Pixel-size, noise and DR.

Started May 12, 2009 | Discussions
richardplondon
richardplondon Forum Pro • Posts: 10,879
I am amazed nobody has mentioned the O field so far

Ralf Ronander wrote:

All of reality, but more specifically, the entire dpreview forums measurement space is permeated with an O-B(x) field with a variable amplitude probability function ensuring that any given observer will encounter a B+ type event or a B- type event, whether or not looking to encounter one.

B(x) is in effect a potential that must crystallise into B1 or B2 outcomes when observed, or merely when the observer is in sufficient proximity. O is a field that exists in an implied manner at all measurement scales, and in all measurements, even if the observer thinks a type N, type C or whatever measurement is being made. N fields and C fields are merely experimental artefacts, only the O field has any ability to interact with what we think of as reality. Even a measurement with no apparent relevance to the O-B(x) field, is merely in an unknowable superposition of B+ and B- states.

In layman's terms, B+ is often termed "boost", and B- is often termed "bash". We call the O field "Olympus"

RP

 richardplondon's gear list:richardplondon's gear list
Pentax K-5 Sigma 10-20mm F4-5.6 EX DC HSM Pentax smc DA 21mm F3.2 AL Limited Pentax smc DA 70mm F2.4 AL Limited Pentax smc DA 35mm F2.8 Macro Limited +6 more
Ralf Ronander
Ralf Ronander Contributing Member • Posts: 796
Re: I am amazed nobody has mentioned the O field so far

I have no knowledge of the O field (in the digital domain; and since the world is digital ((it either exists or not == 1 or 0)) that is all that matters) so the probalility that it exists is closer to 0 than 1.

However, I´m always in search of B+ in relation to R (reality, in the hypothetical sense) but always encounter B-.

In layman´s terms, I want my picutures to look better than the scene at which my camera is pointed, but they never do.

Maybe if the detection of the photons, that might exist or not, could be less randomised...

richardplondon
richardplondon Forum Pro • Posts: 10,879
the O field

In layman´s terms, I want my picutures to look better than the scene
at which my camera is pointed, but they never do.
Maybe if the detection of the photons, that might exist or not, could
be less randomised...

I just wish my own efforts were less random!

RP

 richardplondon's gear list:richardplondon's gear list
Pentax K-5 Sigma 10-20mm F4-5.6 EX DC HSM Pentax smc DA 21mm F3.2 AL Limited Pentax smc DA 70mm F2.4 AL Limited Pentax smc DA 35mm F2.8 Macro Limited +6 more
DSPographer Senior Member • Posts: 2,464
Particles and conservation of mass-energy

I am not a physicist but your description seems to not be consistent with the law of conservation of mass-energy which I understand to mean that if one photon is emitted and thus creates an EM field then no more than one electron would be detected from that field (only extremely fleeting breaking of the conservation law is allowed). Also quantum cryptography relies on the entanglement of individual photons so in some sense it is correct to talk about the detection of specific photons even though between measurements they don't exist as individual particles. Somehow the particle part of the wave-particle duality seems to be missing from your description.

 DSPographer's gear list:DSPographer's gear list
Canon PowerShot G7 X Canon EOS 5D Mark II Canon EF 24mm f/2.8 Canon EF 50mm f/1.8 II Canon EF 200mm f/2.8L II USM +4 more
bobn2
bobn2 Forum Pro • Posts: 61,122
Re: Particles and conservation of mass-energy

DSPographer wrote:

I am not a physicist but your description seems to not be consistent
with the law of conservation of mass-energy which I understand to
mean that if one photon is emitted and thus creates an EM field then
no more than one electron would be detected from that field (only
extremely fleeting breaking of the conservation law is allowed).

I can't see where there is a violation of conservation laws there. An electron is not the same thing as a photon, and very high energy photons will indeed liberate pairs of electrons. Remember, the photons do not (probably) change into electrons, their energy releases electrons from the structure of the material, the electrons were there already, and their release leaves a 'hole' resulting in accumulation of charge.

Also
quantum cryptography relies on the entanglement of individual photons
so in some sense it is correct to talk about the detection of
specific photons even though between measurements they don't exist as
individual particles. Somehow the particle part of the wave-particle
duality seems to be missing from your description.

