A levels

[Great Bustard]

This would not be important except for the fact that it is the amount of light admitted that primarily determines image quality,

An extraordinary claim! Reference, please.

One must include the proper qualifier of "all else equal".

I thought that the qualifier 'primarily' was sufficient.

I don't think "primarily" is enough. For example, ignoring DOF considerations, the SIgma 50 / 1.4A on my 6D at f/2.8 likely has "better" IQ than the Canon 50 / 1.4 at f/2, even though the latter puts twice as much light on the sensor for a given exposure time.

OK, I was assuming a definition of 'IQ' orthogonal from lens effect, mainly about noise.

All else isn't equal and pragmatically the "amount of light determines image quality" rule sounds good except when you start taking photos and then maybe not so good. For instance: in my retirement I go for walks around the neighborhood multiple times a week. I could carry the FF camera but normally I prefer to carry my 1/1.7 sensor compact. Not just because it's smaller and lighter but also because it's so much better in low light and I often walk into the early evening.

Yep, I just said that. Better say it again: My 1/1.7 sensor compact is better in low light than the big FF gun. My lens on that camera is a 5.2 - 17.2 f/1.4 zoom. Let's just use the wide end and if you like call it 83 degrees AOV. On the FF we'll need a 24mm lens for 83 degrees AOV. Crop factor for my compact is 4.6.

As the light dims I'll drop shutter speed on the compact as low as 1/60 th sec. and usually get away with it keeping the ISO at base (100). Nice thing about that small sensor is that at f/2 I'll have all the DOF I could want taking a photo with the 5.2mm end of the zoom. It's a great lens and with a 12 megapixel sensor and 12 bit ADC I get damn good photos with it.

OK, switch cameras. Using the FF camera I'm going to have to stop the lens down to f/9.5 to get the same DOF I get from my compact at f/2. That's going to force me to raise the ISO on the FF camera to 2300 (base ISO 100). Large sensor cameras are much better at high ISO than small sensor cameras -- that's the rule. But they aren't that much better. I get to capture a full sensor exposure at base ISO with my compact camera and with the FF camera I'd be using what -- less than 1/5 of the sensor's capacity or even less? I don't want or need to do that math. I know from experience in that situation I really am going to get better photo IQ from my compact.

Let's discuss your mistakes. First of all, we're presuming that you would need f/9 to get enough DOF at 24mm FF equivalent. That's a pretty deep DOF, and, I dare say, excessive for many situations. But, like I said, let's go with that, and assume that gives you the desired DOF.

Next, you presume that the deeper DOF is preferable to lower noise. This is generally not true. You'll find that people tend to use their lenses at wider apertures in lower light because they prefer lower noise to more DOF. However, let's go with you preferring the deeper DOF.

For the same DOF and exposure time, the same total amount of light will fall on the sensor for all systems. If the sensors record the same proportion of that light (which is basically the case for sensors of the same generation, but BSI tech sensors tend to record a third to half a stop more light than non-BSI tech) and the sensors add in the same amount of electronic noise (there's a lot more variation here, but that doesn't matter until the light is very low, say ISO 6400+ FF equivalent), then the noise will be the same.

Lastly, you forget noise filtering. Unless you're using horrid glass on the FF camera, the FF camera will resolve better. This means that more noise filtering can be used on the FF photo to normalize the level of detail, making the FF photo, in the end, less noisy than the 1/1.7" photo.

So, for sure, if you are comparing an old FF camera to a modern BSI 1/1.7" camera at the same DOF and exposure time, what you say may be true. But, in general, no, it is not true at all.

 

[techjedi]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

This convention has served photography well for many years but with the advent of digital has been somewhat abused by the manufacturers for marketing reasons. In the digital world we have a much greater variety of active frame sizes in the mass market than with film, which was constrained by availability of processing services in a way that digital isn't. The manufacturers like to express their focal lengths in equivalents but then fail to do the same for the aperture. So, to take the example above, a 35mm camera with a 50mm lens gives an angle of view of 46.8 degrees. A lens described as 'f/2' will have a value of 'f' of 50mm, giving an aperture of 25mm. Now if we compare with a 1/2.4" camera. A lens giving the same angle of view, 46.8 degrees, which would be described by the manufacturer as '50mm equivalent' or quite often just '50mm' will actually have a focal length of 8.9mm, and if it has an aperture of 'f/2' the 'f' is now 8.9mm, so the aperture is actually 4.5mm. This is clearly confusing, because if the focal length has been expressed as '50mm' then one would expect the aperture to be 25mm, and let in as much light as expected from a 25mm aperture with a 46.8 degree angle of view. In fact, the camera is letting in 1/32nd the amount of light suggested.

