F-numbers – An outdated concept?

[facts_please]

In the good old days when almost everyone owned a 35mm film camera, the concept of an F-number was extremely useful, giving clear information about low light performance and depth of field, which was directly comparable between different (35mm) cameras. The small number of people using other than 35mm film were generally professionals who knew how to account for different film sizes, so life was simple and everyone was happy.

However, in the age of digital cameras with widely varying sensor sizes, the usefulness of traditional F-numbers is less clear. When sensor size is variable, knowing only the F-number gives no information whatsoever about low light performance or depth of field.

Here is a summary of what F-numbers meant in the past, and what they mean now.


GOOD OLD DAYS – ALMOST EVERYONE HAD A 35mm FILM CAMERA

(a) Low light performance
With a 35mm camera, the smallest F-number attainable by a lens was a direct measure of how much light the lens would capture and deliver to the film. This was great. Photographers of the time quickly gained a “feel” for what an F-number meant in terms of getting a good shot in poor light. F-numbers for different lenses could be compared directly with confidence.

(b) Depth of field
With a 35mm camera, the depth of field depends on the F-number. This was great. Photographers of the time quickly gained a ‘feel’ for what an F-number meant in terms of depth of field. F-numbers for different lenses could be compared directly with confidence.




TODAY – ALMOST EVERONE OWNS A DIGITAL CAMERA

(a) Low light performance
The F-number, on it’s own, tells you nothing. This is a pain, as well as causing much confusion. Regarding the true ‘speed’ of a lens, F-numbers cannot be compared for cameras with different sensor sizes. For example, the lens on my point and shoot camera is F2.0, but it does not gather anywhere near as much light as an F2.8 lens on an SLR style camera.

(b) Depth of field
The F-number, on it’s own, tells you nothing. This is a pain, as well as causing much confusion. Regarding depth of field, F-numbers cannot be compared for cameras with different sensor sizes. For example, F8 gives a completely different depth of field on a point and shoot camera and an SLR.


In summary, F-numbers are at best confusing, and at worst near useless, in the age of digital cameras with their widely differing sensors. The ballgame has changed, but the photographic community has not. The purpose of this thread is to see if there is some other ‘F-parameter’ that avoids these problems.

The obvious solution is to normalize F-numbers with respect to sensor size, as follows :-

Normalized F-number = F-number x Crop Factor
where Crop factor is the sensor size relative to standard 35mm film format.


For brevity, I shall refer to Normalised F-numbers as NF-numbers.
I state without proof that if NF-numbers were used in place of F-numbers, all of the problems described above would instantly disappear. My previous threads explain why this is so. Some examples are the best way to explain this concept.

For lenses on 35mm film cameras, or digital cameras using the same 35mm format, the NF-number will be identical to the F-number, as the crop factor is 1.0

As the majority of digital cameras use sensors smaller than 35mm, the effect will be to normalize the F-number to a higher value. For example:

Typical SLR with APS-C sensor
Crop factor = 1.5
F-number = 2.8 (typical consumer zoom lens)
NF-number = 4.2

The interpretation here is that the F2.8 lens on this SLR , is equivalent to an F4.2 lens on a 35mm camera. That is, the depth of field will be the same, and the all else equal the amount of light gathered by the lens will also be the same. It really does make sense to discard the F-number altogether, and quote the NF-number instead. Here is another example, for a fairly good point-and-shoot camera :-


Canon G2 with 1/1.7” sensor (typical good quality point and shoot)
Crop factor = 4.7
F-number = 2.0
NF-number = 9.4

What this means, is the apparently fast (for a consumer zoom lens) F2.0 lens on the point and shoot is really no such thing, in fact the effective F-number of 9.4 is more than double that of the lens on a typical SLR. (9.4 vs 4.2), even though the point and shoot lens has a lower F-number. For portraits (or whatever) where a poor depth of field is required, we also see how misleading the F2.0 number is – it is actually equivalent to F9.4 on a 35mm camera. Again, it is evident that the NF-number is far more useful than the F-number.

The bottom line is this. If all lenses were specified in terms of NF-numbers, then all of the NF-numbers could be directly compared with each other, just as in the ‘good old days’ when we used 35mm film cameras. That would be exceedingly useful, and eliminate the very considerable present confusion and misunderstanding regarding F-numbers.

However, there are disadvantages as well. An F-number is the ratio of a lenses focal length to aperture diameter, and is therefore a property only of the lens. A NF-number is a property of the lens and the sensor with which it is to be used. For cameras with non-removable lenses there is no problem. However, for removable lenses, the lens NF-numbers will be different depending on the camera and sensor that the lens is used with, so what should be written on the lens? This may not be much of a problem in practice. For example, the new 4/3 lenses are designed very specifically to work with ‘4/3’ cameras having a standard ‘4/3’ size sensor, so there would be little problem here. I guess the key point here is that with removable lenses, NF-numbers would ALWAYS need to be quoted in conjunction with the intended size of sensor. That allows the user to convert back to F-numbers if desired, or recalculate the NF-numbers should the lens be used with a different sensor.

