Relative ISO vs Aperture Light Comparisons?

[waltgary]

The above title is not very accurate but I'm not sure how to improve it.


I'm getting the 5d II and am struggling between the 24-70/2.8 vs 24-105/4 IS. I've owned the 24-70 but needed IS, I like my 17-55/2.8and hesitate to go to f/4. I belive that the reduced noise of the 5d II (vs my 40 and 50d) may compensate for that?


In other words, is there a relative rule of thumb when comparing ISO to aperture settings? I see in the reviews on this site something to the effect that going from 2.8-4 lets in twice as much light? At what ISO change would this be equal so to speak? Low light sports are what I'm thinking, with or without a flash.


I realize these are somewhat stupid questions, but I really need IS (wobbly) and low light capability. If simply going from say, ISO 100 to 200 makes for the same relative exposure, I should be good with the f/4. The $390 price difference is significant too.


I suspect the obvious answer is to mount my 70-200/2.8 ISon the 5d and keep the 50d/17-55 EF-S. That's an appealling choice but I wouldn't mind lugging only one body around rather than two.


I'd like to thank all who take the time to read this uneducated and somewhat rambling post. Thedigitalpicture.com has been very helpful to me through the last few years.


gary

[Mark Elberson]

Someone please correct me if I am wrong but a stop is a stop is a stop. What I mean is the difference between f/2.8 and f/4 is a stop of light. The difference between ISO 100 and ISO 200 is a stop of light. The difference between 1/60 and 1/30 is a stop of light. If you change any one of those variables you will need to change one of the other ones to keep the correct exposure.


For your specific question regarding the 24-70 f/2.8 or the 24-105 f/4 you would need to shoot at ISO 200 with the f/4 as opposed to ISO 100 with the f/2.8 assuming the same shutter speed. Make sense?

[Jon Ruyle]

Not a stupid question.


My take is that since the 5d II sensor is 2.5 times as big as the 50d sensor, you should be able to use an f/4 lens on the 5d II with at least as fast shutter speeds as f/2.8 on the 50d (with similar nosie).


That is, iso 1000 on the 5d II should have the same signal noise ratio (so similar noise) as iso 400 on the 50d. If you use the f/4 lens on the 5D II at iso 1000, you'll have a slightly faster shutter speed than using an f/2.8 lens on the 50d.


Put another way, the 2.5 times as much light hitting the sensor more than compensates for the 1/2 as much light per unit area of the f/2.8 lens. (So the answer to your question, if I understand it, is yes).


I'm not claiming that all of this is actually true in real life, but that would be true other things being equal. It is probably pretty close to being true between 5d II and 50d.

[peety3]

A stop is a stop is a stop, at least with regards to satisfying the meter. If you had a 24-70/2.8 aimed at a still subject at f/2.8 and ISO 100, then replaced it with the 24-105/4IS and changed to f/4 and ISO to 200, you'd have the same shutter speed. You wouldn't have the same image, as elements of the photo might come into sharper focus because of the larger depth-of-field at f/4. With the same shutter speed, you've have the same subject motion blur, but would have no/less camera shake blur thanks to the IS if your shutter speed is between 8/focal-length and 1/focal-length (theoretically no blur) or slower than 8/focal-length (less blur).

[peety3]

In simpler terms, every time you open up/close down the aperture by a factor of 1.4, that's one stop. 4*1.4=5.6, so f/4 to f/5.6 is one stop less. 4/1.4 = 2.8, so f/4 to f/2.8 is one stop more. Every time you halve/double the shutter speed, that's one stop. Every time you halve/double the ISO, that's one stop.


