Why are fixed apertures desirable?

[Ksanti]

Obviously having a constant f/2.8 aperture is better than a varied f/2.8-f/4 or worse, but given the aperture f/2.8 is given as how big the aperture is in comparison to the focal length, surely the same physical aperture (rather than f/ number) would mean a changing f/ number along a zoom range.

The 70-200 f/2.8s for example:

At 200mm, f/2.8, the entrance pupil's diameter is 200/2.8, or roughly 70mm.
At 70mm, f/2.8, the entrance pupil's diameter is 70/2.8, or 25mm.

A 70mm entrance pupil at 70mm focal length would give f/1.0.

Clearly there is something wrong with my thinking, or the optical performance of these lenses falls off of a horrible cliff once they start being that far open at those lengths; if my logic wasn't flawed then the pro lenses would be 70-200 f/1-2.8s rather than f/2.8. I'm just looking for why it is. The glass is there and the entrance pupil is large enough, so why does it not open as much at the short end as at the long end?

 

[StewartR]

The glass is there and the entrance pupil is large enough, so why does it not open as much at the short end as at the long end?

the optical performance of these lenses falls off of a horrible cliff once they start being that far open at those lengths

 

[woof woof]

At 200mm, f/2.8, the entrance pupil's diameter is 200/2.8, or roughly 70mm.
At 70mm, f/2.8, the entrance pupil's diameter is 70/2.8, or 25mm.

AFAIK this is why the f number system was invented, to equal things out. The actual size of the hole will vary with the focal length but the amount of light hitting the sensor will be the same, f2.8, so in effect 70/2.8 will equal 200/2.8.

If you Google there are plenty of explanations on this, AFAIK the f number system isn't 100% accurate but it's good enough.

 

[Ksanti]

AFAIK this is why the f number system was invented, to equal things out. The actual size of the hole will vary with the focal length but the amount of light hitting the sensor will be the same, f2.8, so in effect 70/2.8 will equal 200/2.8.

If you Google there are plenty of explanations on this, AFAIK the f number system isn't 100% accurate but it's good enough.

The f/ number system represents how much light is hitting the imaging medium, not how big the physical pupil or aperture is. My question was whether the reason fixed apertures were desirable was because any wider than f/2.8 (which would still be no bigger than f/2.8 at the long end) on these lenses invariably meant that the optical performance completely tanked.

My question has been answered, I guess.

 

[StewartR]

I think you answered your own question.

Making lenses with very low f/ numbers is very difficult, which makes them very expensive. There are only a couple of f/1.2 lenses available - the Canon 50mm and 85mm - and they are many times more expensive than the f/1.8 equivalents. At least they are both very good optically. Canon used to make a 50mm f/1.0, but that was insanely expensive and not actually that good. All this suggests that f/1.0 really is pushing the boundaries of what is possible (or at least what is possible at a reasonable price).

Zooms are inherently more complex to design, and the fact that there are are no zooms anywhere faster than f/2.8 must be related to this. If you gave Canon's or Nikon's boffins an unlimited budget and challenged them to make a 24-70mm f/1.0-2.8 zoom, I really don't know whether they would be able to do it.

 

[Adam_M]

Zooms are inherently more complex to design, and the fact that there are are no zooms anywhere faster than f/2.8 must be related to this.

Olympus have two f/2.0 zooms

http://www.olympus.co.uk/consumer/dslr_13093.htm

But that's only because their sensor is tiny, but it counts!

You could have an F/2.0 zoom on 35mm, but you would never be able to carry it.

 

[Bristolian]

The f/ number system represents how much light is hitting the imaging medium, not how big the physical pupil or aperture is.

This is COMPLETELY incorrect. The f/number system is purely a measure of the aperture diameter expressed as a ratio to the focal length ... nothing more and nothing less.

The amount of light striking the sensor is determined by the aperture and shutter speed in combination. You can get just as much (or even more) light hitting the sensor at f/32 as you can at f/1/4 ... you just have to adjust the shutter speed accordingly.

Note, I don't include ISO in the above because that doesn't affect the amount of light hitting the sensor, just the way the camera interprets it.