This would not be important except for the fact that it is the amount of light admitted that primarily determines image quality, so many consumers are being conned into thinking that they will get higher image quality than they will in practice.

So, while not for a moment expecting this to happen, I would like to propose a different method of expressing aperture which meets the original objectives of expressing by focal length but is not similarly subject to abuse. I propose that instead of expressing aperture in terms of focal length, we do so directly in terms of the angle of view. Instead of an 'f-number' we have an 'a level'. So, for instance in the cases above the 50mm lens used on FF would be said to have an aperture of a/1.9, where a is the angle of view (diagonal, in degrees) whilst the 1/2.4" camera would be said to have an aperture of a/10.4

I have often had similar thoughts, albeit probably from a more naïve experience in photography.

I don't think it's naïve, the current system is clearly broken.

In terms of marketing or correctness, if you were to start labeling lenses by angle of view, what happens to the angle of view for the same lens mounted to different sensors sizes? Is the angle of view unchanged when the crop is applied on a smaller sensor?

Obviously not, the angle of view changes as you mount the lens on different crops, but this is what is really happening. Quoting the AoV is preferable to an 'equivalent focal length', IMO.

What about some metric of "perspective compression". Wouldn't that be unchanged with he same lens regardless if the image is cropped?

'Perspective compression' goes with angle of view and subject distance.

So if I take a cropped view from a lens, the perspective compression actually changes with angle of view?

I want to clarify, I understand perspective compression is relative to distance, but could you have a measurement kind of like guide number for flashes that includes distance as a variant?

 

[tony field]

Interesting new paradigm for "consumer cameras". Indeed it works. Pity the "consumer camera" owner who upgrades to an "esoteric camera" and has to resort to tradition.

However, It seems to me that, since the "esoteric camera/lens" report the actual focal length and aperture, the firmware could display nominal angle of view. The a/# reported would effectively be a certain type of "quality" index.

IMHO angle of view and a/# would be useful/interesting numbers to report. The a/# as a quality index indicator would give the medium format shooters bragging rights and the m4/3 users would need a handkerchief to dry their tears

 

[bobn2]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view.

Uh...you missed the fourth one Bob.

 

[fishy wishy]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

This convention has served photography well for many years but with the advent of digital has been somewhat abused by the manufacturers for marketing reasons. In the digital world we have a much greater variety of active frame sizes in the mass market than with film, which was constrained by availability of processing services in a way that digital isn't. The manufacturers like to express their focal lengths in equivalents but then fail to do the same for the aperture. So, to take the example above, a 35mm camera with a 50mm lens gives an angle of view of 46.8 degrees. A lens described as 'f/2' will have a value of 'f' of 50mm, giving an aperture of 25mm. Now if we compare with a 1/2.4" camera. A lens giving the same angle of view, 46.8 degrees, which would be described by the manufacturer as '50mm equivalent' or quite often just '50mm' will actually have a focal length of 8.9mm, and if it has an aperture of 'f/2' the 'f' is now 8.9mm, so the aperture is actually 4.5mm. This is clearly confusing, because if the focal length has been expressed as '50mm' then one would expect the aperture to be 25mm, and let in as much light as expected from a 25mm aperture with a 46.8 degree angle of view. In fact, the camera is letting in 1/32nd the amount of light suggested.

This would not be important except for the fact that it is the amount of light admitted that primarily determines image quality, so many consumers are being conned into thinking that they will get higher image quality than they will in practice.

