"Water is not light" contd. :-)

[boardsy]

Can we clear this up? Previous thread is full (i.e. rose by more than one inch!).

> Photozopia said: In reply to RGBaker, 1 hour ago
> Your reply - like others - merely proves you don't get the relationship between standardised unit measurement and overall exposure by area (hence differing volume). As I said, it's an easy test for anyone with a sprinkler/spray hose .... unequal vessels do not record a linear inch measure .... regardless of size.

[me:] It seems intuitive though, that the water level will rise at the same rate in two different-sized trays under a sprinkler or the same rainfall. How do you know your sprinkler was spraying uniform volume per area i.e. equally over each container?

> You are getting mired in unit rainfall v surface area collection and differently sized vessels.If it rains all night, should I put a teacup under the drainpipe to catch the 'inch of water' that fell - or a couple of fifty gallon tanks? according to you all vessels record an equal 1" linear rise so the cup will suffice.

[me:] Hang on now Photozopia, you are now catching in a teacup via drainpipe the total volume of water from the roof, not measuring a vertical rise (ie. if the roof had walls and no drainpipe it would record one inch)!

but can anyone explain why a standard rainfall gauge

1) has a funnel on top

2) has to be standardised in size (area and volume), if any/every vertical-walled vessel records the same vertical rise per rainfall?

 

[RGBaker]

boardsy wrote:
Can we clear this up? Previous thread is full (i.e. rose by more than one inch!).

> Photozopia said: In reply to RGBaker, 1 hour ago
> Your reply - like others - merely proves you don't get the relationship between standardised unit measurement and overall exposure by area (hence differing volume). As I said, it's an easy test for anyone with a sprinkler/spray hose .... unequal vessels do not record a linear inch measure .... regardless of size.

[me:] It seems intuitive though, that the water level will rise at the same rate in two different-sized trays under a sprinkler or the same rainfall. How do you know your sprinkler was spraying uniform volume per area i.e. equally over each container?

> You are getting mired in unit rainfall v surface area collection and differently sized vessels.If it rains all night, should I put a teacup under the drainpipe to catch the 'inch of water' that fell - or a couple of fifty gallon tanks? according to you all vessels record an equal 1" linear rise so the cup will suffice.

[me:] Hang on now Photozopia, you are now catching in a teacup via drainpipe the total volume of water from the roof, not measuring a vertical rise (ie. if the roof had walls and no drainpipe it would record one inch)!

but can anyone explain why a standard rainfall gauge

1) has a funnel on top

2) has to be standardised in size (area and volume), if any/every vertical-walled vessel records the same vertical rise per rainfall?
1) For convenience in measuring small amounts -- the collection area is (for example) 10x as big as the accumulation area, so the rise is magnified 10 fold. Each inch of rainfall is presented as 10" of column, so it is easier to note small increments. Note that the calibration corrects for the apparent length, so every 1/10 of an inch in my example would record 1" of rainfall.

2) There is no standardization -- you can use any straightsided device to measure the rainfall -- a coffee can would do just fine -- because the result is the density of rain, that is how many inches fall over any and every area ... so a barrel, or coffee can, or what have you will deliver the same result, as long as it is straightsided. There is a recommended standard size so that very small amounts of rain will be recorded in a consistent manner -- but there is no requirement that some particular device be used.

Note that I never made the silly claims attributed to me earlier -- a tea cup will not catch the same volume of rain as a rain barrel, or a swimming pool -- but any of them can be used to measure the density of the rainfall, which is what is reported. Rainfall, like light, is measured as a density, not a volume.

Cheers,
GB

 

[nevercat]

boardsy wrote:
Can we clear this up? Previous thread is full (i.e. rose by more than one inch!).

> Photozopia said: In reply to RGBaker, 1 hour ago
> Your reply - like others - merely proves you don't get the relationship between standardised unit measurement and overall exposure by area (hence differing volume). As I said, it's an easy test for anyone with a sprinkler/spray hose .... unequal vessels do not record a linear inch measure .... regardless of size.