 

[Nuffles]

Another benefit of fixed aperture is (effectively) fixed exposure over the length of your zoom. If you're working in low light with the lens wide open your expose stays the same wherever you are in the zoom range. This is great for if you're shooting in manual or if you just want full use of your zoom without constantly juggling ISO/exposure compensation.

 

[gunnar]

This is COMPLETELY incorrect. The f/number system is purely a measure of the aperture diameter expressed as a ratio to the focal length ... nothing more and nothing less.

I'm sorry, but take any lens (esp long tele) and measure the aperture (not entrance pupil) diameter - you will be surprised.

Focal length to aperture diameter ratio works only with single lens.

With multiple lens optical system though, F number expreses the ammount of light single lens with the same focal length would have let it, given the same shuuter speed.

 

[Ksanti]

This is COMPLETELY incorrect. The f/number system is purely a measure of the aperture diameter expressed as a ratio to the focal length ... nothing more and nothing less.

Amount of light hitting the imaging medium over a given period of time.
Happy? :shrug:

...

Yes, I understand that it's difficult but I always felt that the majority of the difficulty was because of the sheer amount of glass needed. Now of course zooms are far more complicated and as such image quality would be horrific at f/1.4 etc., but I would imagine some designs could feasibly offer very usable IQ at slightly larger apertures e.g. f/2.5, f/2.0 at the short end, given the quality of zooms such as the Canon 70-200 f/2.8L even wide open.

 

[woof woof]

The f/ number system represents how much light is hitting the imaging medium, not how big the physical pupil or aperture is. My question was whether the reason fixed apertures were desirable was because any wider than f/2.8 (which would still be no bigger than f/2.8 at the long end) on these lenses invariably meant that the optical performance completely tanked.

Oh dear oh dear.

 

[CT]

Amount of light hitting the imaging medium over a given period of time.
Happy? :shru

You're still wrong - Bristlolian was absolutely right. The amount of light hitting the imaging medium over a given time is a function of the aperture and shutter speed.

There's nothing wrong with large apertures and long focal length lenses - in fact it's very desirable to have a fast long lens. The fact is it gets very difficult and expensive to make long lenses with large maximum apertures. You don't see any 500 and 600mm lenses with max apertures larger than f/4 because to go larger would be expensive (grinding and polishing lenses beyond a certain size is very difficult) and adds enormously to the size and weight of the lens.

As a case in point Canons largest max aperture (physically) is the 400mm 2.8 which isn't their longest lens but it's the heaviest of all the super tele lenses purely by virtue of that larger aperture. You'd probably struggle to pick up a 500 or 600mm lens with the same aperture.

 

[woof woof]

All this suggests that f/1.0 really is pushing the boundaries of what is possible (or at least what is possible at a reasonable price).

A couple of manufacturers have managed to make good f1- lenses. I have a little Voigtlander which is no slouch and there are a couple of others knocking about too at a much higher price although some manufacturers saw f1 as a marketing gimmic when really the quality wasn't there.

 

[CT]

Well if you can find a Canon f/1 today (not to mention afford it) you'd be doing well. The thing with that lens is that whilst it was a pretty crap performer wide open, it also enabled a brighter viewfinder image and easier focusing even if you needed to realistically stop down a bit, so for the low light user who absolutely had to get the shot under all conditions I suppose it was worth the price tag.

 

[woof woof]

Indeed. And what's crap is for the user to decide and IMVHO it's nice to have the option in the bag even if it's not actually, really, great at the setting. Good enough will do and I'm pretty sure that it would be for me

 

[gunnar]

You're still wrong - Bristlolian was absolutely right. The amount of light hitting the imaging medium over a given time is a function of the aperture and shutter speed.
.

Let's make things clear. What we are talking about is illuminance.
Illuminance is a function of F-number, full stop.
Amount of light hitting the sensor is function of illuminance and exposure time (shutter speed).

Now, with single lens design (i.e. simple optic lens) F number is a focal length to aperture diameter ratio.
However this doesn't work with multiple lens desigh, as some of the elements have positive magnification, some are negative, etc etc etc.
Therefore for such case F number answers the question "Given single lens with the same focal length what F number will produce the same illuminance "

You can prove it by taking a long telephoto lens and meassuring it's aperture (not front element) diameter. You'll find that for example 400/2.8L has aperture diameter FAR less than 142mm

 

[CT]

Let's make things clear. What we are talking about is illuminance.
Illuminance is a function of F-number, full stop.
Amount of light hitting the sensor is function of illuminance and exposure time (shutter speed).

