is still applicable to both answers.

If you google and read up on how sensors are made and how they work it will help.

I had a similar question a while back and a few people kind of got me started and then I read up on it until I was satisfied.

So in effect if you took say lens X and put it on a FF camera in the exact same conditions and f stop... then you put it on a crop camera in those exact same conditions... you do not gain 'total light'. The light density does not change. The T stop of the lens is the T stop of the lens.

Of course you have a bigger sensor but the overall light density is the same across the corresponding areas of both sensors.

The variable is the camera's sensor and it's ability of absorb the light at that density of light.

Long story short smaller light receptors that are scrunched in together make the sensor marginally less ISO sensitive than one of a larger size with bigger and un-scrunched light receptors.

It's not a sheer game of mega pixels. There are other factors involved.

---

Last edited by alamo5000; 10-12-2014 at 07:46 PM.

[Pavel_Zhelev]

I'm questioning the concept of "more total light". People more knowledgeable than I have been saying that a FF sensor yields lower noise because there is a greater amount of light falling on the sensor as a whole. Well of course there is, because the sensor is larger. But if you consider the light hitting only that part of an FF sensor that would be taken by an APS-C sensor, one wouldn't expect that to change just because there is "more sensor" beyond this APS-C area.

Hi BFX,
I have spend fair amount of time working with the PMTs (photomultiplier tubes) which are basically devices measuring light. Now having said that as per my understanding is that there is not such a thing as "total light" there is a measure called "amount of photons", and basically it is largely dependent on the surface area of the light detecting device. By surface area I mean not the total area of the sensor, but the area of the sensor units (pixels), as large is the area of the pixel the dependency is on square for the amount of collected photons . This is why not so big difference in size between FF and APS C is translated in to relatively big difference in sensitivity (supporting alamo's explanation). So basically your statement above is not right and reason is generally the more pixels squeezed in smaller area, but not because of the general are, but rather because of the area taken from each pixel.
Back in the day Fuji film made a super CCD with hexagonal pixel allowing bigger surface area pixels on smaller size sensor, this significantly improver their sensitivity.
I hope this helps.

[Fogel70]

The "total light" and "magnification" are actually different answers to the same question. To capture more light using same tech you need to add more light capturing area. The magnification is only important if comparing small portion of images (pixel peeping).

You could print a FF images with much greater magnification (size of a football field) than an APS-C image (size of a stamp) both using same sensor tech, but the FF image will still contain less per picture noise than the APS-C image.

Then there is different ways of capturing more total light. You could use:
- bigger sensor with larger image capturing area
- longer exposure (lower ISO)
- bigger aperture (lower ISO)
- add light with a flash
- wait a few hours for the sun to rise
....

[calsan]

One reason may be that the larger sensor sizes and larger camera bodies could (maybe) better dissipate heat. Heat in the sensor is what causes noise, not "lack of light". The Ricoh GR is actually slightly more noisy than the K5 with the same sensor, simply because the camera body is smaller. Astronomers prove this by having cooling systems for their sensors and take pretty much noise free photos at night time - i.e. the very definition of when there's a lack of light.

Anyway, the year the sensor was designed and made (technological advances) matters more than sensor size and resolution. So the trade off in favour of APS cameras is that for the same amount of money you could upgrade frequently and catch up with FF every second or third year. (whilst gaining other tech improvements...and saving weight)
My K10d has lower resolution but significantly more noise than my K7 which has lower resolution but significantly more noise than my K3. All obviously have the same sensor size. They also disprove the myth that higher resolutions automatically equates to more noise.

---

Last edited by calsan; 10-12-2014 at 10:33 PM.

[Rondec]

I guess the K5 and D800 have roughly the same sensor tech. Really similar dynamic range at base iso. The D800 has a lot bigger sensor, though. If you look at per pixel noise they are basically the same. However, when you "normalize" them (DXO Mark uses 8 megapixels), then you see that the D800 jumps forward. Clearly, this is a magnification issue -- the K5 is magnified quite a bit more for similar size prints and you just see individual pixels quicker.