If you're doing low-light sports, aperture is almost always key, and IS rarely saves the day. Why? You need fast shutter speeds for stopping the action, often 1/250th or better. Hand-held, most people can hold a camera down to shutter speeds of 1/focal-length (so 1/200th for your 70-200). Therefore your need for 1/250th shutter speed means your picture will natively be faster than the typical limits of hand-holdability. IS matters in the range of N/focal-length to 1/focal-length, where N is 4 for early IS units capable of 2 stops of effect, 8 for middle-vintage IS units capable of 3 stops of effect, or 16 for the latest lenses with units capable of 4 stops of effect (i.e. 2 to the X power where X is the number of stops of correction), as mentioned in my previous post.

[Jon Ruyle]

I agee that a stop is a stop, no argument there [:)]


My point was only that you shouldn't expect more noise with a f/2.8 lens on a 1.6fovcf camera vs a f/4 lens on a full frame camera, even if you crank up the iso on the ff to get the same shutter speed... the bigger sensor will- other things being equal- more than compensate for that difference.


As for depth of field, you'll actually get less dof at f/4 with the full frame camera as compared to f/2.8 on the with 1.6fovcf camera... assuming the shot is framed the same.


But your point is taken. There are a lot of other factors that will make the picture different- for better or worse.

[peety3]

Fair points, I was basing my comments on the OP's debate between the 24-70 and 24-105 for his pending 5D2 purchase.

[Daniel Browning]

I'm getting the 5d II and am struggling between the 24-70/2.8 vs 24-105/4 IS. I've owned the 24-70 but needed IS, I like my 17-55/2.8 and hesitate to go to f/4. I belive that the reduced noise of the 5d II (vs my 40 and 50d) may compensate for that?


Yes, it will completely compensate for it. The 24-105 f/4 IS has a wider aperture and more light gathering power than the 17-55 f/2.8 at every equivalent field of view and perspective. The reason for this goes back to the true definition of aperture, which is focal length divided by f-number. F-number is not aperture, though the two concepts are often confused.

I posted about this topic here:
http://community.the-digital-picture.com/forums/p/42/94.aspx#94

For example, the aperture on the 17mm at f/2.8 is 6mm. The L at 27mm f/4 is 7mm: that's wider aperture. The larger front element of the L points toward this fact too.

One important difference is that f/4 does not activate the f/2.8-only autofocus sensors.

The focal ratio (not aperture) is one stop narrower (f/4 vs f/2.8), so the light intensity per area is twice as dim. However, the total amount of light is more than double, thanks to the much larger area and wider aperture. So you can up the ISO or use -1 EC to get the same shutter speed as with f/2.8 and still collect a half-stop more light in total.

At f/2.8 you will be letting in twice the amount of light *per area*, but the total amount of light captured will be much smaller. For example, the lens on a very tiny 1/1.7" Digicam with a 7mm f/2.8 lens also has the same perspective/FOV as your 20mm, and it also captures the same amount of light per area, but the total amount of light falling on the sensor is much less. In fact, f/2.8 on a digicam is the same amount of light as f/14 on your 50D! (And f/19 on the 5D2.) It's no wonder digicams struggle in low light. The reason is sensor area.

In the same way, a full-frame f/4 lens focuses more light than an f/2.8 APS-C lens: it's just spread out over a wider area. So the full-frame camera has about a 1/3 stop advantage in low light at f/4: you could upgrade to the 5D2 and 24-105 f/4 and still get all the same DOF, noise level, and shutter speed.

But the amount of light project by the lens is only half the equation: the other part is the technology of the sensor. The 5D2 is using older sensor technology from the 1DS3, but it has an improved ADC (less read noise) and more permissive color filters (slightly higher sensitivity/QE). But the 50D has newer sensor technology such as gapless microlenses to add on top of the ADC/CFA changes, so the net result is that that 50D is slightly more efficient per area. I'd guess around 3-6%. So given a lens with the same aperture (not f-number), the 50D would be superior to the 5D2. But as I said above, the 24-105 has a slightly *wider* aperture: 1mm at the wide end. This difference is more than the 3-6% technology difference, so the 5D2 is better in the end.

