[jsherman999]

The sensor size is only loosely tied to dynamic range in that it allows for using sensors with larger pixels for the same file size, but what typically happens is that people expect the 'megapixels' to scale with the sensor size, which counteracts the benefit of larger sensors for dynamic range.

This ^ is false.

There is no causal coupling of 'pixel size' to DR. DR is a function of total light captured and sensor QE. A larger sensor (in area, not pixel size) is capable of collecting more total light with available lenses, and a sensor with huge pixels can have worse QE than one with small pixels.

It's about sensor size and efficiency, not really pixel size.

As far as using a prime lens - the type of lens really has little to no impact on dynamic range.

In has an affect in the sense that a faster lens = better DR in the image (potentially) because a larger aperture means more total light.

.

---------- Post added 07-27-15 at 09:25 PM ----------

Philbaum,

Don't they sell that approach as High Key?

Yes. Our wedding photos had a lot of it, and at times the high-key approach really worked - dramatic and interesting work.

---

Last edited by jsherman999; 07-27-2015 at 08:21 PM.

[Ikarus]

There is no causal coupling of 'pixel size' to DR. DR is a function of total light captured ...

Those two statements are mutually contradictory. Since DR is a function of the total light captured, the number of photons collected by a sensor element increases with the size of said element, thus increasing its dynamic range (all else considered equal).

... and sensor QE.

That I agree with.

[Ikarus]

In has an affect in the sense that a faster lens = better DR in the image (potentially) because a larger aperture means more total light.

I have a hard time believing that anybody selects aperture based on dynamic range considerations, so for a particular AOV and DOF, the size of the entrance pupil is given, and hence the total amount of light, which means that dynamic range is not a function of the lens optics (within reasonable limits, i.e. barring local effects from strong vignetting).

[Fogel70]

Those two statements are mutually contradictory. Since DR is a function of the total light captured, the number of photons collected by a sensor element increases with the size of said element, thus increasing its dynamic range (all else considered equal).

Yes, but a larger sensor might also have smaller pixel size than a smaller sensor. Pixel size itself is not connected to DR of the sensor.

---------- Post added 28-07-15 at 10:54 ----------

I have a hard time believing that anybody selects aperture based on dynamic range considerations, so for a particular AOV and DOF, the size of the entrance pupil is given, and hence the total amount of light, which means that dynamic range is not a function of the lens optics (within reasonable limits, i.e. barring local effects from strong vignetting).

If you use a fast prime lens or a slow zoom lens you might prefer to use the larger apertures on the prime lens in low light conditions.
Faster lenses has always primarily been about getting better IQ in low light conditions, where you might prefer to use narrower DOF instead of higher ISO.

---

Last edited by Fogel70; 07-28-2015 at 02:07 AM.

[falconeye]

1.There is no causal coupling of 'pixel size' to DR.
2. DR is a function of total light captured and sensor QE.

1. Correct
2. False, DR is no function of QE

Those two statements are mutually contradictory. Since DR is a function of the total light captured, the number of photons collected by a sensor element increases with the size of said element, thus increasing its dynamic range (all else considered equal).

Wrong, DR is defined for the camera, not the pixel as DR is a function of spatial frequency.

Statements like the ones I cited are an example why I so strongly advocate the equivalence theorem. And it is a theorem really, not a simple conversion rule. Because it holds true in a strict physical sense (and its proof is non trivial) and it allows to understand things like the above WITHOUT the need to dig into physical considerations. Not every photographer is also a physicist ...

In a nutshell:
Eg. a mFT sensor with crop 2 and base ISO 100 has a 35mm-equivalent base ISO of 400 (100 *2^2). And at ISO 400, DR is 2EV less than at ISO 100. It is this simple, really. Just don't try to dig deeper, no need for it. And #pixels or pixel size plays no role at it, don't always pixel peep, not even when thinking about photography.

BTW, note that an ideal camera has DR(ISO,f) which is a known function (the equivalence theorem is 100% strict for ideal cameras only). Here, ISO denotes 35mm-equivalent ISO, f spatial frequency (LW/PH), and an ideal camera has noise=shot noise, QE=1 and base ISO of 0. A real camera's DR will remain below this known function. A lower QE of a real camera would shift the function to lower ISO, including the same shift for its base ISO, yielding in an unaltered DR at base ISO. The DR function is shifted left by QE<1, not up or down. A real camera's non-equivalent base ISO and QE are determined by the CMOS process rules, color discrimination technology and microlens design. Because the latter are technology-driven more than camera-model dependent, equivalence nicely applies to real cameras too. And of course, because real cameras are approaching the ideal camera more and more.

At this point, let me add one more note ... The ideal camera would score infinite DxOmarks because of its zero base ISO. Maybe, it is time for DxO to renormalize their scale such that an ideal camera scores finite.

[mohb]

n a nutshell:
Eg. an FT sensor with crop 2 and base ISO 100 has a 35mm-equivalent base ISO of 400 (100 *2^2). And at ISO 400, DR is 2EV less than at ISO 100. It is this simple, really. Just don't try to dig deeper, no need for it. And #pixels or pixel size plays no role at it, don't always pixel peep, not even when thinking about photography.