I haven't even begun to get my head round what quantum entanglement actually means - I'll leave that to Emil.
--
Bob

DSPographer Senior Member • Posts: 2,464
Re: Particles and conservation of mass-energy

bobn2 wrote:

DSPographer wrote:

I am not a physicist but your description seems to not be consistent
with the law of conservation of mass-energy which I understand to
mean that if one photon is emitted and thus creates an EM field then
no more than one electron would be detected from that field (only
extremely fleeting breaking of the conservation law is allowed).

I can't see where there is a violation of conservation laws there. An
electron is not the same thing as a photon, and very high energy
photons will indeed liberate pairs of electrons. Remember, the
photons do not (probably) change into electrons, their energy
releases electrons from the structure of the material, the electrons
were there already, and their release leaves a 'hole' resulting in
accumulation of charge.

I wasn't clear enough in my description. I was thinking of a sensor that needs the energy in a photon of the wavelength of visible light we are talking about in order to release an electron (the amount of energy and thus wavelength of light depends on the band-gap of the material). That energy is absorbed when the photon is detected and the number of detected photon's therefore has to obey the conservation of mass-energy.

 DSPographer's gear list:DSPographer's gear list
Canon PowerShot G7 X Canon EOS 5D Mark II Canon EF 24mm f/2.8 Canon EF 50mm f/1.8 II Canon EF 200mm f/2.8L II USM +4 more
bobn2
bobn2 Forum Pro • Posts: 61,122
Re: Particles and conservation of mass-energy

DSPographer wrote:

bobn2 wrote:

DSPographer wrote:

I am not a physicist but your description seems to not be consistent
with the law of conservation of mass-energy which I understand to
mean that if one photon is emitted and thus creates an EM field then
no more than one electron would be detected from that field (only
extremely fleeting breaking of the conservation law is allowed).

I can't see where there is a violation of conservation laws there. An
electron is not the same thing as a photon, and very high energy
photons will indeed liberate pairs of electrons. Remember, the
photons do not (probably) change into electrons, their energy
releases electrons from the structure of the material, the electrons
were there already, and their release leaves a 'hole' resulting in
accumulation of charge.

I wasn't clear enough in my description. I was thinking of a sensor
that needs the energy in a photon of the wavelength of visible light
we are talking about in order to release an electron (the amount of
energy and thus wavelength of light depends on the band-gap of the
material). That energy is absorbed when the photon is detected and
the number of detected photon's therefore has to obey the
conservation of mass-energy.

I think the mistake you're making is seeing this as necessarily a one to one process. There are three possibilities:

1. A photon has insufficient energy to release an electron, in which case it is dissipated as heat.

2. A photon has sufficient energy to release an electron. The excess over that required by the bandgap will end up being dissipated as heat.

3. A photon has sufficient energy to release an electron, but fails to interact with one, in which case its energy will end up being dissipated as heat.

The point is that the conversion is not 100% efficient, the photon energy not used releasing electrons ends up as heat, and energy is conserved.
--
Bob

-- hide signature --

Bob

ejmartin Veteran Member • Posts: 6,274
Re: Particles and conservation of mass-energy

DSPographer wrote:

I am not a physicist but your description seems to not be consistent
with the law of conservation of mass-energy which I understand to
mean that if one photon is emitted and thus creates an EM field then
no more than one electron would be detected from that field

What I wrote is entirely consistent with conservation of energy; not sure what led you to conclude otherwise. There is a basis of states for the electromagnetic field that describes it in terms of definite numbers of photons for a given wavelength. One could prepare a state of the system in this "number basis"; being states of a definite number of photons, such states do not obey Poisson statistics, nor are they localized in space or time (due to the uncertainty principle, since one has precisely specified the momentum and energy). Typical emission processes, reasonably localized in space and time, generate a superposition of "number basis" states where the probability to be in any partiular "number state" follows Poisson statistics.

Edit: The property of Poisson statistics is not inherent to the nature of light, rather it is a property of generic states of light that one encounters in the everyday world.

(only extremely fleeting breaking of the conservation law is allowed).