Adoption of NF-numbers would not mean that the concept of F-numbers is redundant. It would still be true, for example, that the cost of making a quality lens depends very much on the F-number, not the NF-number. Even if NF-numbers were adopted, I’m sure many users would still want to know the F-numbers for all sorts of reasons that I haven’t thought of.

In practice, it is incredible unlikely that the photographic industry would ever print NF-numbers on lenses, or in camera specifications, and even less likely that any manufacturer would build a digital camera where the entire menu and parameter display system worked with and displayed NF-numbers rather than F-numbers. I wish they would, but there is just too much inertia for this to change – too many existing products described in terms of F-numbers, and generally just too much familiarity with F-numbers, despite the advent of digital cameras that makes the usefulness of F-numbers appear questionable. I am not asking whether I think industry adoption of NF-numbers will ever occur. It won’t. This is an exercise in free thought. I am asking whether NF-numbers would be a better way of describing lenses, or at least a useful alternative way of describing lenses, not whether such a change will ever happen. Of course, anyone is free now to quickly and easily calculate NF-numbers for lenses/camera of interest – I certainly intend to calculate and think in terms NF-numbers from here on because in my view they are so much more useful, and I sure wish all my cameras displayed NF-numbers, too.

I enjoy ‘what if’ scenarios. If all digital cameras adopted NF-numbers rather than F-numbers, then I bet more people at the lower end of the market would spend more money to get a better camera, as it would be very obvious just how bad (=slow) the lenses on small point and shoot camera really are, whereas at the moment this fact is hidden because the F-numbers for these cameras look OK, on account of the small sensor size. More people would buy SLRs, too.

This thread is closely related to my previous threads, where I was trying to throw around some ideas regarding how best to compare the light gathering ability of different camera, which is distinctly confusing when sensor sizes is not a constant. An excellent start would be if F-numbers were replaced by NF-numbers.

Comments?
Any disadvantages of working with NF-numbers rather than F-numbers? (I bet there are)
What do YOU use F-numbers for, that NF-numbers would not be suitable for?

[Davey]

I think people don't like anything "new" so even if adoption (which as you say is unlikely) were to happen many would still use traditional f numbers, even if the new one was more fitting and useful people still cling to what they know.

I think it would be a good method for comparing but would take people time to get used to and would be better in my opinion to be viewed as a value that has a different funtion to original f number rather than a replacement. As a value for comparing different lenses it would work but makes things more complex I think in other respects, especially for thos familiar with the old system. Same reason people in UK still use imperial measurments over metric (for instance I still measure height in inches etc despite the fact I grew up with decimalisation because it's familiar for reference).

I think any camera owners who know little if anyhthing about f numbers wouldn't care either way and are unlikely to decide on an upgrade etc due to such comparissons. These people are more inclined to buy products due to marketing strategies so unless companies pushed that it wouldn't affect them. Companies not likely to do that since low end entry level up to mid range mainstream (or the bottom end of mid range specifically) are the real proportion of the market. This means a value that makes such products look poor would not be good choice for a marketing pitch. The majority of people don't look into the actual mechanics and technology of what they buy, so they are unlikely to learn what an nf number is an actual representaion of and research it.

For the photographers and amateurs/hobbyists alike who look into the actual tech side of things a value for comparisson is good. However most if not all understand the drift between f numbers on dslr and p&s cams, between lenses/sensors etc and that it's relative value. I think people tend to get used to the result in cam with a particular lens (in the case of dslr), eg. I understand my depth of field etc is much broader at 3.2 than a dslr and appropriate lens at 3.2. I understand I cannot just compare a lens based on f number alone and there are more factors involved (from both sensor size, lens size and specs etc). For me I use it as guide value based on experience of what result I actually get out on my camera at any given f number, for instance knowing at f7.1 across a distance I'm familiar with the whole scene will be sharp (based on experience of previous similar results) where as at 3.2 there will be some softening no matter where I focus (again based on experience).

Although I often learn about values a lot of my interactions with stuff comes from experience and I suspect others are similar. I can't think of a photography example with a lot of calculation needed since I don't have slr with multiple lens choices and only own a single point and shoot so for want of a better example (although this illustrates my point) I know the chemistry of styrene resins and catalysts and although I have actually calculated exactly how much MEK to what weight resin gives what work time until gel point and until set point and fully cure time I don't bother now. I just know from experience this many ml to that many g will give me 15 minutes before it goes off. In fact chemistry wise it's hard to calculate gel point or full cure time accurately since it's temp dependent. Same goes for photography in that caluclating exactly how much light relative to other lens/body combo doesn't affect many photographers and they are unlikely to work out the physics of it to find actual comparable value, and even for those who do it's not on a regular basis as they will tend to go on experience rather than theory (I suspect you may yourself).