The whole "tech" aspect of things tends to obfuscate the issue. Technology does improve over time and obviously there is a big jump forward from the K10 to the K5. But if you compare sensors of similar generation, you will find a similar gap in performance.

With regard to per pixel noise, it is not particularly important when it comes to output and in fact, smaller pixels tend to give noise that is either less noticeable or easier to clean up, even if the overall SNR for the sensor is the same (eg, K5 versus K3).

[dakight]

Two factors are involved. One is heat as was mentioned above. When the pixels are more tightly packed in a smaller sensor the heat generated by the sensor is not dissipated as efficiently. More heat = more noise. The other factor is electromagnetic coupling among adjacent pixels; the closer together they are the more effect they will have on one another. The higher the iso setting the more pronounced the effects become. Both of these factors can be addressed in the design of the sensor circuits but there are physical limits. All things equal, a larger sensor will produce a cleaner picture, especially at higher amplification (iso).

[bxf]

The same answer I gave above about pixel density and light receptors is still applicable to both answers.

If you google and read up on how sensors are made and how they work it will help.

I had a similar question a while back and a few people kind of got me started and then I read up on it until I was satisfied.

So in effect if you took say lens X and put it on a FF camera in the exact same conditions and f stop... then you put it on a crop camera in those exact same conditions... you do not gain 'total light'. The light density does not change. The T stop of the lens is the T stop of the lens.

Of course you have a bigger sensor but the overall light density is the same across the corresponding areas of both sensors.

The variable is the camera's sensor and it's ability of absorb the light at that density of light.

Long story short smaller light receptors that are scrunched in together make the sensor marginally less ISO sensitive than one of a larger size with bigger and un-scrunched light receptors.

It's not a sheer game of mega pixels. There are other factors involved.

As you say, "you do not gain total light", and yet we keep on being told that this is a (if not "the") primary reason for the superiority of the FF sensors.

Then, whereas I am well aware of the concept of higher noise due to higher pixel density, I am under the impression that with more recent sensor technologies, this is no longer quoted as a reason for the FF sensor superiority. Perhaps I am wrong here.

---------- Post added 13-10-14 at 12:57 ----------

Hi BFX,
I have spend fair amount of time working with the PMTs (photomultiplier tubes) which are basically devices measuring light. Now having said that as per my understanding is that there is not such a thing as "total light" there is a measure called "amount of photons", and basically it is largely dependent on the surface area of the light detecting device. By surface area I mean not the total area of the sensor, but the area of the sensor units (pixels), as large is the area of the pixel the dependency is on square for the amount of collected photons . This is why not so big difference in size between FF and APS C is translated in to relatively big difference in sensitivity (supporting alamo's explanation). So basically your statement above is not right and reason is generally the more pixels squeezed in smaller area, but not because of the general are, but rather because of the area taken from each pixel.
Back in the day Fuji film made a super CCD with hexagonal pixel allowing bigger surface area pixels on smaller size sensor, this significantly improver their sensitivity.
I hope this helps.

But if you take FF and APS-C sensors of equal pixel densities, the APS-C portion of the FF sensor would not have any greater "sensor unit area" than the APS-C sensor, yet it would have better noise performance.

---------- Post added 13-10-14 at 13:06 ----------

The "total light" and "magnification" are actually different answers to the same question. To capture more light using same tech you need to add more light capturing area. The magnification is only important if comparing small portion of images (pixel peeping).

You could print a FF images with much greater magnification (size of a football field) than an APS-C image (size of a stamp) both using same sensor tech, but the FF image will still contain less per picture noise than the APS-C image.

Then there is different ways of capturing more total light. You could use:
- bigger sensor with larger image capturing area
- longer exposure (lower ISO)
- bigger aperture (lower ISO)
- add light with a flash
- wait a few hours for the sun to rise
....