I would rather say that the conservation law is always valid, but that the observer doesn't have a precise notion of the energy of the system under study.

Some calculational methods (eg Rayleigh-Schrodinger perturbation theory) are structured to seem as though energy is being violated for brief periods due to interactions, but that is an artifact of using a basis of states for the system whose energy is not fixed but rather a superposition of states of different energies, and of certain approximations made in the description of the dynamics. Modern field theoretic formulations of quantum electrodynamics have energy and momentum conservation as an inherent property.

Also quantum cryptography relies on the entanglement of individual photons
so in some sense it is correct to talk about the detection of
specific photons even though between measurements they don't exist as
individual particles. Somehow the particle part of the wave-particle
duality seems to be missing from your description.

One can indeed prepare states of the electromagnetic field that are highly "non-classical", and encode subtle correlations among the particles. Quantum computing attempts to make use of such correlations. But again, typical processes that generate photons don't generate such delicate correlations (and in fact generic interactions with the environment destroy such correlations, which is why quantum computing hasn't made it to market so far). Thus it's a safe bet that unless you're inside an NSA research lab or one like it, the photons coming into your camera are incoherent (as I hope I have not been

-- hide signature --
kwik Forum Member • Posts: 84
On slits

bobn2 wrote:

There's a simple experiment that every physics undergraduate gets to
do. You take two slits and...

I'm aware of what undergraduates do with slits.

... send light through them, and get a diffraction pattern.

Er, uh, yes, of course -- diffraction patterns. Well, yeah, there is that, too. Interestingly, you don't need two slits (although it's certainly nice) -- a single slit, or hole, if you will, will do nicely. In the case of a round opening (aperture), we get diffraction patterns in the form of Airy Disks, which are responsible for the diffraction softening at small apertures.

Basically, the "mystery" of QM is that we can observe a single photon at a time, but the location of that photon is described by an interference pattern, as if the photon were actually in more than one location at once and interfering with itself, or, alternatively, vritual photons of altenative realities.

The latter interpretation is interesting to consider since we can think instead that all possible universes actually happen, and the wave function merely gives us the probability of being in a universe where the particular event we are measuring occurs.

kwik Forum Member • Posts: 84
Another elegant explanation

ejmartin wrote:

I would rather say that the conservation law is always valid, but
that the observer doesn't have a precise notion of the energy of the
system under study.

Well said!

Some calculational methods (eg Rayleigh-Schrodinger perturbation
theory) are structured to seem as though energy is being violated for
brief periods due to interactions, but that is an artifact of using a
basis of states for the system whose energy is not fixed but rather a
superposition of states of different energies, and of certain
approximations made in the description of the dynamics. Modern field
theoretic formulations of quantum electrodynamics have energy and
momentum conservation as an inherent property.

Bingo.

DSPographer Senior Member • Posts: 2,464
Re: Particles and conservation of mass-energy

ejmartin wrote:

DSPographer wrote:

I am not a physicist but your description seems to not be consistent
with the law of conservation of mass-energy which I understand to
mean that if one photon is emitted and thus creates an EM field then
no more than one electron would be detected from that field

What I wrote is entirely consistent with conservation of energy; not
sure what led you to conclude otherwise.

I went back to your previous post and I see that I did not read it carefully enough. I thought you were criticizing more of the post you were replying to than was actually the case.

There is a basis of states
for the electromagnetic field that describes it in terms of definite
numbers of photons for a given wavelength. One could prepare a state
of the system in this "number basis"; being states of a definite
number of photons, such states do not obey Poisson statistics, nor
are they localized in space or time (due to the uncertainty
principle, since one has precisely specified the momentum and
energy). Typical emission processes, reasonably localized in space
and time, generate a superposition of "number basis" states where the
probability to be in any partiular "number state" follows Poisson
statistics.

Edit: The property of Poisson statistics is not inherent to the
nature of light, rather it is a property of generic states of light
that one encounters in the everyday world.

(only extremely fleeting breaking of the conservation law is allowed).

I would rather say that the conservation law is always valid, but
that the observer doesn't have a precise notion of the energy of the
system under study.