So yeah it's a good value but aside from comparissons for knowledgable people when buying stuff it's not necessary, and even then they will check out many features and factors and will use some rough indication guide (like photo comparissons to other lenses at given f-stop etc) to the amount of light a lens lets in over physical values. I think unless you actually need a standardised value to compare things (such as if you were to write a calculator for something that was lens independent) then it would have little impact even though in theory it's great idea the real world probably wouldn't find use for it. This is just my opinion however and I could be wrong. I'd be interested in what others think and of whether you agree or disagree with any of my points. Thanks

[facts_please]

Hi Davey,

First, thanks for replying, as I think few would find this sort of thought exercise interesting!

As a value for comparing different lenses it would work but makes things more complex I think in other respects ....

Can you give details as to which things would be more complex? I mean inherently more complex, not more complex just bcause you are accustomed to f-numbers. As someone with very little photography background, a clean slate so to speak, what strikes me about nf-numbers is just how much simpler everything would be. But I don't claim to have really thought this through.

Basically, I agree with pretty much all you say, much of which is concerned with pointing out that people are not going to change for a whole host of reasons. Agreed.

However, this was intended only as a thought exercise. If the idea of f-numbers was not already established, would nf-numbers be better? Clearly I must have too much time on my hands to think and write about such stuff ...

As I'm sure you realize, the simple calculations I showed comparing the light gathering capability of various cameras is really exactly the same thing. However, overall I think the idea of nf-numbers is a better way of explaining and expressing the same idea.

Colin

[Davey]

Hehehehe explaining that to someone who regularly wonders about such odd things is not necessary. Most people just look at me blankly when I ask them have they never wondered about [insert unusual observation of mundane thing]. Clearly they haven't, many seem to just ignore 99% of what i say and just agree, perhaps a good thing I reckon hehe.

Complicated from the point of view f numbers are used in many calculations where they work well. Again I'm no photographer so doesn't affect me, and more so because I use a point and shoot but for many it's an extra value to convert. Also it's a similar value to f numbers but not identical, this makes it awkward since it's closely related so will could be interchanged sometimes if became used almost as if it was synonymous with f numbers (the way many similar but essentially different terms are).

I cannot find an inherent complication though, it's dependent on the old f system and other factors. The complication is due to the old system being the one that calculations are composed for, simple changes could no doubt be made to use nf number instead, or equations altered to take the difference into account and change nf to f as part of the equation rather than convertion to f before inputing data. Still many pros and amateurs alike will use f numbers still for this since they work, the new one would only need to be introduced if you are calculating something involving multiple sensor sizes etc in one equation (unlikely).

The second complication is as a comparative value that remains constant across the range it might be awkward to figure out some real world effects of it. That is it's a constant value between different lens/sensor sizes where as the current digital f number is relative to the equipment is a plus point I'd say. You can theoretically work out the dof and how much light (relative to the aperature at least) is let in by the lens and so on easier. But the issue is if you are familiar with high f number to cut light by however much, extend range of dof and so on then it remains similar-ish between cameras. Yet with nf numbers since it's not a relative figure it means learning a wider range and without familiarity it's difficult.

These are NOT inherent to nf numbers though, I cannot see an inherent problem to using such a value (although there may be one, just I can't see it if there is). In fact nf numbers are to digital cameras what f numbers are to 35mm film, a more accurate constant value that is applicable across the board. The problem is more the variety of digital setups coupled with an old system being used where a new one would have been better. If there was no f numbering system for digital and nf was used straight off then there would be no issues since everyone would be familiar and accept the widely used standard and accept it was different from f numbers in 35mm film /full frame system (as they do with exposing for highlights instead of shadows which is first thing I noticed with digital and is commonly accepted). Obviously the real problem is either varied sensor& lens specs making f numbers relative, or familiarity and wide acceptance of an old system. As unlikely as it is to be put on lenses it is possible such a value may be used as a new comparative value (when comparing varied setups) since there currently isn't one where aperature is concerned.

The only issue I personally have with it is it could be dangerous for me to talk loudly around those ignorant of the standard since nf is a far right racist group in the UK, I have a skinhead (close to the bone ala boneheads although I'm not), wear boots (14 hole invaders to be precise) and often wear black (again a nazi skin thing, trojans and afa oi crowd tend to avoid all black due to this). I already have issues with ignorant people calling me a nazi, I don't need another excuse for them to

[Henry Peach]

For example, the lens on my point and shoot camera is F2.0, but it does not gather anywhere near as much light as an F2.8 lens on an SLR style camera.

You are wrong. F/2 is f/2 and allows the same amount of light to pass whether using your compact digital camera or a 4x5 camera. That's the point of using f/#'s: it's the same amount of light whatever lens or camera you are using. Notice that hand held meters don't ask you which format you are using, yet they work fine with cameras from sub-miniature to ULF.

The size of the photo receptors on the sensor may influence how many photons are gathered, but at a given f/# all lenses allow the same amount of light to pass.

(b) Depth of field. The F-number, on it’s own, tells you nothing. This is a pain, as well as causing much confusion. Regarding depth of field, F-numbers cannot be compared for cameras with different sensor sizes. For example, F8 gives a completely different depth of field on a point and shoot camera and an SLR.