Of course, since we're only talking about differeneces of 5-10%, it's better to say that they are effectively the same. So you shouldn't think that 24-105 f/4 on the 5D2 will be significantly less noise than 17-55 f/2.8 on 50D. But at least it will be the *same* (i.e. you're not taking a step backwards.) To really *improve* noise would require more aperture, that is, f/2.8 on the 5D2.

There is one web page that explains all of this, and more, in excrutiating detail.

http://www.josephjamesphotography.com/equivalence/




I see in the reviews on this site something to the effect that going from 2.8-4 lets in twice as much light?


Right.



At what ISO change would this be equal so to speak?


One stop would mean double the ISO (f/2.8 ISO 100 would become f/4 ISO 200, just as you stated).



I realize these are somewhat stupid questions


Not at all.



If simply going from say, ISO 100 to 200 makes for the same relative exposure, I should be good with the f/4.


I think you'll love it!



Someone please correct me if I am wrong but a stop is a stop is a stop.


I disagree.



What I mean is the difference between f/2.8 and f/4 is a stop of light.


I would not describe it as a stop of "light", but rather a stop of "light intensity per area". That is an important difference. F-number doesn't describe the total amount of light in the resulting image. If it did, there would be no difference between a tiny f/2.8 lens on my camera phone, digicam, APS-C, 35mm, medium format, or a 10-thousand pound telescope. All those are f/2.8, and they all have the same intensity of light per area, but the difference in area and total light makes for a very different image in the end.



I'm not claiming that all of this is actually true in real life, but that would be true other things being equal.


FWIW, I claim that it's true in real life.



A stop is a stop is a stop, at least with regards to satisfying the meter. If you had a 24-70/2.8 aimed at a still subject at f/2.8 and ISO 100, then replaced it with the 24-105/4IS and changed to f/4 and ISO to 200, you'd have the same shutter speed. You wouldn't have the same image, as elements of the photo might come into sharper focus because of the larger depth-of-field at f/4.


Actually, the depth of field projected by the 24-105 at f/4 is *thinner* than the same field of view projected by the 17-55 at f/2.8. It would require stopping down to f/4.5 in order to get the same DOF. The reason, as above, is physical aperture.

I hope that helps.

[peety3]

A stop is a stop is a stop, at least with regards to satisfying the meter. If you had a 24-70/2.8 aimed at a still subject at f/2.8 and ISO 100, then replaced it with the 24-105/4IS and changed to f/4 and ISO to 200, you'd have the same shutter speed. You wouldn't have the same image, as elements of the photo might come into sharper focus because of the larger depth-of-field at f/4.


Actually, the depth of field projected by the 24-105 at f/4 is *thinner* than the same field of view projected by the 17-55 at f/2.8. It would require stopping down to f/4.5 in order to get the same DOF. The reason, as above, is physical aperture.





Seriously, read the first post. The OP is buying a 5D2, and is debating between the 24-70L and the 24-105L. Let's give the OP some honest, logical, understandable advice. To do so, let's stop focusing on the 17-55 lens he has, and stick to the two lenses he's choosing amongst. Once you do that, I think you'll all have a hard time saying the 24-105/4 lets in more light than a 24-70/2.8. Clearly there's an assumption that the two lenses are being used on the same 5D2 camera that I don't think the OP has purchased yet.

[Daniel Browning]

Seriously, read the first post.


I did. Very carefully. Before I posted.






The OP is buying a 5D2, and is debating between the 24-70L and the 24-105L.





Did you read the first post? He said a lot more than just that. He even considered keeping the 50D with 17-55 because of the I.S., but would prefer not to.






Let's give the OP some honest, logical, understandable advice.





I did.






To do so, let's stop focusing on the 17-55 lens he has, and stick to the two lenses he's choosing amongst.





No, I think it's important to also discuss the lens he has. His principle question was whether the 24-105+5D2 and higher ISO could match the 50D+17-55. It can. So now he can choose between getting similar images to what he has now (24-105) or getting images with less noise but no IS (24-70).