[me:] It seems intuitive though, that the water level will rise at the same rate in two different-sized trays under a sprinkler or the same rainfall. How do you know your sprinkler was spraying uniform volume per area i.e. equally over each container?

> You are getting mired in unit rainfall v surface area collection and differently sized vessels.If it rains all night, should I put a teacup under the drainpipe to catch the 'inch of water' that fell - or a couple of fifty gallon tanks? according to you all vessels record an equal 1" linear rise so the cup will suffice.

[me:] Hang on now Photozopia, you are now catching in a teacup via drainpipe the total volume of water from the roof, not measuring a vertical rise (ie. if the roof had walls and no drainpipe it would record one inch)!

but can anyone explain why a standard rainfall gauge

1) has a funnel on top

2) has to be standardised in size (area and volume), if any/every vertical-walled vessel records the same vertical rise per rainfall?

 

[SQLGuy]

Regarding the rainfall meter, if you want it to be one inch high for each inch its recording, then it doesn't need a funnel, but will make readings of small amounts of rainfall difficult. The funnel amplifies the vertical scale proportionally to it's opening size compared to the area of the opening of the measurement tube. For instance, if the measurement tube is 1 square inch, and the funnel opening is 10 square inches, then a 1/4" of rainfall will raise the level in the tube 2.5" - much easier to read.

For the exposure thing, one of the things I pointed out, which went ignored, was that one way you could look at it is that it takes energy to expose a silver halide crystal, or a pixel. If your frame size is double the surface area, it will collect twice as much light = twice as much energy, but it will also have twice the photosites to expose, which requires twice as much energy, so the whole thing balances out. With CCDs or CMOS, those pixels are acting just like silver halide crystals with respect to how much energy it takes to stimulate them.

 

[Pal2012]

Its all just calibration, same as for viscosity, Zahn #2, 4 or a Ford cup, etc all same purpose but different area and orifice but calibrated to reveal measured flow of material in relation to time.

 

[wcdennis]

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space. So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

 

[Tom2572]

wcdennis wrote:

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space. So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

Yep, I'm the dumb*** who thought that full frame sensors took cleaner low light photos than crop frame sensors at a given aperture. Boy was I mistaken. Glad you guys proved me wrong by pointing out the sham regarding full frame cameras.

Again, thank you for proving that the low light capabilities of cameras like the Nikon D3s, Sony A99 and Canon EOS 1Dx are only imaginary and have nothing to do with the size of their sensors. I suppose the real loser in all this are the owners of these impostor cameras, and I truly feel sorry for those poor saps who spent such good money on them.

You really should get the word out by hopping on the Canon and Nikon full frame forums and informing them of your discovery and the error of their ways.

 

[twald]

To keep up with the rainfall analogy, camera sensors have a little funnel (albeit one that funnels light) called microlens over every photosite for the very reason you mentioned; the sensor houses circuitry in addition to the photosites. Therefore, it does not matter what percentage of the sensor is covered by photosites, so long as the entire sensor is covered by microlenses and all light collected by each microlens is directed to its corresponding photosite.

 

[Bart Hickman]

SQLGuy wrote:

Regarding the rainfall meter, if you want it to be one inch high for each inch its recording, then it doesn't need a funnel, but will make readings of small amounts of rainfall difficult. The funnel amplifies the vertical scale proportionally to it's opening size compared to the area of the opening of the measurement tube. For instance, if the measurement tube is 1 square inch, and the funnel opening is 10 square inches, then a 1/4" of rainfall will raise the level in the tube 2.5" - much easier to read.

For the exposure thing, one of the things I pointed out, which went ignored, was that one way you could look at it is that it takes energy to expose a silver halide crystal, or a pixel. If your frame size is double the surface area, it will collect twice as much light = twice as much energy, but it will also have twice the photosites to expose, which requires twice as much energy, so the whole thing balances out. With CCDs or CMOS, those pixels are acting just like silver halide crystals with respect to how much energy it takes to stimulate them.