Some calculational methods (eg Rayleigh-Schrodinger perturbation
theory) are structured to seem as though energy is being violated for
brief periods due to interactions, but that is an artifact of using a
basis of states for the system whose energy is not fixed but rather a
superposition of states of different energies, and of certain
approximations made in the description of the dynamics. Modern field
theoretic formulations of quantum electrodynamics have energy and
momentum conservation as an inherent property.

Also quantum cryptography relies on the entanglement of individual photons
so in some sense it is correct to talk about the detection of
specific photons even though between measurements they don't exist as
individual particles. Somehow the particle part of the wave-particle
duality seems to be missing from your description.

One can indeed prepare states of the electromagnetic field that are
highly "non-classical", and encode subtle correlations among the
particles. Quantum computing attempts to make use of such
correlations. But again, typical processes that generate photons
don't generate such delicate correlations (and in fact generic
interactions with the environment destroy such correlations, which is
why quantum computing hasn't made it to market so far). Thus it's a
safe bet that unless you're inside an NSA research lab or one like
it, the photons coming into your camera are incoherent (as I hope I
have not been

-- hide signature --

emil

Thanks for the explanation. In reading it I am in the position I criticize some business managers of being in where I recognize most of the terms and think I understand you but I don't have real experience with the technical details. I fear I may have misunderstandings I am not even aware of.

 DSPographer's gear list:DSPographer's gear list
Canon PowerShot G7 X Canon EOS 5D Mark II Canon EF 24mm f/2.8 Canon EF 50mm f/1.8 II Canon EF 200mm f/2.8L II USM +4 more
WahTech2 New Member • Posts: 15
Re: It's more bizarre than that!

Thanks for trying, I think some people are just not meant to understand QM and I’m one of them. I head is still spinning from reading your posts, my brain has melted into a pool of goo, I best stop before it runs out of my ears;)

Unfortunately my day job doesn’t involve me staying at any Holiday Inn...

WahTech2 New Member • Posts: 15
Re: On slits

Er, uh, yes, of course -- diffraction patterns. Well, yeah, there is
that, too. Interestingly, you don't need two slits (although it's
certainly nice) -- a single slit, or hole, if you will, will do
nicely. In the case of a round opening (aperture), we get
diffraction patterns in the form of Airy Disks, which are responsible
for the diffraction softening at small apertures.

Basically, the "mystery" of QM is that we can observe a single photon
at a time, but the location of that photon is described by an
interference pattern, as if the photon were actually in more than one
location at once and interfering with itself, or, alternatively,
vritual photons of altenative realities.

The latter interpretation is interesting to consider since we can
think instead that all possible universes actually happen, and the
wave function merely gives us the probability of being in a universe
where the particular event we are measuring occurs.

I saw a TV program once (few years back) about this, they showed that you’ll get a diffraction pattern even with a single “stream” of photons. You would think that if the photons have been bent in some way, then with a single stream of photons would all goes to one spot, but no, the same pattern show up, as if the photons know where they needed to go. Amazing.

bobn2
bobn2 Forum Pro • Posts: 61,122
Re: On slits

kwik wrote:

bobn2 wrote:

There's a simple experiment that every physics undergraduate gets to
do. You take two slits and...

I'm aware of what undergraduates do with slits.

I'm very sure I couldn't even begin to think of what you're getting at here.

... send light through them, and get a diffraction pattern.

Er, uh, yes, of course -- diffraction patterns. Well, yeah, there is
that, too. Interestingly, you don't need two slits (although it's
certainly nice) -- a single slit, or hole, if you will, will do
nicely. In the case of a round opening (aperture), we get
diffraction patterns in the form of Airy Disks, which are responsible
for the diffraction softening at small apertures.

Sure, there is a diffraction patten from a single slit. The nice point about using two is that if there's only one photon in the system, and you want to think bullets, you have to think of the bullet going through both slits at the same time. With a single slit, it's easier to convince yourself that the pattern is just a statistical summation of the bullets going through the slit, with some 'hitting' the edges. Two slits gives a good breaker of false metaphor.

-- hide signature --

Bob

richardplondon
richardplondon Forum Pro • Posts: 10,879
visualising DR

WahTech2 wrote:

richardplondon wrote:

That is why "per pixel" measurement of anything is - not meaningless

  • but, not enough.