Once you do that, I think you'll all have a hard time saying the 24-105/4 lets in more light than a 24-70/2.8.





I never said that. In fact, I said the opposite.

[Colin]

It's nice when nobody disagrees about the facts


Never gets in the way of an argument though...[:)]

[Keith B]

I fear Daniel Browning's knowledge.

[waltgary]

I fear Daniel Browning's knowledge.


Me too. But, his obvious knowledge and willingness to share w/me is damned nice.


I printed out this thread and actually studied it. Exceptionally good information, and exactly what I was after. While not extensive, my understanding of aperture and stops of light are considerably improved. BTW, what is the unit of a "stop of light"? Anyways, I've ordered a 5d II at 17Photo and the 24-70L. It came down to what I (think) I needed/wanted. More often than not I'm shooting fast moving dogs, etc so IS will be of no benefit there. Given the explanation that by going from f/4 to f/2.8 I can double the shutter speed at any given ISO, I'm going to see how it works. My old 24-70L was mounted on a 350 Rebel. Why on earth I sold it is still beyond me other than continuing in my tradition of buy high/sell low. Anyone interested in a lightly used 50d?


You guys hankering for a 24-70L/IS version should thank me. By purchasing today, I feel I'm guaranteeing the new model will be out shortly. At which point, I'll upgrade and spend a bundle more in the process.[:P] You're welcome.


Seriously, I'd like to thank again all those who helped enlighten me on this subject. gary

[Daniel Browning]

Thanks Keith and Gary.



BTW, what is the unit of a "stop of light"?


Photons. Somewhere on the order of 2e+12 photons per photograph, give or take a few orders of magnitude.

[Jon Ruyle]

BTW, what is the unit of a "stop of light"?


Photons. Somewhere on the order of 2e+12 photons per photograph, give or take a few orders of magnitude.





A "stop" is a factor of 2. (So a "stop", or even a "stop of light" is actually unitless)


So one stop more light means double the light. One stop less light means half the light. A lens is one stop faster if it lets in twice as much light. Moving the iso up one stop means doubling the iso. Stopping down one stop means closing the lens aperture so half as much light gets in. Etc.


I don't see how you can have a fixed number of photons and say that is a stop of light. I'm sure I'll be corrected if I'm wrong [:)]

[Daniel Browning]

I don't see how you can have a fixed number of photons and say that is a stop of light.


You're right; that wouldn't make sense. I guess I interpreted the question as "what is the unit of light?". I.e. half/double the photons per stop.

[Matthew Gilley]

This post has created more questions than answers for me. In the end, I agree that the 5dII is the better solution, but many of the points that Daniel makes do not appear logical. I suspect it is
a misunderstanding on my part. Below is my
understanding of this thread's topics - please correct any errors as I'm very eager to learn more.


I'm attempting to compare apples to apples, here are some ground rules for my conclusion at the end:


The 1.6x 50d has half the sensor area of a FF 5dII - therefore half as much light still produces the same light intensity at the 1.6x sensor likewise the FF requires twice as much light.


A 17-55 on a 50d is effectively a 27-88mm from a full frame perspective - DoF not considered. I mention this for the 1/shutter speed rule.


The 24-105 has the same 77mm filter size as the 17-55. This is arguably the same the front element, correct? I did not understand Daniel's comparison.


On a 50d much of the 24-105's (effectively a 38-168) image would be projected off the sensor - wasted light.


In Bryan's review of the 17-55 he says it is arguably the most hand holdable lens Canon makes, this excludes camera body and its' ISO perforamnce. The combination of wide aperture, IS, and wide angle are my understanding as to why this is true. This powerful combination is unmatched by any other canon lens for static subjects.