Not sure what you mean by "balancing out". Conservation of information requires that the same amount of energy on the same number of photo sites creates the same exact image regardless of surface area. The only difference is the larger surface area can handle a larger amount of total energy before over-flowing and thus is capable of larger SNR.

If you instead choose to increase the number of photo sites in proportion to the surface area (keep photo site size constant), then the larger surface area actually gathers more information with the same amount of energy.

Bart

 

[SQLGuy]

Tom2572 wrote:

wcdennis wrote:

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space. So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

Yep, I'm the dumb*** who thought that full frame sensors took cleaner low light photos than crop frame sensors at a given aperture. Boy was I mistaken. Glad you guys proved me wrong by pointing out the sham regarding full frame cameras.

Again, thank you for proving that the low light capabilities of cameras like the Nikon D3s, Sony A99 and Canon EOS 1Dx are only imaginary and have nothing to do with the size of their sensors. I suppose the real loser in all this are the owners of these impostor cameras, and I truly feel sorry for those poor saps who spent such good money on them.

You really should get the word out by hopping on the Canon and Nikon full frame forums and informing them of your discovery and the error of their ways.

And I'm going to tell you again, for what it's worth, that pixels don't know or care how many neighbors they have.

As others have amazingly patiently attempted to explain to you, a bigger pixel (all other things being equal) will be more sensitive to light, just as a bigger silver halide crystal will be more sensitive to light.

If all you're comparing is sensor size, you have your head in the sand for some reason.

One more example, based on one of the cameras you mentioned... maybe this will help: If you install a DX lens on a Nikon D3s, it will crop to the APS-C portion of its full frame sensor. Now, tell me, do you really think the exposure is going to be any different for the DX 35/1.8 @ 1.8 on a D3s, versus an FX 50/1.8 @ 1.8 on the SAME camera? The resolution will be different, the field of view will be the same, the pixel-by-pixel noise will be the same, and the exposure will be the same, even though the 35DX shot was on a smaller sensor.

 

[blue_skies]

Tom2572 wrote:

wcdennis wrote:

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space. So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

Yep, I'm the dumb*** who thought that full frame sensors took cleaner low light photos than crop frame sensors at a given aperture. Boy was I mistaken. Glad you guys proved me wrong by pointing out the sham regarding full frame cameras.

Again, thank you for proving that the low light capabilities of cameras like the Nikon D3s, Sony A99 and Canon EOS 1Dx are only imaginary and have nothing to do with the size of their sensors. I suppose the real loser in all this are the owners of these impostor cameras, and I truly feel sorry for those poor saps who spent such good money on them.

You really should get the word out by hopping on the Canon and Nikon full frame forums and informing them of your discovery and the error of their ways.

Lol,

it is so simple, really. A smaller pixel-site is LESS SENSITIVE to light, and under LOW LIGHT requires more time to collect the proper amount of light to avoid 'noise'.

This is why the max-ISO rating of a sensor goes down when pixel-sites become smaller. This has nothing to do with the sensor size, other than marketing demanding certain resolutions, which invariably means similar pixel count between sensor sizes, hence pixel-site sizes correlate to sensor sizes to some extent.

If you have a less sensitive pixel-site, you'd need to collect more light to overcome this. But you cannot, because more light means over-exposure. What you do instead, while keeping the Aperture the SAME, is to lower the ISO and lengthen the Shutter time. So, on a FF sensor, I could shoot at 1/60th and ISO 6400. I would have to change this to 1/30th and ISO 3200 for a smaller sensor, and 1/15th and ISO 1600 for an even smaller one and so forth.

• (Technically, it gets a bit more complicated than just this, but generally this analogy should work)

When I keep the Aperture constant, but change my Shutter time, I do change the amount of 'rainfall' collected. This is where I think you have your thoughts: a larger sensor requires less 'rainfall' to produce a quality image than a smaller sensor.