So, while not for a moment expecting this to happen, I would like to propose a different method of expressing aperture which meets the original objectives of expressing by focal length but is not similarly subject to abuse. I propose that instead of expressing aperture in terms of focal length, we do so directly in terms of the angle of view. Instead of an 'f-number' we have an 'a level'. So, for instance in the cases above the 50mm lens used on FF would be said to have an aperture of a/1.9, where a is the angle of view (diagonal, in degrees) whilst the 1/2.4" camera would be said to have an aperture of a/10.4

--
Tinkety tonk old fruit, & down with the Nazis!
Bob

You have a problem here of trying to force a new standard on the market which suits the selling of FF over everything else. There's no reason why the manufacturers of exclusively smaller format cameras like Panasonic and Olympus should ever submit to this.

It would also overload ignorant consumers who at best can remember the field of view of 35mm camera zoom ranges.

I never see people have a problem with the concept that cameras with a bigger sensor have potentially better quality and better noise characteristics simply because they collect more light, meaning that a camera with a small sensor runs out of steam at a relatively low ISO.

On the other hand, people do not recognise the matter of aperture, depth of field and cost. A lot of time gets wasted on people so pleased with their swanky new little camera system, tirelessly recommending it to others but omitting the fact that to get similar performance to larger systems you need MUCH faster lenses which are MUCH more expensive. i.e. the 25mm f1.2 on Olympus m4/3 for $1200, can be used in exactly the same way, SAME angle of view and distance as a FF 50mm lens, but the FF lens need only be f2.5 to give the same depth of field as the Olympus 50mm f1.2, and ultimately give the same theoretical noise performance from the FF sensor too. You could probably find a 50mm f2.5 non-macro for $20 somewhere. You can get 50mm on FF a whole stop faster new for under $100 after all.

 

[Just Tim 3]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

As you correctly point out here, aperture is set and the f-stop system was devised to maintain consistent exposure across all formats and focal lengths. It's primary function is not and never has been as a measure of dof.

Think about it, f5.6 gives you consistent what? Exposure across all formats and focal lengths, it does not give any consistency to dof which is dependant on not only f-stop and focal length, (combined they are aperture diameter), but also is highly dependant on subject distance and format size, (magnification of image to print for viewing has to be taken into account when calculating dof so must also be taken into account by any numerical system indicating dof. Have you factored that into your new dof measuring system?).

You also fail to note that the subject distance is by far the most important and that "equivalent" apertures depends on dof changing proportionally to the four parameters mentioned above, which it only does across a relatively small range of distances. Outside this range dof fails to change in a linear manner and the "crop factor multiple of f-stop" simply fails to give any indication of dof, or change of dof between formats.

At very close focus distances the effective change in focal length cannot be ignored and because the f-stop is calibrated at infinity focus it no longer is in proportion to focal length and therefore can no longer give consistent exposure. It certainly throws "equivalent aperture" right out of the window.

Once you exceed the hyperfocal distance to infinity dof does not increase proportionally to f-stop. If your camera format gave a dof (hyperlocal to infinity) of 3' to infinity at f8 with a 20mm lens just how much do you think it increases if you stop down to f11? If you had a larger format that gave a dof of 5'-35' with the same fov and exposure it would still give a proportional increase in dof when stopping down.

The point is that in landscape and distant subjects it is very often the case that with smaller formats the range of apertures available do very little to alter the dof of the scene because you subject is already within the hyperfocal distance to infinity. Whereas with a larger format with the range of apertures available the dof still changes in proportion to f-stop. Therefore there is no value in "equivalent aperture" because changing f-stop to achieve exposure with the limited range of apertures available to you with that lens in that light across different formats does not always produce a linear or even equivalent relationship.

So what I end up with is a camera with a set of numbers that do not indicate either consistent exposure or an indication of dof when shooting distant landscapes.

 

[D Cox]

On the contrary, my system removes the need to know about 'equivalence' entirely.

There was never any need in the first place.

If that's your opinion then you should enthusiastically support my proposal, which removes the need even to debate equivalence.

It is a hobby for theologians.

And people who'd like to be able to predict what their photo will look like.

Why would you need your esoteric theory for this ? Just use a camera or lens for an hour or two, and you know what the results look like.

It is mainly a matter of having a good visual imagination.

 

[D Cox]

In terms of marketing or correctness, if you were to start labeling lenses by angle of view, what happens to the angle of view for the same lens mounted to different sensors sizes? Is the angle of view unchanged when the crop is applied on a smaller sensor?