DOF is controlled by 4 factors:

1) aperture size
2) focal length
3) focusing distance
4) format size (people argue this, I guess it depends on how you look at it, but in the real world if I'm composing a scene with a 35mm camera using f/8 on a normal lens for the format it's going to have significantly more DOF than a similarly composed scene with a 4x5 camera at f/8 on a normal lens for the format).

First, I think that f/#'s were mainly created to help calculate exposure not DOF, and as I said above, format size has no effect on exposure. Second, how do your normalized numbers account for the effects of focal length and focusing distance on DOF? Look up the formula for calculating DOF, and you'll see it's alot more complicated than your formula accounts for.

I have a better idea: since we are carrying computers, why not have them calculate the DOF and display it on the LCD or in the viewfinder? I've installed the CHDK hack in my Canon Powershots, and I love the digital DOF scale. I don't understand why my DSLRs don't have it.

Photography is going on 185ish years old. During that entire time there has been one brief time when the masses only used 35mm: the 1980s. Pre-1980 consumer cameras were made in 110 format, 35mm format, 120, and 620 formats, and probably a few others. Medium format consumer cameras disappeared in the 1960s. 110 was pretty much gone by the 80s, but by the 90s we had APS format. Photography enthusiasts have been juggling multiple formats since the beginning; it's not as hard to figure out as you make it out to be.

[McQ]

You are wrong. F/2 is f/2 and allows the same amount of light to pass whether using your compact digital camera or a 4x5 camera. That's the point of using f/#'s: it's the same amount of light whatever lens or camera you are using. Notice that hand held meters don't ask you which format you are using, yet they work fine with cameras from sub-miniature to ULF.

The size of the photo receptors on the sensor may influence how many photons are gathered, but at a given f/# all lenses allow the same amount of light to pass.

DOF is controlled by 4 factors:

1) aperture size
2) focal length
3) focusing distance
4) format size (people argue this, I guess it depends on how you look at it, but in the real world if I'm composing a scene with a 35mm camera using f/8 on a normal lens for the format it's going to have significantly more DOF than a similarly composed scene with a 4x5 camera at f/8 on a normal lens for the format).

First, I think that f/#'s were mainly created to help calculate exposure not DOF, and as I said above, format size has no effect on exposure. Second, how do your normalized numbers account for the effects of focal length and focusing distance on DOF? Look up the formula for calculating DOF, and you'll see it's alot more complicated than your formula accounts for.

I fully agree with Henry's last post. f/-numbers have a great universality when it comes to describing the flux of light (exposure) hitting a camera sensor -- regardless of focal length or sensor size. This was their original purpose. Honestly, precise depth of field is often a secondary concern to precise exposure...you need the former before the latter.

[William W]

I also agree, with what Henry wrote .

Specifically:

“First, I think that f/#'s were mainly created to help calculate exposure not DOF, and as I said above, format size has no effect on exposure.”

***

This is very important. F numbers are not outdated and they perform a vital role.

IMO, having “new F numbers” would be silly; and, moreover using F numbers we have, is vital – F numbers are part of the universal Trilogy of Exposure, all Photographers can immediately understand.

Aperture (“F Number”) / Shutter Speed (Exposure Time) / ISO (Sensitivity) – these three numbers form the base language of the technical premise of all image making – we can communicate so much with just three numbers - camera size, film size or sensor format has nothing to do with the communication of information via the use of these three numbers.

***

To bring this to a practicality, let’s just take a fact and then invent a scenario:

I know, I need at least 1/640s to establish a reasonable freeze of the swimmers at a Backstroke Start.

Assuming the lighting is even throughout the pool.

If I meter the light at the Indoor Pool with my DSLR and I get F2.0 @ 1/640s @ ISO800 then I know many things, three of which are:

> Firstly I know if I choose to use my EF70-200F2.8, I will need ISO1600, at least to shoot a Backstroke Start

> Secondly I know that if I forgot to bring my DSLR kit and I only had a 645 with a 180mm/F4 lens, I would need to push the film to ISO3200, at least, for a Backstroke Start.

> Thirdly if the 645 failed and I borrowed my Wife’s Canon Powershot 5 IS – (max aperture F2.7 at the wide FL – max ISO1600) I would need to shoot only at the wide angle FL, and I would have to use ISO1600.

And also, let’s pretend a little more: assume this is an Inter-Solar System swimming meet:

If all the other Photographers are from Mars, and can speak no Earth Language, and all their Light Meters have broken . . . any one of them can just is look at my exposure settings to know:

They can shoot the Medal Ceremony on 5x4 View Camera using an F5.6 lens wide open at 1/40s exposure using ISO400 film.

WW

[facts_please]

Hi Henry,

Originally Posted by facts_please
For example, the lens on my point and shoot camera is F2.0, but it does not gather anywhere near as much light as an F2.8 lens on an SLR style camera.

You are wrong. F/2 is f/2 and allows the same amount of light to pass whether using your compact digital camera or a 4x5 camera. That's the point of using f/#'s: it's the same amount of light whatever lens or camera you are using. Notice that hand held meters don't ask you which format you are using, yet they work fine with cameras from sub-miniature to ULF.
The size of the photo receptors on the sensor may influence how many photons are gathered, but at a given f/# all lenses allow the same amount of light to pass.