But you were changing the parameters in your thinking: if you freeze FL, Aperture, Shutter time, ISO, then all sensors, of all sizes, receive the SAME amount of light per area. For low ISO (ISO 100), this is fine, all pictures look rather similar (except for FOV). For high ISO (ISO 3200), this is not so: the smaller the sensor, the more noise you will be seeing. And again, reducing the noise can be done by lowering the ISO and lengthening the shutter time.

But the earlier thread was not about this, it was about the simple question: same exposure = same result, regardless of sensor size. And this holds true. (again, in terms of light per area collected). But the quality of the result can vary, especially under low light, this also holds true.

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

It is only for low light and shallow DOF (creative or subject isolation) that a larger sensor camera is desirable. Plus, for long range (tele), a crop sensor can in fact be a better choice, as you get more reach (as long as you have sufficient lighting).

And, when process-nodes move forwards, all pixel-sites become more sensitive, making the above only a valid comparison among similar process-nodes, but not over (calendar) time.

And manufacturer's 'max-ISO' rating is not what I would use as max-ISO, there is a point at which you begin to notice noise, which I would consider the max-ISO from a user perspective.

 

[wcdennis]

blue_skies wrote:

Tom2572 wrote:

wcdennis wrote:

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space. So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

Yep, I'm the dumb*** who thought that full frame sensors took cleaner low light photos than crop frame sensors at a given aperture. Boy was I mistaken. Glad you guys proved me wrong by pointing out the sham regarding full frame cameras.

Again, thank you for proving that the low light capabilities of cameras like the Nikon D3s, Sony A99 and Canon EOS 1Dx are only imaginary and have nothing to do with the size of their sensors. I suppose the real loser in all this are the owners of these impostor cameras, and I truly feel sorry for those poor saps who spent such good money on them.

You really should get the word out by hopping on the Canon and Nikon full frame forums and informing them of your discovery and the error of their ways.

Lol,

it is so simple, really. A smaller pixel-site is LESS SENSITIVE to light, and under LOW LIGHT requires more time to collect the proper amount of light to avoid 'noise'.

This is why the max-ISO rating of a sensor goes down when pixel-sites become smaller. This has nothing to do with the sensor size, other than marketing demanding certain resolutions, which invariably means similar pixel count between sensor sizes, hence pixel-site sizes correlate to sensor sizes to some extent.

If you have a less sensitive pixel-site, you'd need to collect more light to overcome this. But you cannot, because more light means over-exposure. What you do instead, while keeping the Aperture the SAME, is to lower the ISO and lengthen the Shutter time. So, on a FF sensor, I could shoot at 1/60th and ISO 6400. I would have to change this to 1/30th and ISO 3200 for a smaller sensor, and 1/15th and ISO 1600 for an even smaller one and so forth.

• (Technically, it gets a bit more complicated than just this, but generally this analogy should work)

When I keep the Aperture constant, but change my Shutter time, I do change the amount of 'rainfall' collected. This is where I think you have your thoughts: a larger sensor requires less 'rainfall' to produce a quality image than a smaller sensor.

But you were changing the parameters in your thinking: if you freeze FL, Aperture, Shutter time, ISO, then all sensors, of all sizes, receive the SAME amount of light per area. For low ISO (ISO 100), this is fine, all pictures look rather similar (except for FOV). For high ISO (ISO 3200), this is not so: the smaller the sensor, the more noise you will be seeing. And again, reducing the noise can be done by lowering the ISO and lengthening the shutter time.

But the earlier thread was not about this, it was about the simple question: same exposure = same result, regardless of sensor size. And this holds true. (again, in terms of light per area collected). But the quality of the result can vary, especially under low light, this also holds true.

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

It is only for low light and shallow DOF (creative or subject isolation) that a larger sensor camera is desirable. Plus, for long range (tele), a crop sensor can in fact be a better choice, as you get more reach (as long as you have sufficient lighting).