Now, with single lens design (i.e. simple optic lens) F number is a focal length to aperture diameter ratio.
However this doesn't work with multiple lens desigh, as some of the elements have positive magnification, some are negative, etc etc etc.
Therefore for such case F number answers the question "Given single lens with the same focal length what F number will produce the same illuminance "

You can prove it by taking a long telephoto lens and meassuring it's aperture (not front element) diameter. You'll find that for example 400/2.8L has aperture diameter FAR less than 142mm

You start off by saying that the f-number to focal length ratio doesn't work with a zoom then go on to use a prime as an example of why it doesn't apply?

I don't have a 400 2.8L but I've just measured the front element of my 500mm f/4L as accurately as I can and I get 128mm......128 X 4 = 512. :shrug:

 

[DemiLion]

You don't see any 500 and 600mm lenses with max apertures larger than f/4 because to go larger would be expensive (grinding and polishing lenses beyond a certain size is very difficult) and adds enormously to the size and weight of the lens.

Sorry to be a PITA Cedric, but....

 

[CT]

Sorry to be a PITA Cedric, but....

I knew someone would post that. A pretty extreme example though. Not mangy people would want to cart that around.

 

[arad85]

Clearly there is something wrong with my thinking,

Good question, but the assumption you have made is that the aperture stays stationary.

Take a zoom lens off a camera and hold it up to the light. Look at it from the sensor end and zoom in. The "aperture" stays constant. Turn the lens around and zoom in. The aperture changes in size. This change in size is a combination of two things:

1) Magnification making it look bigger
2) The size being made bigger by the mechanism as it zooms in/out.

The reason: because it is easier and cheaper to build like this.

 

[HoppyUK]

Let's make things clear. What we are talking about is illuminance.
Illuminance is a function of F-number, full stop.
Amount of light hitting the sensor is function of illuminance and exposure time (shutter speed).

Now, with single lens design (i.e. simple optic lens) F number is a focal length to aperture diameter ratio.
However this doesn't work with multiple lens desigh, as some of the elements have positive magnification, some are negative, etc etc etc.
Therefore for such case F number answers the question "Given single lens with the same focal length what F number will produce the same illuminance "

You can prove it by taking a long telephoto lens and meassuring it's aperture (not front element) diameter. You'll find that for example 400/2.8L has aperture diameter FAR less than 142mm

It's the focal length divided by the diameter of the entrance pupil - which is actually quite hard to measure externally.

Sometimes this coincides with the diameter of the aperture, or size of the front element, eg a typical 50mm f/1.8 and some longer primes, but not always. Take Nikon/Canon 50 1.4s and compare to the Sigma 50 1.4 - Sigma's front element is much bigger.

With zooms, it's pretty much impossible to physically measure anything meaningful in terms of f/number. My 17-40L 4 has a front element about 62mm across...

None of this answers the OP's question, though I think Stewart's posts did. Another quirk is that zooms with a variable max aperture often change the diaphragm with focal length.

 

[andya700]

Here is a link to an f1.1 Nokton lens

http://www.luminous-landscape.com/reviews/lenses/50mm-nokton.shtml

Minolta produced a couple of highly regarded f1.2 lenses back in the day - the Rokkor 50 1.2 and the Rokkor 58 1.2.

 

[gunnar]

You start off by saying that the f-number to focal length ratio doesn't work with a zoom then go on to use a prime as an example of why it doesn't apply?

You can measure zoom's aperture as well,

I don't have a 400 2.8L but I've just measured the front element of my 500mm f/4L as accurately as I can and I get 128mm......128 X 4 = 512. :shrug:

But why did you measure the front element? Try to measure aperture (diaphrapgm).

 

[gunnar]

My 17-40L 4 has a front element about 62mm across...

Here we go. This is precisely what I'm tlking about- classic optics formula focal_length/aperture_diameter doesn't work with multiple lens design

 

[Dale_tem]

The f value relates to the entrance pupil. This is the size of the aperture as it appears through the front of the lens. Measuring the front element will not help you with the f value.