The label on the box would say something like "50mm, 50 degrees on full frame".

This might give some idea of the perspective to expect.

I don't think the angle of view as such has any effect on exposure. The distance from lens to sensor does -- that is why you need to increase the shutter time when using extension tubes.

 

[D Cox]

It's a 'relative' scale, not an 'equivalent' one. f-numbers are strictly the reciprocal of the relative aperture - as in 'the aperture is f/2.8', the aperture is expressed relative to the focal length ('f'). In this example the 'relative aperture' is 1/2.8 (more commonly written 1:2.8 as you see on the front ring of a lens) and the 'f-number' is 2.8. My proposal is also a 'relative aperture', but the aperture is given relative to the angle of view.

It is f/2.8, not 1/2.8.

Yes, it would be best to stop talking about "equivalent focal lengths". It confuses everyone. The angle of view of a lens should be shown as, for example, 45 degrees on a 36mm wide sensor. (Or use diagonals as for TV screens.) The true focal length should be given too.

Yes, my proposal is to move to angles of view and angle-of-view-relative apertures. So for a FF 35mm camera with a 50mm lens with a 25mm aperture, you would quote it as a '46.8° a/1.9' and for a 1/2.3" camera with a 8.9mm lens, you would quote it as a '46.8° a/10.4'. I expect in the unlikely case of this ever being taken seriously, we would use more convenient values, so things would settle on an a/2 and a/11 apertures, with the sequence going in stops pretty much as it does for f-numbers.

I'm not sure why, when this system is in place, anyone would need to know the actual focal length.

You need it for macro work, if nothing else.

 

[D Cox]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

As you correctly point out here, aperture is set and the f-stop system was devised to maintain consistent exposure across all formats and focal lengths. It's primary function is not and never has been as a measure of dof.

Think about it, f5.6 gives you consistent what? Exposure across all formats and focal lengths, it does not give any consistency to dof which is dependant on not only f-stop and focal length, (combined they are aperture diameter), but also is highly dependant on subject distance and format size, (magnification of image to print for viewing has to be taken into account when calculating dof so must also be taken into account by any numerical system indicating dof. Have you factored that into your new dof measuring system?).

You also fail to note that the subject distance is by far the most important and that "equivalent" apertures depends on dof changing proportionally to the four parameters mentioned above, which it only does across a relatively small range of distances. Outside this range dof fails to change in a linear manner and the "crop factor multiple of f-stop" simply fails to give any indication of dof, or change of dof between formats.

At very close focus distances the effective change in focal length cannot be ignored and because the f-stop is calibrated at infinity focus it no longer is in proportion to focal length and therefore can no longer give consistent exposure. It certainly throws "equivalent aperture" right out of the window.

Once you exceed the hyperfocal distance to infinity dof does not increase proportionally to f-stop. If your camera format gave a dof (hyperlocal to infinity) of 3' to infinity at f8 with a 20mm lens just how much do you think it increases if you stop down to f11? If you had a larger format that gave a dof of 5'-35' with the same fov and exposure it would still give a proportional increase in dof when stopping down.

The point is that in landscape and distant subjects it is very often the case that with smaller formats the range of apertures available do very little to alter the dof of the scene because you subject is already within the hyperfocal distance to infinity. Whereas with a larger format with the range of apertures available the dof still changes in proportion to f-stop. Therefore there is no value in "equivalent aperture" because changing f-stop to achieve exposure with the limited range of apertures available to you with that lens in that light across different formats does not always produce a linear or even equivalent relationship.

So what I end up with is a camera with a set of numbers that do not indicate either consistent exposure or an indication of dof when shooting distant landscapes.

You need swing and tilt (or focus stacking) to get enough DoF both when using large formats and when doing macro. 35mm users don't need movements for normal distances such as a few metres, but 5x4 users do. That is why the larger cameras have movements built in.

 

[fishy wishy]

In terms of marketing or correctness, if you were to start labeling lenses by angle of view, what happens to the angle of view for the same lens mounted to different sensors sizes? Is the angle of view unchanged when the crop is applied on a smaller sensor?

The label on the box would say something like "50mm, 50 degrees on full frame".