You are correct in that F2 always produces the same intensity on the sensor. (=photons/area/time) - that is not in dispute.

However, I am also correct in stating that my F2.0 point and shoot does not deliver as much light (=number of photons) onto the sensor, as an F2.8 lens on an SLR, and that is what matters in a digital camera. Image noise depends on the absolute number of photons striking the sensor, NOT on the intensity of light striking the sensor per se. This is not analogous to film. With a film camera, the 'speed' of the lens (meaning fnumber) was all that mattered, regardless of film format size.

For example, in a film camera, if an F2.0 lens at 1/00th second produces correct exposure on a 35mm camera, then an F2.0 lens at 1/00th second will also produce correct exposure on any other format size film camera as well, assuming of course the same type of film is used and the same subject is photographed at the same field of view. Even so, the lens in the larger format camera will require a larger absolute aperture diameter (=bigger lens), and have inferior depth of field. Nothing is free - you need more total photons for the larger format, and you pay for that with a larger lens.

In a digital camera, lens fnumber cannot be considered in isolation. What matters is the performance (image noise, speeed, or however you want to say it) of the system comprising the lens and the sensor as a whole. I stand behind what I said. By any meaningful measure, my F2.0 point and shoot camera is slower than the F2.8 SLR - it requires a brighter subject for the same image noise. If NF-numbers are used, then you reach the correct conclusion that the F2.8 SLR is faster.

I have actually introduced a whole new concept here. Exactly what does 'speed' mean in the digital camera age? We are traditionally accustomed to referring only to the speed of a lens, and in the film era the speed of the lens and the speed of the camera were in effect the same thing. However, for a digital camera it makes more sense to speak of the 'speed' of the lens+sensor combination, of the entire camera. At the end of the day, I am not a 'numbers' person - what I care about is how things perform in the real world. By any meaningful measure, my F2.0 point and shoot camera is slower than an F2.8 SLR - it requires a brighter subject for the same image noise. The rest is just semantics.

I also stand behind my DOF statements. If two cameras (=lens+sensor) have the same NF-number, then they will produce the same depth of field under the same shooting conditions, meaning same distance from object and same field of view. It really is that simple - a very compelling case for thinking in terms of NF-numbers.

I'll address other posts, as time allows. I don't seriously think that f-numbers have no use, by the way, but NF-numbers are very useful indeed for comparing the effective speed and DOF between digital camera having different sensor sizes. For many other things, I'm sure that f-numbers still rule.

PS!! This added after original posting. When I speak of 'aperture', I always mean absolute aperture diameter. When I speak of f-number, I mean f-number. I urge everyone to adopt this convention, otherwise misunderstanding and confusion is inevitable.

[William W]

Attempting to slip into what I believe you are getting at and applying it to film only. . .

The "speed" of lenses (used exclusively for film cameras) do have some kind of "a relationship" to the size of the film being used - because it takes an HUGE chunk of glass for get an F2.8 lens to cast an image circle on the negative of a 5x4 view camera, and a much smaller chunk of glass to get an F2.8 lens to cover the neg of a 35mm SLR.

Think of all the photons that are stimulating that 5x4 neg when such a lens is used at F2.8, compared to much fewer number of Photons hitting the film of the 35mm SLR.

I think you are confusing some issues, within your "digital" postulations, in that you believe that the digital sensor is a special recording medium which requires different rules.

I might be misunderstanding your theory, though.

WW

[facts_please]

Colin:
... the lens in the larger format (film) camera will require a larger absolute aperture diameter (=bigger lens), and have inferior depth of field. Nothing is free - you need more total photons for the larger format, and you pay for that with a larger lens.

William:
... it takes an HUGE chunk of glass for get an F2.8 lens to cast an image circle on the negative of a 5x4 view camera, and a much smaller chunk of glass to get an F2.8 lens to cover the neg of a 35mm SLR.

Clearly we agree on that point.


I think the most productive way forward is for me to ask you (and others interested in thread) if you agree with this statement I made earlier. You may need to refer back to the rest of my posting for further details.

By any meaningful measure, my F2.0 point and shoot camera is slower than the F2.8 SLR - it requires a brighter subject for the same image noise.

Cheers, Colin

[Dave Humphries]

Ok, to add a slightly different slant on this, two points came to mind reading this thread;

1) I can see an equivalent argument to wanting to know focal length converted to 35mm standard, my 6.2 - 66.7mm lens means nothing to most of you (or me), but if I say it does 28 - 300mm, everyone knows what I mean. In terms of depth of field, the NF equivalent aperture (sorry, f-number) would again mean more to most people, although the need to compare DoF seems rather less useful, but maybe that's because there wasn't an easy way before.

2) NF numbers would be a nightmare for most lens manufaturers, but especially Sigma, Tokina, Tamron, etc. because their lens may go onto different crop factor bodies (1.5, 1.6) and even the big boys who make cameras too use c/fs of 1, 1.3 and 1.5/1.6.