And, when process-nodes move forwards, all pixel-sites become more sensitive, making the above only a valid comparison among similar process-nodes, but not over (calendar) time.

And manufacturer's 'max-ISO' rating is not what I would use as max-ISO, there is a point at which you begin to notice noise, which I would consider the max-ISO from a user perspective.

 

[GaryW]

blue_skies wrote:

Tom2572 wrote:

wcdennis wrote:

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space.

Back-lit sensors help with this on the very small sensors. I recall a statement where Sony said that beyond the typical P&S size, it wasn't worth it. So, comparing APC-C to FF, I probably wouldn't worry so much about space between the receptors.

So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

Yep, I'm the dumb*** who thought that full frame sensors took cleaner low light photos than crop frame sensors at a given aperture. Boy was I mistaken. Glad you guys proved me wrong by pointing out the sham regarding full frame cameras.

Again, thank you for proving that the low light capabilities of cameras like the Nikon D3s, Sony A99 and Canon EOS 1Dx are only imaginary and have nothing to do with the size of their sensors. I suppose the real loser in all this are the owners of these impostor cameras, and I truly feel sorry for those poor saps who spent such good money on them.

You really should get the word out by hopping on the Canon and Nikon full frame forums and informing them of your discovery and the error of their ways.

Lol,

it is so simple, really. A smaller pixel-site is LESS SENSITIVE to light, and under LOW LIGHT requires more time to collect the proper amount of light to avoid 'noise'.

This is why the max-ISO rating of a sensor goes down when pixel-sites become smaller. This has nothing to do with the sensor size, other than marketing demanding certain resolutions, which invariably means similar pixel count between sensor sizes, hence pixel-site sizes correlate to sensor sizes to some extent.

If you have a less sensitive pixel-site, you'd need to collect more light to overcome this. But you cannot, because more light means over-exposure. What you do instead, while keeping the Aperture the SAME, is to lower the ISO and lengthen the Shutter time. So, on a FF sensor, I could shoot at 1/60th and ISO 6400. I would have to change this to 1/30th and ISO 3200 for a smaller sensor, and 1/15th and ISO 1600 for an even smaller one and so forth.

• (Technically, it gets a bit more complicated than just this, but generally this analogy should work)

When I keep the Aperture constant, but change my Shutter time, I do change the amount of 'rainfall' collected. This is where I think you have your thoughts: a larger sensor requires less 'rainfall' to produce a quality image than a smaller sensor.

But you were changing the parameters in your thinking: if you freeze FL, Aperture, Shutter time, ISO, then all sensors, of all sizes, receive the SAME amount of light per area. For low ISO (ISO 100), this is fine, all pictures look rather similar (except for FOV). For high ISO (ISO 3200), this is not so: the smaller the sensor, the more noise you will be seeing. And again, reducing the noise can be done by lowering the ISO and lengthening the shutter time.

I think you need to add that this assumes that both sensors have the same # of pixels. So, the smaller photosites on the smaller sensor collect less information, which explains the higher noise rate.

But the earlier thread was not about this, it was about the simple question: same exposure = same result, regardless of sensor size. And this holds true. (again, in terms of light per area collected). But the quality of the result can vary, especially under low light, this also holds true.

This is the bottom line. I still hold, though, that as you fix the aperture and shutter between two cameras, the ISO value varies because manufacturers don't stick to the proper standard, which just adds to the overall confusion.

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

I wish this were completely true, but I have P&S photos with blown highlights that say otherwise. ;-) Seriously, look at DxO, and it appears to me that larger sensors have more dynamic range. Here again, I think it's the advantage of larger photosites/pixels -- you can capture more "water" with the same exposure. ;-) The advantage is not just less noise at high ISO, but a bigger bucket to work with gives finer control.

But in general, yes, in broad daylight, you should get fine photos out of even a small P&S, but it's hard to ignore how often a larger dynamic range helps.

It is only for low light and shallow DOF (creative or subject isolation) that a larger sensor camera is desirable. Plus, for long range (tele), a crop sensor can in fact be a better choice, as you get more reach (as long as you have sufficient lighting).