500/4 = 125mm, so from the front the aperture hole should measure 125mm across. From the back of the lens it will be much smaller.

http://en.wikipedia.org/wiki/Entrance_pupil

 

[ausemmao]

Here we go. This is precisely what I'm tlking about- classic optics formula focal_length/aperture_diameter doesn't work with multiple lens design

The front element is not the entrance pupil.


Faster than 2.8 zooms are made. More than 3x zoom 2.8 or faster lenses are made. They're just not all that practical for photography, given the resultant weight.

Here is a 18-85mm (4.7x) T2.0 (so likely f1.7) zoom. You could mount this to your camera with a PL adapter. At 7kg you probably wouldn't want to handhod it though

Or this super 24-70 +70-200 - a 24-180 T2.6 (bear in mind a 70-200 f2.8 is usually T3.1-T3.4). That's 9kg though.

In case you were wondering, a T-stop is F-stop corrected for light loss through the elements. Something that photographers can ignore, but video people care about more. And also for reference, a 400 f2.8 is around 5kg.

 

[HoppyUK]

The front element is not the entrance pupil.


Faster than 2.8 zooms are made. More than 3x zoom 2.8 or faster lenses are made. They're just not all that practical for photography, given the resultant weight.

Here is a 18-85mm (4.7x) T2.0 (so likely f1.7) zoom. You could mount this to your camera with a PL adapter. At 7kg you probably wouldn't want to handhod it though

Or this super 24-70 +70-200 - a 24-180 T2.6 (bear in mind a 70-200 f2.8 is usually T3.1-T3.4). That's 9kg though.

In case you were wondering, a T-stop is F-stop corrected for light loss through the elements. Something that photographers can ignore, but video people care about more. And also for reference, a 400 f2.8 is around 5kg.

Interesting examples. But those lenses won't cover FF, which makes optical design much easier. Are they cine half-frame 35mm?

Olympus can only make an f/2 zoom because they only have to cover 4/3rds format, which is one quarter the area of full frame.

 

[ausemmao]

Interesting examples. But those lenses won't cover FF, which makes optical design much easier. Are they cine half-frame 35mm?

Olympus can only make an f/2 zoom because they only have to cover 4/3rds format, which is one quarter the area of full frame.

They cover full frame
They're the lenses used in the Zacuto shootouts.

 

[HoppyUK]

They cover full frame
They're the lenses used in the Zacuto shootouts.

Spec says they have a field of view of 53 degrees at 24mm. On full frame, 24mm is 84 degrees. So they can't cover full frame, unless I'm missing something.

 

[ausemmao]

Spec says they have a field of view of 53 degrees at 24mm. On full frame, 24mm is 84 degrees. So they can't cover full frame, unless I'm missing something.

The 18-85 is definitely full frame, because it's used on the 5D2. Can't say for sure about the 24-180, but I'm fairly certain it is too.
One thing to remember is that the AoV is set by the the combination of the lens and the film/sensor behind the lens, and they will quote for 24x18 as it's still (just, I think) dominant for cinema. Video people are weird

 

[HoppyUK]

The 18-85 is definitely full frame, because it's used on the 5D2. Can't say for sure about the 24-180, but I'm fairly certain it is too.
One thing to remember is that the AoV is set by the the combination of the lens and the film/sensor behind the lens, and they will quote for 24x18 as it's still (just, I think) dominant for cinema. Video people are weird

Well yes, when cine folks say 35mm, they're referring to the film, not the format. Since the film runs vertically through the gate, it's what we call half-frame, 24x18mm as you say, and not the 24x36mm that we know as full frame.

Specs for the 18-55 also sugest it's half-frame, and though I don't doubt it's used on a 5D2, it won't cover the whole sensor (check the angle of view). Lenses like that would be extremely difficult to make for FF - the smaller image circle makes optical life very much easier.

 

[ausemmao]

Well yes, when cine folks say 35mm, they're referring to the film, not the format. Since the film runs vertically through the gate, it's what we call half-frame, 24x18mm as you say, and not the 24x36mm that we know as full frame.

Specs for the 18-55 also sugest it's half-frame, and though I don't doubt it's used on a 5D2, it won't cover the whole sensor (check the angle of view). Lenses like that would be extremely difficult to make for FF - the smaller image circle makes optical life very much easier.