BTW, note that an ideal camera has DR(ISO,f) which is a known function (the equivalence theorem is 100% strict for ideal cameras only). Here, ISO denotes 35mm-equivalent ISO, f spatial frequency (LW/PH), and an ideal camera has noise=shot noise, QE=1 and base ISO of 0. A real camera's DR will remain below this known function. A lower QE of a real camera would shift the function to lower ISO, including the same shift for its base ISO, yielding in an unaltered DR at base ISO. The DR function is shifted left by QE<1, not up or down. A real camera's non-equivalent base ISO and QE are determined by the CMOS process rules, color discrimination technology and microlens design.

I am so glad that taking photographs doesn't include the need to understand this sort of thing.

[kh1234567890]

Statements like the ones I cited are an example why I so strongly advocate the equivalence theorem. And it is a theorem really, not a simple conversion rule. Because it holds true in a strict physical sense (and its proof is non trivial) and it allows to understand things like the above WITHOUT the need to dig into physical considerations. Not every photographer is also a physicist ...

I used to do physics and other science and engineering stuff for a living. Statements like this set off my bullshit detector ringing. References to this 'non trivial' proof please ?

[Ratcheteer]

Since DR is a function of the total light captured

should we really refer to it as a function when by definition its just a measurement?

ok i might be wrong in saying this but dynamic range sounds more like exposure latitude from the days of film (if it is exposure latitude, when and why did we start calling it dynamic range? marketing reasons maybe? so they can put a h infront of it when its abbreviated and make it sound epic? :P)

[mohb]

dynamic range sounds more like exposure latitude from the days of film (if it is exposure latitude, when and why did we start calling it dynamic range?

I sometimes wonder how those forever calling for increased dynamic range and high ISO would have coped with slide film where 1/2 a stop either way was crucial and ISO 100 was fast

[Parallax]

I sometimes wonder how those forever calling for increased dynamic range and high ISO would have coped with slide film where 1/2 a stop either way was crucial and ISO 100 was fast

They would have coped by arguing equivalence factors for 35mm Kodachrome and 120 Kodacolor.

[Ikarus]

Wrong, DR is defined for the camera, not the pixel as DR is a function of spatial frequency. Statements like the ones I cited are an example why I so strongly advocate the equivalence theorem.

It's a bit ironic that I find myself in disagreement with some of the equivalentalists on this, since I consider myself as one. I don't profess to be an imaging technology expert, so I'm referring to those who are with the article "How Small Should Pixel Size Be?", Proc. SPIE 3965, Sensors and Camera Systems for Scientific, Industrial, and Digital Photography Applications, 451 (May 15, 2000); doi:10.1117/12.385463" by T. Chen et al., available online here:

Increasing pixel size improves the sensor by increasing dynamic range and signal-to-noise ratio. [...] DR increases roughly as the square root of pixel size, since both C and reset noise (kTC) increase approximately linearly with pixel size.

For the math, see equation (1) and for a plot, see figure 2a.

[Ratcheteer]

I sometimes wonder how those forever calling for increased dynamic range and high ISO would have coped with slide film where 1/2 a stop either way was crucial and ISO 100 was fast

you would be surprised at how many people with dslr's who have never touched a film camera (they are usually the ones asking for high iso)

the people asking for higher dynamic range are usually people that have shot with slide film before

They would have coped by arguing equivalence factors for 35mm Kodachrome and 120 Kodacolor.

actually they argued which manufacture made the best film, papers, chemicals blah blah blah etc

[falconeye]

Statements like this set off my bullshit detector ringing. References to this 'non trivial' proof please ?

I provide one article containing a glimpse of a proof on my site, there are a few more elsewhere. What makes the proof non trivial is that one has to cope with all aspects of an image which includes, e.g., SNR, diffraction effects and possible aberrations. Not only DoF and FoV. Moreover, equivalence needs to introduce the concept of the "effective crop factor" when dealing with finite magnifications.

I consider all this to be beyond the scope of this thread and will refrain from referring to a proof or discussing it. If required, leave a comment on my corresponding blog article.

It's a bit ironic that I find myself in disagreement with some of the equivalentalists on this, since I consider myself as one.

It wasn't my intention to argue with anybody here.

[Ikarus]

It wasn't my intention to argue with anybody here.

I must say you have a rather interesting way of not arguing ("correct, false, wrong"). I would still like to see you respond to the article I was referring to in support of the notion that dynamic range does in fact decrease with pixel size, which, correct me if I'm wrong, is in direct contradiction to what you were claiming above.

[Ratcheteer]

I must say you have a rather interesting way of not arguing ("correct, false, wrong"). I would still like to see you respond to the article I was referring to in support of the notion that dynamic range does in fact decrease with pixel size, which, correct me if I'm wrong, is in direct contradiction to what you were claiming above.

that article needs updating cmos technology has improved quite a bit in 14 years