So, what I think would be far more useful is if camera manufacturers expressed/advertised all camera bodies with a published crop-factor number, then those that know could apply this to focal length and/or f-number as they see fit. This would be especially helpful for compacts and bridge cameras where people like us* might want to get a P&S as a go-anywhere back up to their DSLR, comparing different models in terms they can calculate and equate to their DSLR, for F/L, DoF and noise performance. For most fixed lens cameras, it takes some searching to find the sensor size and even then it's expressed as something meaningless like "1/1.7" or "1/2.5" inches, which are about as helpful as a chocolate teapot.

* not that we're that important to the camera manufacturers

Just my take on it,

[facts_please]

Re normalizing f-numbers to 35mm equivalent, DaveH wrote:

I can see an equivalent argument to converting focal length to 35mm standard, my 6.2 - 66.7mm lens means nothing to most of you (or me), but if I say it does 28 - 300mm, everyone knows what I mean.

That's a neat analogy. Nobody objects to normalizing focal lengths to 35mm equivalent.

NF numbers would be a nightmare for most lens manufaturers ....
So, what I think would be far more useful is if camera manufacturers expressed/advertised all camera bodies with a published crop-factor number .....

Agreed. Unlike me, you keep your feet firmly on the ground, making well considered suggestions that would be useful in the real world. You also clearly understand what I am on about, which is heartening.

[facts_please]

I think you are confusing some issues, within your "digital" postulations, in that you believe that the digital sensor is a special recording medium which requires different rules.

This cuts to the heart of the matter. Indeed, digital sensors DO require slightly different rules, and NF-numbers provide those rules. A digital sensor is similar in many ways to film, but it is not quite the same. In particular, the concept of ‘exposure’ and ‘ISO’ do not exactly transfer from film to digital camera.

(a) Using film
For film, the ‘graininess’ (= image noise) is a function only of the film that is physically within the camera, so image noise is always the same for film of the same ISO and type. Typically, more sensitive film (=high ISO), produces a grainier image, and requires less photons/area for correct exposure. Also, for a given ISO, correct exposure is always obtained with the same number of photons/area, regardless of format size. To state that another way, for a given shutter speed and ISO, the required intensity of light striking the film is always the same, regardless of format size. That is why f-numbers are so well suited to film cameras, because all lenses with the same f-number produce the same light intensity (photons/area/time) on the film.

(b) Using digital sensor
There is an analogy of sorts with digital sensors, but the situation is NOT the same. With digital cameras, image noise is most definitely NOT the same for different cameras set on the same ISO!! Everybody knows that a compact point and shoot will produce a much noisier image on ISO1600 compared to an SLR on ISO1600. Apparently very few people have noticed that this is not at all analogous to film cameras. Just quietly, adoption of NF-numbers re-defines the meaning of ISO on digital cameras to be analogous with film cameras. That is, if NF-numbers were adopted, all digital cameras set to the same ISO, regardless of sensor size, would receive the same number of photons on their sensors, and thus produce similar image noise. Another plus for NF-numbers.


Adoption of NF-numbers also re-defines the meaning of ‘exposure’ in a more logical way as it applies to digital sensors, in a way that makes the term more analogous to it’s meaning with film.

For film, exposure is proportional to the total number of photons/area that are dumped onto the film, independent of film format size. Furthermore, if the film ISO is doubled, the required number of photons/area dumped on the film is halved.

When NF-numbers are adopted, exposure is proportional to the total number of photons that are dumped onto the sensor, independent of sensor size. Furthermore, if the camera ISO setting is doubled, the number of photons dumped on the sensor is halved. For digital sensors, this is a more logical way of defining exposure, because image noise is a function of the total number of photons.

First, I think that f/#'s were mainly created to help calculate exposure ....

This is very important. F numbers are not outdated and they perform a vital role.
IMO, having “new F numbers” would be silly; and, moreover using F numbers we have, is vital – F numbers are part of the universal Trilogy of Exposure, all Photographers can immediately understand.

Aperture (“F Number”) / Shutter Speed (Exposure Time) / ISO (Sensitivity) – these three numbers form the base language of the technical premise of all image making – we can communicate so much with just three numbers - camera size, film size or sensor format has nothing to do with the communication of information via the use of these three numbers.

Many people pointed out that traditional F-numbers are indispensible for calculating correct exposure, and that there is no way that photographers will discard the elegant trilogy of exposure time, f-number and ISO. I totally agree.

Fortunately, when NF-numbers are used with digital cameras, the interplay between these 3 variables still works out just fine, except more powerful, and more closely analogous to film cameras.

WilliamW presented an excellent practical scenario demonstrating the use of f-numbers. As time permits I will work through the same example using NF-numbers, and show the procedure is just as simple, but more powerful.

[Henry Peach]

You go ahead and start the revolution, my friend. If your ideas make sense to others then they'll adopt them. I'll just stick to what I understand and what works for me. I can't make heads or tails of what you are talking about.