It covers the whole sensor - if it didn't then there'd be vignetting on this test for the 5D2 and the Arri Alexa:

http://www.zacuto.com/the-great-camera-shootout-2011/episode-one

3-4 minutes in they discuss the setup and lens used, later on there are the various tests. Which do show why DSLRs have become so popular for video. They're not quite as good as the Alexa or others in that price bracket, but at 1/30th the price...

 

[HoppyUK]

It covers the whole sensor - if it didn't then there'd be vignetting on this test for the 5D2 and the Arri Alexa:

http://www.zacuto.com/the-great-camera-shootout-2011/episode-one

3-4 minutes in they discuss the setup and lens used, later on there are the various tests. Which do show why DSLRs have become so popular for video. They're not quite as good as the Alexa or others in that price bracket, but at 1/30th the price...

Well, I don't know then. I'm just equating the Fuji specs for angle of view and focal length, which doesn't match full frame but, without doing the calcs, I'm guessing it's close to half-frame.

So either those specs are misleading in some way, or they've cropped the vignetting out with the 5D2?

I'll catch the video link later - looks interesting

 

[ausemmao]

Well, I don't know then. I'm just equating the Fuji specs for angle of view and focal length, which doesn't match full frame but, without doing the calcs, I'm guessing it's close to half-frame.

So either those specs are misleading in some way, or they've cropped the vignetting out with the 5D2?

I'll catch the video link later - looks interesting

Yeah, the difficulty goes away if you assume they're quoting FoV for 24x18 as a quick reference for their customers (who will all be used to those numbers, like stills cameras quote 35mm equivalent for crop and P&S sensors), not as an absolute maximum FoV.

It is. It also shows why the 1DX could be very good for video as it seems to address a good number of the shortcomings that were highlighted.

 

[HoppyUK]

Yeah, the difficulty goes away if you assume they're quoting FoV for 24x18 as a quick reference for their customers (who will all be used to those numbers, like stills cameras quote 35mm equivalent for crop and P&S sensors), not as an absolute maximum FoV.

It is. It also shows why the 1DX could be very good for video as it seems to address a good number of the shortcomings that were highlighted.

Idea! They zoomed the lens, which they must have done to match the different crop factors of all the various cameras anyway. Towards the long end, those Fuji lenses would probably cover full frame okay, and it looks like they were testing cameras for dynamic range and ISO rather than looking at the lenses?

 

[ausemmao]

Idea! They zoomed the lens, which they must have done to match the different crop factors of all the various cameras anyway. Towards the long end, those Fuji lenses would probably cover full frame okay, and it looks like they were testing cameras for dynamic range and ISO rather than looking at the lenses?

Could be, but I'm inclined to think that's just far too complicated. I think they cover it natively. Look at the TS lenses as an example - they're F/EF mount, designed for 35mm, and their AoV if stills companies stated it would be x degrees. But put them on an adapter, we know they will cover medium format, providing a wider AoV. That's what I think is happening there. There's no need for that much glass otherwise (and my justification for that is the Olympus 70-200).

The other example are large format lenses - you use the same lens to cover 6x7 as you do 8x10 as the lenses are designed to ahve a massive image circle.

 

[HoppyUK]

Could be, but I'm inclined to think that's just far too complicated. I think they cover it natively. Look at the TS lenses as an example - they're F/EF mount, designed for 35mm, and their AoV if stills companies stated it would be x degrees. But put them on an adapter, we know they will cover medium format, providing a wider AoV. That's what I think is happening there. There's no need for that much glass otherwise (and my justification for that is the Olympus 70-200).

The other example are large format lenses - you use the same lens to cover 6x7 as you do 8x10 as the lenses are designed to ahve a massive image circle.

Yes, maybe. The answer is around here somewhere, and though I suspect only you and I are too interested, it's another example of how you can slice and dice these things up optically and trade one advantage for another.

The compromises that seem to go in opposite directions are things like you can't have a big zoom range and a low f/number, or you can't have a zoom with a very low f/number and big format coverage, if you want a very very low f/number, it's got to be a prime, and if you also want small and cheap, you can't have much at all!