Well, it's a more affordable and more compact choice, to be sure. (I'm sure that there are some capable, but huge, FF tele lenses....)

And, when process-nodes move forwards, all pixel-sites become more sensitive, making the above only a valid comparison among similar process-nodes, but not over (calendar) time.

Ok, you lost me at "process-node".

And manufacturer's 'max-ISO' rating is not what I would use as max-ISO, there is a point at which you begin to notice noise, which I would consider the max-ISO from a user perspective.

I think this is too limiting. Sometimes it's worth tolerating a bit more noise. I don't like to go too aggressive on the NR, preferring to leave a bit of "grain" in the photo many times, particularly using my RAW processor which produces a fine grain rather than the usual chunky look.

 

[RGBaker]

Part of the confusion is that this is one area where film and digital are different:


Historically, film ISO was determined by one thing -- size of the crystals, which correlated to the graininess of an image. Everyone used the same technology, so a 'faster' film always meant a coarser grain. But as this grain size was fixed, the choice to use a film in a large format camera resulted in a correspondingly reduced grain to image size result. i.e. a 400 ASA film used in a 35mm camera delivered a grainy image when enlarged to 8x10; the same 400 ASA film used in a 2 1/4 x 2 3/4 camera delivered a noticeably less grainy 8x10.




Today, we are still in the developing stages of sensor chip technology. Some manufacturers may use modestly different methods to coax more ISO out of their sensor, not as simple as just using larger sensor sites as they would have in the days of film crystals. Doubtless there will come a day when the technology has matured, and each every camera will offer something near identical in individual sensor site performance .... When that day comes, the relationship between low light performance, 'resolution' as is measured against 'graininess', and image quality will reach the simple equation of days past:


Resolution will be directly measured against number of pixels (photo sites) per area, assuming a lens capable of delivering said resolution; low light performance will be the inverse of number of pixels -- a high density sensor like the NEX7 delivers high resolution at the expense of some low light performance, therefore a sensor with the same TOTAL number of photo sites in a larger area (i.e. FF) would deliver 'better' resolution measured against a final same-sized print and better low light performance. A sensor that delivered the same density of photo sites would deliver increased resolution and matched low light capability, that is the increased resolution would come at the expense of low light as the 'grain' became finer.





So we are left with some eternal truths and some shifting ones -- more pixels equals increased resolution, assuming the lens can match the pixels, regardless of sensor size, when an image is printed to the same dimensions. This is film-like, where an 8x10 view camera delivered stunning resolution in a Playboy centrefold against the best possible result from a 35mm camera. Larger pixels deliver better low light performance -- this too is film-like, where coarse grain films outperform fine grain films in low light situations. Where things have changed is in the inability to make matching baselines between devices -- two camera with the same sensor size but a different number of pixels will forever have different resolution capabilities, and different low light performance. Two cameras with different sensor sizes but the same density of pixels may have either the same resolution, or the same low light performance ... but not both. And you can't swap sensors the way you swapped film to change the cameras abilities based on the circumstance.




A FF NEX with 24MP will deliver better resolution than the APS-C NEX, but the first will have better low light performance as each pixel will be correspondingly larger. The VG900 exists today offering exactly that against the NEX7. A (hypothetical) FF NEX with 36MP would deliver even better resolution but the same low light performance as each individual pixel (photo site) would be the same size as an APS-C 24MP.




This of course assumes no 'break throughs' in sensor design, and at this stage that is still a big assumption. Today's sensors are remarkably 'better' than those of ten years ago -- a roll of Kodachrome in 1985 was pretty much the same as a roll in 1975, and the equivalent roll of Fuji film was pretty much the same as any roll of Kodak. Sensor technology will likely continue to advance, so direct model to model comparisons will still be valid, but in general the notion of photo site density being directly related to low light performance will hold.