[William W]

Many people pointed out that traditional F-numbers are indispensible for calculating correct exposure, and that there is no way that photographers will discard the elegant trilogy of exposure time, f-number and ISO. I totally agree.

Fortunately, when NF-numbers are used with digital cameras, the interplay between these 3 variables still works out just fine, except more powerful, and more closely analogous to film cameras.

WilliamW presented an excellent practical scenario demonstrating the use of f-numbers. As time permits I will work through the same example using NF-numbers, and show the procedure is just as simple, but more powerful.

No need for my benefit.

I have re read the whole thread.

I get your point - it came clear to me when you mentioned noise differences between digital cameras set at the same ISO.

I think your argument would be better served:

1. If you referred to the number of photons hitting the sensor as the "NP" number.

2. Keep the F number as is (a pure lens-function geometric relationship)

3. Draw a table, or similar, of relationships, between lens's F numbers and Sensor's NP numbers, for all the different sensor sizes, at all the different ISO's.

If I follow your line of thought correctly, you will find your argument, and resultant values, will not only be dependent upon the lens's F number being used and sensor size and the ISO; but also, just to throw another two variables in to the mix: it will also be dependent upon the sensor TYPE.

And also the ANGLE at which the photons strike the sensor – (so lens design comes into play too)


***


BTW: your argument about same ISO film having the same grain is flawed.

Grain and Acutance is dependent upon many other variables than ISO, some of which include:

. Reciprocity of Exposure and allowance for it

. Developer used

. Development time and temperature (for exposure at the same ISO)

. Age (of the film) - and how it was kept

. Temperature (of the film) at the time of exposure

. Type of Film (e.g. ISO100 Pan Film is not the same as ISO100 Ortho Film etc).


***

Also I believe you will find there are studies already completed on the sensitivity / noise relationships of various sensors at various ISO including all the variables previously mentioned in this discussion.

These could be extrapolated directly as a linear function to a particular lens's F number variance for any sensor at a particular ISO. If one wanted to pick nits, then for any particular lens, one would derive the T-stop for every F-stop and use the T-Stop value, which is the actual measurement of light, taking into account loss within the lens.

(T-Stops are commonly used in cinematography - less so now, with the superior lenses used, though light transfer is never 100%, through any lens.)

Canon, Nikon and Sigma, amongst others have included these facts to a greater or lesser extent in their White Papers, on Sensor Design and Functionality and the latter, in Papers discussing the Sigma "DG" and "DC" Lens Design protocols.


WW

[Raycer]

And here I thought, f-number is f-number. I thought it is just like any other dimensionless number where it is a physical ratio. Normalizing a non-dimensional number? hmmm... why? not too sure about the whole theory.

remember, sunny 16 rule works for any format.

[facts_please]

Henry wrote:

You go ahead and start the revolution, my friend. If your ideas make sense to others then they'll adopt them. I'll just stick to what I understand and what works for me. I can't make heads or tails of what you are talking about.

Sorry about that. I hope you are able to look at the examples that I shall give.

Practical, real world examples are the best way to explain the simplicity and power of NF-numbers. If real examples can't make the advantages of NF-numbers clear, then the concept of NF-numbers should be discarded.

In real world usage, NF-numbers work almost identically to F-numbers, which should not be too surprising, as they are exactly the same thing for a 35mm camera.

If you are already familiar with how to use f-numbers, exposure time and ISO (and you obviously are) then the transition to NF-numbers is painless. All the reciprocity stuff works exactly as it did before.

I hope you find time to look at the forthcoming practical examples. Then, you be the judge.

Colin

PS. My discussions/recommendations are actually on two levels.

On the first level, the practical level, I point out that converting f-numbers to 35mm equivalent is very useful for comparing the effective 'speed' of cameras with different sizes of sensor. This would be especially useful for many of us when considering a camera purchase.

The second level is really just a thought exercise, for digital cameras calibrated in NF-numbers will never be manufactured. However, I personally find it interesting to consider how everything would 'work out' in a hypothetical world where our digital cameras are calibrated in NF-numbers. As it turns out, everything seems to work out rather well, with some real advantages. When a thought exercise like this is pursued, much is learned along the way, new ideas and ways of looking at things are introduced .... and I think that is all to the good.

[facts_please]

Colin:
WilliamW presented an excellent practical scenario demonstrating the use of f-numbers. As time permits I will work through the same example using NF-numbers, and show the procedure is just as simple, but more powerful.

No need for my benefit. I have re read the whole thread. I get your point

I will still show some real life examples, because for many others that is the only way that my ideas will make any sense. I appreciate the considerable time you must have spent re reading the whole thread ....

Grain and Acutance is dependent upon many other variables than ISO ...

Yes, of course, although this does not invalidate the usefulness of NF-numbers, any more than it invalidates the usefulness of F-numbers. It is equally true that image noise from a digital sensor depends to some extent on factors other than the number of incident photons, so the analogy that I drew between film cameras and digital cameras is as strong as ever.