Thanks, but I'm afraid none of the above manage to explain why a part of a FF sensor equal to the size of an APS-C sensor "gets more light" than an APS-C sensor of the same pixel density.

[bxf]

One reason may be that the larger sensor sizes and larger camera bodies could (maybe) better dissipate heat. Heat in the sensor is what causes noise, not "lack of light". The Ricoh GR is actually slightly more noisy than the K5 with the same sensor, simply because the camera body is smaller. Astronomers prove this by having cooling systems for their sensors and take pretty much noise free photos at night time - i.e. the very definition of when there's a lack of light.

Interesting idea. If this is correct, then I'd expect some electronics engineer out there to be able to confirm it. But even so, it has nothing to do with the concept of "more total light".

Anyway, the year the sensor was designed and made (technological advances) matters more than sensor size and resolution. So the trade off in favour of APS cameras is that for the same amount of money you could upgrade frequently and catch up with FF every second or third year. (whilst gaining other tech improvements...and saving weight)
My K10d has lower resolution but significantly more noise than my K7 which has lower resolution but significantly more noise than my K3. All obviously have the same sensor size. They also disprove the myth that higher resolutions automatically equates to more noise.

I don't believe this is correct, though I can't confirm right now. I suspect that DXO would show high ISO noise figures for older FF sensors still significantly better than current APS-C ones.

[Fogel70]

Thanks, but I'm afraid none of the above manage to explain why a part of a FF sensor equal to the size of an APS-C sensor "gets more light" than an APS-C sensor of the same pixel density.

It doesn't. If using a APS-C portion of a FF sensor you will get the same performance as on a APS-C sensor using the same sensor tech.
Cropping will affect total amount of noise in images as total amount of light used is affected. FI using 16MP of a 36MP FF sensor will make the noise performance in the image drop with a bit over 1 stop compared to using the full sensor area.

Cropping works the same way as using different sized sensors.

[bxf]

I guess the K5 and D800 have roughly the same sensor tech. Really similar dynamic range at base iso. The D800 has a lot bigger sensor, though. If you look at per pixel noise they are basically the same. However, when you "normalize" them (DXO Mark uses 8 megapixels), then you see that the D800 jumps forward. Clearly, this is a magnification issue -- the K5 is magnified quite a bit more for similar size prints and you just see individual pixels quicker.

The whole "tech" aspect of things tends to obfuscate the issue. Technology does improve over time and obviously there is a big jump forward from the K10 to the K5. But if you compare sensors of similar generation, you will find a similar gap in performance.

With regard to per pixel noise, it is not particularly important when it comes to output and in fact, smaller pixels tend to give noise that is either less noticeable or easier to clean up, even if the overall SNR for the sensor is the same (eg, K5 versus K3).

All this makes sense to me (meaning I find it logical).

So if this is the correct explanation, are we saying that statements pertaining to FF superiority due to "more total light", including that made by DPR, are plainly wrong?

---------- Post added 13-10-14 at 13:21 ----------

Two factors are involved. One is heat as was mentioned above. When the pixels are more tightly packed in a smaller sensor the heat generated by the sensor is not dissipated as efficiently. More heat = more noise. The other factor is electromagnetic coupling among adjacent pixels; the closer together they are the more effect they will have on one another. The higher the iso setting the more pronounced the effects become. Both of these factors can be addressed in the design of the sensor circuits but there are physical limits. All things equal, a larger sensor will produce a cleaner picture, especially at higher amplification (iso).

But these differences do not exist when the sensors are of the pixel density, yet the FF sensor has better high ISO performance.

---------- Post added 13-10-14 at 13:30 ----------

It doesn't. If using a APS-C portion of a FF sensor you will get the same performance as on a APS-C sensor using the same sensor tech.
Cropping will affect total amount of noise in images as total amount of light used is affected. FI using 16MP of a 36MP FF sensor will make the noise performance in the image drop with a bit over 1 stop compared to using the full sensor area.