This might give some idea of the perspective to expect.

I don't think the angle of view as such has any effect on exposure. The distance from lens to sensor does -- that is why you need to increase the shutter time when using extension tubes.

But that is in effect because you're squandering some angle of view, projecting a larger image circle than the designers expected, wasting some of the total amount of light.

 

[bobn2]

It's a 'relative' scale, not an 'equivalent' one. f-numbers are strictly the reciprocal of the relative aperture - as in 'the aperture is f/2.8', the aperture is expressed relative to the focal length ('f'). In this example the 'relative aperture' is 1/2.8 (more commonly written 1:2.8 as you see on the front ring of a lens) and the 'f-number' is 2.8. My proposal is also a 'relative aperture', but the aperture is given relative to the angle of view.

It is f/2.8, not 1/2.8.

The aperture is f/2.8 (the focal length divided by 2.8)

The relative aperture is 1/2.8 (or 1:2.8) that is, the aperture is 1/2.8 of the focal length.

The f-number is 2.8 (the denominator of the relative aperture)

I'm not sure why, when this system is in place, anyone would need to know the actual focal length.

You need it for macro work, if nothing else.

Not so, you can work out your macro work in terms of angle of view and AoV relative aperture if you wish.

 

[Dutch Newchurch]

I live in a country where we teach our children metres and kilometres, until they learn to drive, when they need to understand miles per hour. They then buy petrol for their cars, in litres, but manufacturers still tell us fuel consumption in miles per gallon.

The supermarket sells me beer measured in ml, but - thank god - I will have a pint in the pub in an hour's time.

Yes, f-stops are complicated, and I'd like to see diagonal angles of view engraved on lenses (but that gets complicated when a lens will fit two different cameras, and give a different angle of view on both).

I think we'll just keep the - imperfect - system we've got. And that suits me fine. Because anything else will just make it more complicated.

 

[astwood1285]

Great post Bob. Have you ever thought about writing something similar to debunk the exposure triangle and to answer the "what is ISO anyway question". I'd like to read a concise informed opinion and never mind that sooooo many heads would explode.

Cheers

 

[bobn2]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

As you correctly point out here, aperture is set and the f-stop system was devised to maintain consistent exposure across all formats and focal lengths. It's primary function is not and never has been as a measure of dof.

Sure, but then we just move onto a discussion of the relevance of exposure when applied across different formats

Think about it, f5.6 gives you consistent what? Exposure across all formats and focal lengths, it does not give any consistency to dof which is dependant on not only f-stop and focal length, (combined they are aperture diameter), but also is highly dependant on subject distance and format size, (magnification of image to print for viewing has to be taken into account when calculating dof so must also be taken into account by any numerical system indicating dof. Have you factored that into your new dof measuring system?).

My new system is not only about DOF, it is also about consistency across formats of the light entering the lens, the amount which determines the essential signal to noise ratio of your image.

You also fail to note that the subject distance is by far the most important and that "equivalent" apertures depends on dof changing proportionally to the four parameters mentioned above, which it only does across a relatively small range of distances.

Sure, but so far as I'm aware, the present f-number system also does not 'take into account' the subject distance. My system works just fine with subject distance and allows you to calculate DoF.

Outside this range dof fails to change in a linear manner and the "crop factor multiple of f-stop" simply fails to give any indication of dof, or change of dof between formats.

My system eliminates the need to consider the 'format' when calculating DoF. The image format is an arbitrary scale in the process from scene to final image, it is an engineering detail. My system puts all those calculations in the object field, where they belong.

At very close focus distances the effective change in focal length cannot be ignored and because the f-stop is calibrated at infinity focus it no longer is in proportion to focal length and therefore can no longer give consistent exposure. It certainly throws "equivalent aperture" right out of the window.

Yes, this is a basic failure of the f-number system. It occurs because at close distances focal length ceases to be a good approximation to angle of view. My a level system eliminates that problem because it uses angle of view directly. That is another of its advantages.

Once you exceed the hyperfocal distance to infinity dof does not increase proportionally to f-stop. If your camera format gave a dof (hyperlocal to infinity) of 3' to infinity at f8 with a 20mm lens just how much do you think it increases if you stop down to f11? If you had a larger format that gave a dof of 5'-35' with the same fov and exposure it would still give a proportional increase in dof when stopping down.