If I follow your line of thought correctly, you will find your argument, and resultant values, will not only be dependent upon the lens's F number being used and sensor size and the ISO; but also, just to throw another two variables in to the mix: it will also be dependent upon the sensor TYPE.
And also the ANGLE at which the photons strike the sensor – (so lens design comes into play too)

If I understand you correctly, then yes, the image noise does to some extent depend on variables other than the number of incident photons, just as film grain depends on variables other than the film ISO. This does not invalidate the usefulness of NF-numbers, any more than it invalidates the usefulness of F-numbers.

To state what this all means in practice, when NF-numbers are adopted, all digital cameras set to the same ISO number will produce images with approximately the same image noise, with the differences being due to sensor differences in dead area, QE, effectiveness of the microlenses, electrical noise in shadow areas, and so on.

When F-numbers are adopted, all film cameras using film with the same ISO rating produce images with (very) approximately the same image noise (grain), with the differences being due to film differences such as the type and brand of film, development variables, and so on.

This is all just as I stated previously, that adoption of NF-numbers has the effect of re-defining the meaning of ISO on digital camera to be more analogous with film cameras.

If photographic enthusiasts with a film background were unbiased, then I would have thought that this subtle change in the meaning of ISO on digital camers, to be more analogous with film cameras, would be regarded as an interesting idea, maybe even a useful idea, but in reality I suspect that the vast majority of photographers have not thought about these issues at such depth, and never realised in the first place that ISO on a digital camera has a slightly different meaning.

Purely for interest, this re-defining of ISO on digital cameras is an inevitable consequence of adopting NF-numbers - if you do one you must do the other, otherwise the reciprocity relationships become dimensionally (and functionally) wrong. For the same reason, the meaning of exposure must be modified as I described. I gave a fair bit of thought to all this before starting the thread.

[facts_please]

The following ‘quote’ is a typical question that could be asked by someone starting out in photography.

Newbie question:
I have noticed that many of the shots from my credit-card-sized camera have a noisy, speckled appearance, and my friend told me this is because there is not enough light getting to the sensor. That seems to make sense, because the shots taken in bright daylight seem OK.

I want to buy a better camera that suffers less from this problem. I have narrowed down my choice to two models of camera, which have the other features I want, and are within my price range.

One of these has an F/2.0-2.5 lens, and the other an F/2.8-4.0 lens, and both cover the same zoom range.

My friend says that lower f/numbers mean a faster lens that will be better for low light shooting, so I should buy the camera with the F/2.0-2.5 lens. Is this correct?

I confidently predict that a high proportion of ‘average’ photographers would not hesitate to concur with the friend’s advice, and recommend the first camera with the F/2.0-2.5 lens.

The correct answer is that, on the basis of the information given, it is impossible to say ANYTHING AT ALL about which camera would perform better in low light.

Raycar:
Normalizing a non-dimensional number? hmmm... why?

Hmmm. I trust that the example above make it clear that f-numbers on digital camera are commonly misunderstood, and create confusion. Quoting a lens f-number on a digital camera, without reference to sensor size, gives no information about low light performance, or depth of field. That is why there is an incentive for normalizing non-dimensional F-numbers.



Now I will repeat the above example, in a hypothetical world where NF-numbers are printed on the camera, alongside the F-numbers. NF-numbers give meaningful information about the performance of the lens system, taking into account the effect of sensor size, and can therefore be directly compared between cameras.

I want to buy a better camera that suffers less from this (low light) problem. I have narrowed down my choice to two models of camera, which have the other features I want, and are within my price range.

One of these cameras is rated at NF 9.4-11.8 , and the other camera, with standard lens, is rated at NF 4.2-6.0, and both cover the same zoom range.

My friend says that lower NF-numbers mean a faster lens system that will be better for low light shooting, so I should buy the camera rated at NF 4.2-6.0 Is this correct?

Now the decision is a no-brainer – all you need to do is compare NF-numbers, end of story. The friend is right, the second camera is significantly better at NF4.2 vs. NF9.4, an improvement in NF-number by over a factor of two, meaning that over 4 times as much light can be collected and dumped on the sensor. Note that traditional interpretation of F-numbers gave the wrong result, as the lens on the faster camera actually had slower F-numbers.

Depth-of-field comparison between the cameras is also completely straightforward using NF-numbers. The higher the NF number, the greater the depth of field – end of story. Again, looking at F-numbers in this example gives the wrong result.

This is the simplest example I could think of, intended as much as anything to illustrate the problem with F-numbers, as I suspect that many people didn’t understand the problem, much less the solution which NF-numbers provide.

Subsequent examples will explore the operation and benefits of NF-numbers in more depth.

PS. Just for interest, the first camera in the given example was a reasonable quality point and shoot Canon G2, while the second is a typical APS-C consumer grade SLR.

[Raycer]

Okay, I'm starting to see what you are trying to get at.
But if at the end of the day, the goal is to compare IQ at low light, why don't you direct them to sample comparison sites? Afterall, the end result is what you are after. I'm sure there are small sensors that has better low light performance than large sensors.