Cropping works the same way as using different sized sensors.

But this total amount of light is spread over a larger area. The APS-C-equivalent-area of the FF sensor gets the same amount of light as would an APS-C sensor. So why the difference in high ISO performance?

[Fogel70]

But this total amount of light is spread over a larger area.

Yes, but it's still more light. It's the total amount of light that is important.
You can also capture more light by using smaller sensor size using longer exposure or using larger aperture to achieve the same thing.
FI if using ISO6400 on a FF camera you capture the same amount of light by using ISO2800 on a APS-C. Then noise will be equal on both images (if both use the same sensor tech).

The APS-C-equivalent-area of the FF sensor gets the same amount of light as would an APS-C sensor. So why the difference in high ISO performance?

When both capture the same amount of light they will have equal ISO performance. A image from a FF sensor cropped to APS-C size will have same ISO performance as a APS-C sensor image. (if using same sensor tech).

[Stone G.]

A sensor is not like a rather "inert" solar panel, producing heat or electrons in proportion to its total area under a given intensity of the incident sun light.

In a photographic system, it is the lens that captures the light and focuses it onto the sensor.

So, take a 50mm f/2.0 lens on an APS-C and a 75mm f/2.8 lens on an FF and you will have two comparable system (lens + sensor) that have a) the same field of view and b) very nearly the same absolute aperture = area of the entrance pupil. Thus, it is clear that in equal spans of time, these two systems will collect the same amount of light (number of photons). And if the pixel count is the same for both sensors, each pixel will also be hit (on average) by the same number of photons.

But the area of the APS-C sensor is only half (1/2.25 to be exact) of that of the FF-sensor so the number of photons per unit area hitting the APS-C will be twice that of the FF. So my intuition from the film days tells me to either A) reduce the ISO of my APS-C by a factor of 2(.25) or B) reduce my exposure by the same factor.

In situation A) we have an example of two equivalent cameras highlighting that one has to take all factors into account - crop factor, pixel count, true focal lenght, absolute aperture/entrance pupil and sensor sensitivity - when one talks about camera equvalence. In in situation A) I *should* get indistinguisable images when I view them on screen or print them to the same size - at least, that's what the theory says.

But let's turn to the alternative Situation B then and see if there might be some indications of answers to the O.P.s questions to be found here:

Answer to Question 1
In situation B) my APS-C sensor obviously only get hit by half the number of photons - because I reduce the expore by a factor of two - and thus, also half the amount of photons hits any corresponding pixel/photodiode and in anwer to the O.P.'s question 1 I belive that is why it is said that an FF sensor "captures more light" - but do remember, that *correct* exposure really depends upon the lens and sensor senitivity combined!

Answer to Question 2
Photon capture (conversion into electrons) is not a 100% efficient process, there is only a certain probability - quantum effeciency - that whenever a photon hits a photodiode it will also be converted into an electron. Further, the distribution of light/photons hitting the sensor is also a statistical process and finally, all electrical devices produce a minimum of noise only dependent upon the temperature. Thus, when you have fewer photon counts to work with - as you have in situation B - then the statistical nature of the photon capture process will leade to a poorer signal-to-noise ratio.

One may continue playing with other lens/aperture/sensor combinations. Take for example the same lens (same true focal length) at the same aperture on an FF and an APS-C. In this case, the total light captured and focused on the APS-C will be half of that focused on the FF, but the amount of light captured by the APS-C will be exactly the same as captured by the corresponding part of the FF covering the same part of the scene, and it does not make sense (to me at least) to talk about "the total light capturing capability" in comparisons between such two (non-equivalent) systems.

Again, sensor size, sensor sensitivity, pixel count, lens focal length, lens absolute aperture - you can't discuss one and leave the others out of the equation.