Again, another advantage of my system, It does not suffer from that problem. The DoF calculations are different and more accurate, they all happen in the object field.

The point is that in landscape and distant subjects it is very often the case that with smaller formats the range of apertures available do very little to alter the dof of the scene because you subject is already within the hyperfocal distance to infinity. Whereas with a larger format with the range of apertures available the dof still changes in proportion to f-stop. Therefore there is no value in "equivalent aperture" because changing f-stop to achieve exposure with the limited range of apertures available to you with that lens in that light across different formats does not always produce a linear or even equivalent relationship.

So what I end up with is a camera with a set of numbers that do not indicate either consistent exposure or an indication of dof when shooting distant landscapes.

Then you need to switch to a levels instead of f-numbers and those problems no longer exist.

 

[bobn2]

Great post Bob. Have you ever thought about writing something similar to debunk the exposure triangle and to answer the "what is ISO anyway question". I'd like to read a concise informed opinion and never mind that sooooo many heads would explode.

Cheers

 

[bobn2]

This would not be important except for the fact that it is the amount of light admitted that primarily determines image quality,

An extraordinary claim! Reference, please.

One must include the proper qualifier of "all else equal".

I thought that the qualifier 'primarily' was sufficient.

I don't think "primarily" is enough. For example, ignoring DOF considerations, the SIgma 50 / 1.4A on my 6D at f/2.8 likely has "better" IQ than the Canon 50 / 1.4 at f/2, even though the latter puts twice as much light on the sensor for a given exposure time.

OK, I was assuming a definition of 'IQ' orthogonal from lens effect, mainly about noise.

All else isn't equal and pragmatically the "amount of light determines image quality" rule sounds good except when you start taking photos and then maybe not so good. For instance: in my retirement I go for walks around the neighborhood multiple times a week. I could carry the FF camera but normally I prefer to carry my 1/1.7 sensor compact. Not just because it's smaller and lighter but also because it's so much better in low light and I often walk into the early evening.

Yep, I just said that. Better say it again: My 1/1.7 sensor compact is better in low light than the big FF gun. My lens on that camera is a 5.2 - 17.2 f/1.4 zoom. Let's just use the wide end and if you like call it 83 degrees AOV. On the FF we'll need a 24mm lens for 83 degrees AOV. Crop factor for my compact is 4.6.

As the light dims I'll drop shutter speed on the compact as low as 1/60 th sec. and usually get away with it keeping the ISO at base (100). Nice thing about that small sensor is that at f/2 I'll have all the DOF I could want taking a photo with the 5.2mm end of the zoom. It's a great lens and with a 12 megapixel sensor and 12 bit ADC I get damn good photos with it.

OK, switch cameras. Using the FF camera I'm going to have to stop the lens down to f/9.5 to get the same DOF I get from my compact at f/2. That's going to force me to raise the ISO on the FF camera to 2300 (base ISO 100). Large sensor cameras are much better at high ISO than small sensor cameras -- that's the rule. But they aren't that much better.

Small sensors (very small ones, such as your 1/1.7" sensor) do have an advantage, they tend to have higher quantum efficiency, maybe around 70% rather than the 50% common on 4/3 and upwards. This is because the market for these sensors is mach larger, they use the more modern fab lines with finer geometries and their designers take all sorts of design steps that they don't bother with for larger sensors, that is things like BSI, metal grid light guides, internal light pipes and other measures.

I get to capture a full sensor exposure at base ISO with my compact camera and with the FF camera I'd be using what -- less than 1/5 of the sensor's capacity or even less? I don't want or need to do that math. I know from experience in that situation I really am going to get better photo IQ from my compact.

Well, your compact is performing better that strict 'equivalence' due to the factors above, but that doesn't stop you using f-numbers on the compact, even though they fail to account for that efficiency difference, does it? My a levels are no different. Where they are different is that they do account for the light at an image normalised scale.