I don’t know enough to know what else to measure DR in, other then at
per pixel level.

I keep coming back to the rain gauge analogy, which at least seems to match up reasonably well with experience. If we have an array of square rain collectors, then when one or two inches of rain fall, each one will have one or two inches of rain in it - regardless of whether it is big or small. The entire surface has a capacity to record the amount of rain that is falling, because it is entirely covered in rain collectors. Of course if different amounts of rain fall in different parts of the surface, we can record that also, so long as these differences are not too blurred by the collector size (all the water within a very large single collector would have run together, where the water within smaller collectors amounting to the same area would have been kept distinct).

The capacity of our entire rain collecting surface is - lets say - six inches. That is the height of all our individual gauges, large or small (the quantity of rainfall that will make them fill up). One might contain a maximum of one pint, another might contain a maximum of half a pint, but these volumes will correspond to the same ability to record how much rain has fallen, if the heights are the same . The volume is just an accidental sideeffect.

So we can express the top of the useful raincatching range of the surface, with (for example) how many pints it can contain per square foot - how much rainfall is it designed to store, as a surface, before you need to go round emptying all the collectors that cover it. We can express the bottom of the useful raincatching range, by saying how light a precipitation (pints per square foot) is really to be considered as measurable rain without all absorbing, evaporating etc. These are all statements about rain and about the overall surface, not about the sizes of individual collectors.

The dynamic range is the space between these limits. An unsuitable collector surface (sensor) might have 1/16th" high rain collectors, and hardly be useful in practice. Another unsuitable design might have 40" inch high collectors, and when we inspected them, we would just get "more or less empty" all the time, below our + - 5% reading accuracy. A sensibly chosen design will give us a clear indication when a likely amount of rain falls - something like "half full".

The argument is that you can accumulate DR from a lot of small
photosites, and get approximately the same result as the DR from a
large photosite. The two situations are of course physically
different - and the realities will depart from the theory in separate
ways, each with different advantages and disadvantages, in the two
cases. But the logic is sound AFAICT.

Yes, you are right. You can simply sum up a number of pixels to give
you a “larger” pixel.

This is also happening inside the photosite itself. It has a given surface area, it is not just a mathematical point - that is the entire issue right there. It gets more or less "volume" depending how big it is, but you can't separately tell what happened on each bit of that, because the electrons all mix and collect into one undifferentiated charge.

I think it's pixel binning. As far as I know it
doesn’t have much effect on DR because it needs to underexpose the
sensor to avoid clipping.

?? A larger pixel is not inherently less sensitive. It has (in principle) more opportunity to have light affect it, due to its larger area (though current sensors don't always take up all of that opportunity, in differing degrees). The same logic applies to merging and averaging readouts from a number of photosites, into one summarised value.

But the Fuji EXR sensor have very clever way expanding DR, its
basically two sensors interlaced together, to expand DR it simply
underexpose one of the sensor, so at lest half of the sensor isn’t
clipped.

Yes, there are many interesting strategies, and probably some others too that haven't yet been explored.

I doubt that any camera's (live view) exposure metering reads just a
single pixel; it will certainly average across an area of many
pixels. But even if it did take a single one, this just reports an
averaged value of what is happening across its own surface. We could
equally well have divided that into smaller subpixels, averaged their
values, and got the same answer - in principle. The reality is of
course more complex, and there are practical extremes where the
equivalence breaks down. But it's a reasonable enough approximation,
IMO, for normal purposes.

I was thinking about a hand held light meter and yes all it does is
to give you an average. I guess you can add up a number of small
pixels to give you the same value but, why? if all I want is an
average value:)

We use pixels when we want to have separate "light meters" reading each part of the view. When we don't, we don't. You could I suppose use a camera as a light meter (say, putting an incident light diffuser over the lens), but you couldn't use a lightmeter as a camera; to overstate the obvious...