Cheers,
GB

 

[Tom2572]

wcdennis wrote:

blue_skies wrote:

Tom2572 wrote:

wcdennis wrote:

where this entire thread went wrong is someone decided the smaller sensor is noisier and generally capable of less image quality because it receives less total light. This idea seems so flat-out correct, that they couldn't be dissuaded. The actual issue is that as pixels get smaller on a chip the area that receives light not only gets smaller, but the borders between the receptors, which can only be made so small, take up more space. So, even though the same amount is falling on the surface, some of it ends up lost in the gaps between. Note that the same thing happens on full frame sensors with a similar pixel pitch--totally negating the the "more light on a large sensor" theory. I love the fact that it all dissolved into a discussion of measuring rainfall!

Yep, I'm the dumb*** who thought that full frame sensors took cleaner low light photos than crop frame sensors at a given aperture. Boy was I mistaken. Glad you guys proved me wrong by pointing out the sham regarding full frame cameras.

Again, thank you for proving that the low light capabilities of cameras like the Nikon D3s, Sony A99 and Canon EOS 1Dx are only imaginary and have nothing to do with the size of their sensors. I suppose the real loser in all this are the owners of these impostor cameras, and I truly feel sorry for those poor saps who spent such good money on them.

You really should get the word out by hopping on the Canon and Nikon full frame forums and informing them of your discovery and the error of their ways.

Lol,

it is so simple, really. A smaller pixel-site is LESS SENSITIVE to light, and under LOW LIGHT requires more time to collect the proper amount of light to avoid 'noise'.

This is why the max-ISO rating of a sensor goes down when pixel-sites become smaller. This has nothing to do with the sensor size, other than marketing demanding certain resolutions, which invariably means similar pixel count between sensor sizes, hence pixel-site sizes correlate to sensor sizes to some extent.

If you have a less sensitive pixel-site, you'd need to collect more light to overcome this. But you cannot, because more light means over-exposure. What you do instead, while keeping the Aperture the SAME, is to lower the ISO and lengthen the Shutter time. So, on a FF sensor, I could shoot at 1/60th and ISO 6400. I would have to change this to 1/30th and ISO 3200 for a smaller sensor, and 1/15th and ISO 1600 for an even smaller one and so forth.

• (Technically, it gets a bit more complicated than just this, but generally this analogy should work)

When I keep the Aperture constant, but change my Shutter time, I do change the amount of 'rainfall' collected. This is where I think you have your thoughts: a larger sensor requires less 'rainfall' to produce a quality image than a smaller sensor.

But you were changing the parameters in your thinking: if you freeze FL, Aperture, Shutter time, ISO, then all sensors, of all sizes, receive the SAME amount of light per area. For low ISO (ISO 100), this is fine, all pictures look rather similar (except for FOV). For high ISO (ISO 3200), this is not so: the smaller the sensor, the more noise you will be seeing. And again, reducing the noise can be done by lowering the ISO and lengthening the shutter time.

But the earlier thread was not about this, it was about the simple question: same exposure = same result, regardless of sensor size. And this holds true. (again, in terms of light per area collected). But the quality of the result can vary, especially under low light, this also holds true.

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

It is only for low light and shallow DOF (creative or subject isolation) that a larger sensor camera is desirable. Plus, for long range (tele), a crop sensor can in fact be a better choice, as you get more reach (as long as you have sufficient lighting).

And, when process-nodes move forwards, all pixel-sites become more sensitive, making the above only a valid comparison among similar process-nodes, but not over (calendar) time.

And manufacturer's 'max-ISO' rating is not what I would use as max-ISO, there is a point at which you begin to notice noise, which I would consider the max-ISO from a user perspective.

 

[Tom2572]

GaryW wrote:

blue_skies wrote:

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

I wish this were completely true, but I have P&S photos with blown highlights that say otherwise. ;-) Seriously, look at DxO, and it appears to me that larger sensors have more dynamic range. Here again, I think it's the advantage of larger photosites/pixels -- you can capture more "water" with the same exposure. ;-) The advantage is not just less noise at high ISO, but a bigger bucket to work with gives finer control.