 

[pwilly]

--
Paul
Just an old dos guy

 

[bobn2]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

This convention has served photography well for many years but with the advent of digital has been somewhat abused by the manufacturers for marketing reasons. In the digital world we have a much greater variety of active frame sizes in the mass market than with film, which was constrained by availability of processing services in a way that digital isn't. The manufacturers like to express their focal lengths in equivalents but then fail to do the same for the aperture. So, to take the example above, a 35mm camera with a 50mm lens gives an angle of view of 46.8 degrees. A lens described as 'f/2' will have a value of 'f' of 50mm, giving an aperture of 25mm. Now if we compare with a 1/2.4" camera. A lens giving the same angle of view, 46.8 degrees, which would be described by the manufacturer as '50mm equivalent' or quite often just '50mm' will actually have a focal length of 8.9mm, and if it has an aperture of 'f/2' the 'f' is now 8.9mm, so the aperture is actually 4.5mm. This is clearly confusing, because if the focal length has been expressed as '50mm' then one would expect the aperture to be 25mm, and let in as much light as expected from a 25mm aperture with a 46.8 degree angle of view. In fact, the camera is letting in 1/32nd the amount of light suggested.

This would not be important except for the fact that it is the amount of light admitted that primarily determines image quality, so many consumers are being conned into thinking that they will get higher image quality than they will in practice.

So, while not for a moment expecting this to happen, I would like to propose a different method of expressing aperture which meets the original objectives of expressing by focal length but is not similarly subject to abuse. I propose that instead of expressing aperture in terms of focal length, we do so directly in terms of the angle of view. Instead of an 'f-number' we have an 'a level'. So, for instance in the cases above the 50mm lens used on FF would be said to have an aperture of a/1.9, where a is the angle of view (diagonal, in degrees) whilst the 1/2.4" camera would be said to have an aperture of a/10.4

 

[Just Tim 3]

The amount of light in an image is determined by three things - the luminance of the scene, the size of the aperture and the angle of view. As you expand the angle of view the amount of light projected onto the image plane increases. This is inconvenient when you have a sensitive medium that requires a given amount of light, since it means that you have to reset the camera as the angle of view changes. Therefore it has become customary to express the aperture diameter in terms of the angle of view when it comes to labelling settings, though we use the focal length of the lens as a proxy for angle of view. So, rather than setting the aperture in millimetres, we set it as a fraction of the focal length. As an example, if we had a 50mm lens and wanted an aperture of 25mm, we would choose a setting of f/2, where 'f' stands for the focal length, in this case 50mm.

As you correctly point out here, aperture is set and the f-stop system was devised to maintain consistent exposure across all formats and focal lengths. It's primary function is not and never has been as a measure of dof.

Sure, but then we just move onto a discussion of the relevance of exposure when applied across different formats

Think about it, f5.6 gives you consistent what? Exposure across all formats and focal lengths, it does not give any consistency to dof which is dependant on not only f-stop and focal length, (combined they are aperture diameter), but also is highly dependant on subject distance and format size, (magnification of image to print for viewing has to be taken into account when calculating dof so must also be taken into account by any numerical system indicating dof. Have you factored that into your new dof measuring system?).

My new system is not only about DOF, it is also about consistency across formats of the light entering the lens, the amount which determines the essential signal to noise ratio of your image.

But optimum exposure is intensity of light across all formats (f-stop), it is only sensor performance that is measured with consistent light. Whatever format you use you will aim for optimal exposure as this will also give you best noise performance. What does you system do but allow us to reduce the performance of the greater of the two systems to match that of that of the lesser? This is something you do not do when taking photos.

You also fail to note that the subject distance is by far the most important and that "equivalent" apertures depends on dof changing proportionally to the four parameters mentioned above, which it only does across a relatively small range of distances.

Sure, but so far as I'm aware, the present f-number system also does not 'take into account' the subject distance. My system works just fine with subject distance and allows you to calculate DoF.

No, because f-stop is primarily a control of exposure and not dof. Dof is affected by aperture diameter, and also subject distance and magnification to final image size/viewing distance.

Outside this range dof fails to change in a linear manner and the "crop factor multiple of f-stop" simply fails to give any indication of dof, or change of dof between formats.

My system eliminates the need to consider the 'format' when calculating DoF. The image format is an arbitrary scale in the process from scene to final image, it is an engineering detail. My system puts all those calculations in the object field, where they belong.