RP

 richardplondon's gear list:richardplondon's gear list
Pentax K-5 Sigma 10-20mm F4-5.6 EX DC HSM Pentax smc DA 21mm F3.2 AL Limited Pentax smc DA 70mm F2.4 AL Limited Pentax smc DA 35mm F2.8 Macro Limited +6 more
kwik Forum Member • Posts: 84
Re: On slits

bobn2 wrote:

Sure, there is a diffraction patten from a single slit. The nice
point about using two is that if there's only one photon in the
system, and you want to think bullets, you have to think of the
bullet going through both slits at the same time. With a single slit,
it's easier to convince yourself that the pattern is just a
statistical summation of the bullets going through the slit, with
some 'hitting' the edges. Two slits gives a good breaker of false
metaphor.

Well, even with the single slit, and some photons "hitting the egdes", you won't get the observed diffraction pattern. But I know what you mean -- with two slits, the "weirdness" is more clearly visible.

That is, one would expect that the observed pattern of photons with two slits would simply be a superposition of the observed patterns with single slits. But, instead, we get a different diffraction pattern. Yet how can that be so? After all, we are dimming the source so that only one photon at a time gets through, so what is there to interfere with?

Even more troubling for some is that when we put a detector on one of the slits, the diffraction pattern goes away, as if the photons know they are being watched. However, that is simply resolved when one realizes that the act of detecting changes everything. That is, to "look" at a photon, we drastically affect its behavior.

So, one slit or two, what, then, is the photon interfering with? One way to "visualize" this phenomena is to think of the photon as being a wave that eventually collapses into a particle. Another way is to think of the photon as taking all possible paths and interfering with itself. Either way, we are left with the bitter taste that neither explanation is quite right, and that something else must be going on. However, as we cannot conceptualize photons as anything other than a wave or particle, we are left wanting.

So, yeah, cool stuff.

bobn2
bobn2 Forum Pro • Posts: 61,122
Re: On slits

kwik wrote:

bobn2 wrote:

Sure, there is a diffraction patten from a single slit. The nice
point about using two is that if there's only one photon in the
system, and you want to think bullets, you have to think of the
bullet going through both slits at the same time. With a single slit,
it's easier to convince yourself that the pattern is just a
statistical summation of the bullets going through the slit, with
some 'hitting' the edges. Two slits gives a good breaker of false
metaphor.

Well, even with the single slit, and some photons "hitting the
egdes", you won't get the observed diffraction pattern. But I know
what you mean -- with two slits, the "weirdness" is more clearly
visible.

It's really weird. It's like going hunting and shooting two bears with one bullet, whereas in the macroscopic world you're quite likely not even to hit one bear with a bullet, and the bear may end up interfering with you.

-- hide signature --

Bob

kwik Forum Member • Posts: 84
Re: On slits

bobn2 wrote:

It's really weird. It's like going hunting and shooting two bears
with one bullet, whereas in the macroscopic world you're quite likely
not even to hit one bear with a bullet, and the bear may end up
interfering with you.

That reminds me of a joke, "Guy goes hunting bear..." Perhaps you know it?

Ehrik Veteran Member • Posts: 8,014
LX3 pixels are 2.025µm

... according to this document from Panasonic:
http://panasonic.net/avc/lumix/popup/lx3_interview/engineering.html

Just my two oere
Erik from Sweden

igb Senior Member • Posts: 2,637
Since we're talking slits

WahTech2 wrote:

Er, uh, yes, of course -- diffraction patterns. Well, yeah, there is
that, too. Interestingly, you don't need two slits (although it's
certainly nice) -- a single slit, or hole, if you will, will do
nicely. In the case of a round opening (aperture), we get
diffraction patterns in the form of Airy Disks, which are responsible
for the diffraction softening at small apertures.

Basically, the "mystery" of QM is that we can observe a single photon
at a time, but the location of that photon is described by an
interference pattern, as if the photon were actually in more than one
location at once and interfering with itself, or, alternatively,
vritual photons of altenative realities.

The latter interpretation is interesting to consider since we can
think instead that all possible universes actually happen, and the
wave function merely gives us the probability of being in a universe
where the particular event we are measuring occurs.

I saw a TV program once (few years back) about this, they showed that
you’ll get a diffraction pattern even with a single “stream” of
photons.

I think that the appropriate term would be "squirt", of photons, of course.

-- hide signature --
Keyboard shortcuts:
FForum MMy threads