LOL, after all the heartburn someone finally acknowledging my esoteric rainfall analogy....

 

[Photozopia]

Tom2572 wrote:

GaryW wrote:

blue_skies wrote:

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

I wish this were completely true, but I have P&S photos with blown highlights that say otherwise. ;-) Seriously, look at DxO, and it appears to me that larger sensors have more dynamic range. Here again, I think it's the advantage of larger photosites/pixels -- you can capture more "water" with the same exposure. ;-) The advantage is not just less noise at high ISO, but a bigger bucket to work with gives finer control.

LOL, after all the heartburn someone finally acknowledging my esoteric rainfall analogy....

This is why the last thread descended into chaos - everyone saying 'container size' has no effect on the depth of water - 'everything collects one inch' .... but then contradict their analogy by saying bigger 'buckets' hold more water ... or that smaller ones overflow .... when they believe 'one inch' is collected equally in each.


Can't have it both ways - either everything gets the same exposure by AREA (an inch of water - no matter the size of receptacle) or it doesn't.

Everything gets equal exposure (rain/light/snow/duck-sh*t hits).

Water certainly does not act as light, nor in analogy ....


Which is why I 'retitled' the last thread ... used as the title here, in continuity - "Water is not light".

It isn't ... and should not be used as analogy .... and you lot should stop wasting your time holding contradictory views based upon it. Or accept that equal AREA gets equal exposure and processor efficiency is what you are all trying to reconcile in 'noisier' images .... as in RGBaker post above :-D

 

[Philip Corlis]

Bravo!

 

[Philip Corlis]

Again, Bravo!

 

[RGBaker]

Photozopia wrote:

Tom2572 wrote:

GaryW wrote:

blue_skies wrote:

Now, not to confuse anything, but until you hit the max-ISO, you can use FF and crop size sensors similarly, and the results will be quite comparable. This is why P&S cameras can produce such great pictures in the middle of the day, at ISO 100 they are very sharp, with extreme DOF. A larger size sensor has no benefit here other than producing shallower DOF (which may even be undesirable).

I wish this were completely true, but I have P&S photos with blown highlights that say otherwise. ;-) Seriously, look at DxO, and it appears to me that larger sensors have more dynamic range. Here again, I think it's the advantage of larger photosites/pixels -- you can capture more "water" with the same exposure. ;-) The advantage is not just less noise at high ISO, but a bigger bucket to work with gives finer control.

LOL, after all the heartburn someone finally acknowledging my esoteric rainfall analogy....

This is why the last thread descended into chaos - everyone saying 'container size' has no effect on the depth of water - 'everything collects one inch' .... but then contradict their analogy by saying bigger 'buckets' hold more water ... or that smaller ones overflow .... when they believe 'one inch' is collected equally in each.

Can't have it both ways - either everything gets the same exposure by AREA (an inch of water - no matter the size of receptacle) or it doesn't.

Everything gets equal exposure (rain/light/snow/duck-sh*t hits).

Water certainly does not act as light, nor in analogy ....

Which is why I 'retitled' the last thread ... used as the title here, in continuity - "Water is not light".

It isn't ... and should not be used as analogy .... and you lot should stop wasting your time holding contradictory views based upon it. Or accept that equal AREA gets equal exposure and processor efficiency is what you are all trying to reconcile in 'noisier' images .... as in RGBaker post above :-D

No, everything does collect one inch of water, regardless of the container size. Rainfall, like light, is measured as density. An inch of rain delivers a 'blanket' of rain an inch thick over everything it falls on. It is measured by its thickness, not its volume -- and so an 1" of rain means the same to every point it fall on, regardless of size. Light is measured the same way, which is why my post above holds, not despite. Both light and rainfall are measured/described as the density of the fall.





In the light comparison, the sensor is responding not to a 'volume' of light but to a density of light.



HTH,
GB