No it doesn't because it fails to take into account that dof is not measured at the sensor but is a perceptual quality also determined by final print size and viewing distance.

At very close focus distances the effective change in focal length cannot be ignored and because the f-stop is calibrated at infinity focus it no longer is in proportion to focal length and therefore can no longer give consistent exposure. It certainly throws "equivalent aperture" right out of the window.

Yes, this is a basic failure of the f-number system. It occurs because at close distances focal length ceases to be a good approximation to angle of view. My a level system eliminates that problem because it uses angle of view directly. That is another of its advantages.

No, focal length is still a good approximation. The approximations that fail are the numbers printed on the lens as these are calculated at infinity focus, just the same as the numbers you propose to replace them with. Or will you have different sets of numbers for different subject distances that reflect the change in focal length when shooting close up?

Once you exceed the hyperfocal distance to infinity dof does not increase proportionally to f-stop. If your camera format gave a dof (hyperlocal to infinity) of 3' to infinity at f8 with a 20mm lens just how much do you think it increases if you stop down to f11? If you had a larger format that gave a dof of 5'-35' with the same fov and exposure it would still give a proportional increase in dof when stopping down.

Again, another advantage of my system, It does not suffer from that problem. The DoF calculations are different and more accurate, they all happen in the object field.

No, they're not. One of the basic and incorrect assumptions people make with equivalent apertures and trying to link sensor size to aperture is that dof behaves proportionally to aperture diameter/subject distance. It only does this within a limited range of subject distances. Outside this dof ceases to be proportional or even behave in a linear manner. Again the exact relationship between sensor size and dof is highly dependant on subject distance.

Consider: Two digital formats, one has a 35mm lens on a D70 that when set to f5.6 has a dof that extends from 18.8'' to infinity (subject distance 40'), if I stop this down then to f8 I only get an increase of a few feet, (15.5'-infinity). If I focus it a little closer (27' subject distance) then I get a dof of 15.4'-111' at f5.6 and 13'-infinity at f8. How does your system of numbers reflect this non-linear response to dof dependant on subject distance, even with the same lens at the same two f-stops on the same camera? With one distance dof increases by only 3.3' and the other closer distance it increases by many millions of miles, ( ;-) well considerably more proportionally and for a subject that's closer and therefore should have less dof?).

Now the other format is FF and with my 55mm lens I have a range of apertures from f3.5-f16 before diffraction sets in which is handy as the "equivalent" apertures are f9-f12.7 ish. But I can also choose f3.5-f-8 so pretty much the whole range of apertures produces a proportional response to dof. But with my kit lens on the D70 f5.6-f11 give me exactly the same dof for my distant landscape and I've really only got a stop and a half out of the range of apertures available that actually give me any adjustment. Does your system represent this?

The point is that in landscape and distant subjects it is very often the case that with smaller formats the range of apertures available do very little to alter the dof of the scene because you subject is already within the hyperfocal distance to infinity. Whereas with a larger format with the range of apertures available the dof still changes in proportion to f-stop. Therefore there is no value in "equivalent aperture" because changing f-stop to achieve exposure with the limited range of apertures available to you with that lens in that light across different formats does not always produce a linear or even equivalent relationship.

So what I end up with is a camera with a set of numbers that do not indicate either consistent exposure or an indication of dof when shooting distant landscapes.

Then you need to switch to a levels instead of f-numbers and those problems no longer exist.

Yes they do, because you can't calculate dof without including subject distance which cannot be linked to sensor size because it simply is not always proportional to sensor size. As f-stop is primarily a measure of exposure it does not attempt to give you an indication of dof alone, you must always know subject distance and focal length as well. So what does your system do? A different set of numbers that don't indicate either exposure or the non-proportional relationship of dof dependant on subject distance. It tells me that the D70 has greater dof for my landscapes, which it doesn't as I can achieve the dof I need with the FF. It doesn't tell me that the range of f-stops on the D70 make very little difference to dof.

You simply can't invent a simple system that does other than the tried and tested one that already exists. Which works across all formats and all medias.

And you cannot calculate dof at the sensor, it can only be done with a finished print with a known viewing distance.