All of this relates to optics, which I felt Daniel touched on for the most part. I believe there is an light per area advantage of the 2.8 vs the 4.0. A very specific example will clearly show my misunderstanding or my understanding of this. Assuming equilavent ISO, equal shutter speeds, equal subjects, and the 5dII at 4.0 properly exposes the image. Will the 50d at 2.8 (remember the assumptions) overexpose, underexpose or equivalently expose the image? I belive it will overexpose it.


Ultimately, the question was - will the 5dII's ISO performance compensate for the light advantage the 2.8 has? Similarly asked - Is the 5dII's ISO performance that much better than the 50d's? Now the question becomes - How much better is "that much" better?


For reference, Bryan shows the 50d vs 5dII ISO comparison well in the 5dII review.:


http://www.the-digital-picture.com/Reviews/Canon-EOS-5D-Mark-II-DSLR-Digital-Camera-Review.aspx


When comparing the 5dII and the 24-105 4.0 3 stop IS to the 50d and the 17-55 2.8 3 stop IS - I believe that the ISO of the 5dII needs to be one stop better than the 50d's. Why? Because the 2.8 vs 4.0 focal ratio, the illusive "stop". In my opinion, the charts easily show the difference between sensor noise is at least 1 stop of improvement, but you should be the judge. Compare ISO 6400 of the 5dII with ISO 3200 of the 50d - the difference in noise is substantial on both the cloth and color card. I agree with Daniel's point that the efficiency of the 50d is greater than the 5dII, but this is a minor 5dII disadvantage comparatively to the large advantage the 5dII makes up with the ISO charts.


My ISO conclusion - 5dII's ISO performance easily outperforms the 50d's even though the 24-105 4.0 IS optically has a light intensity disadvantage compared to the 17-55 2.8 IS.


If weather sealing, build quality, and expanded range the 24-105 offers are also considerations it makes the victory that much sweeter.

[Daniel Browning]

The 1.6x 50d has half the sensor area of a FF 5dII - therefore half as much light still produces the same light intensity at the 1.6x sensor likewise the FF requires twice as much light.


Yes. (It's actually two and a half times smaller area, but half is close enough: (36x24)/(22.3x14.9)=2.6X).



The 24-105 has the same 77mm filter size as the 17-55. This is arguably the same the front element, correct? I did not understand Daniel's comparison.


I was wrong. Thanks for the correction. Yes, they are the same size. The front element has to be at least as large as the aperture, but it can be any size according to the lens design. I should have left that point only to telephoto lenses where it is more often true. I should have stuck to aperture: if you hold both lenses up with the end caps off, and look through them from the front at a bright background, you'll see the same size aperture (hole) through them. With the 17-55 f/2.8 at 55mm f/2.8, for example, the hole will be the same size as the 24-105 set to 88mm f/4.5 (though f/4 is close enough to look the same).



In Bryan's review of the 17-55 he says it is arguably the most hand holdable lens Canon makes, this excludes camera body and its' ISO perforamnce. The combination of wide aperture, IS, and wide angle are my understanding as to why this is true. This powerful combination is unmatched by any other canon lens for static subjects.


Excluding the camera body and ISO performance, yes. If those are taken into account (as you did below), then I'd argue that the 24-105 on full frame eeks out slightly more hand-holdability (as you intimated) thanks to the wider field of view, slightly wider aperture, and the same 3-stop I.S.



Assuming equivalent ISO, equal shutter speeds, equal subjects, and the 5dII at 4.0 properly exposes the image. Will the 50d at 2.8 (remember the assumptions) overexpose, underexpose or equivalently expose the image? I belive it will overexpose it.


Agreed: overexpose.



Now the question becomes - How much better is "that much" better?


I'd say it's in close proportion with the difference in area, except for a small difference allowed for the slightly higher technology of the 50D. In the case of the 24-105, the aperture is 1/3 stop wider than it needs to be for this difference (2.6X area difference means 1+1/3 stop). Generally, I think we can expect it to scale with the area of the sensor, then we factor in differences in technology/performance on top of that expectation.

Great post, Matthew.

[Jon Ruyle]

It seems to me like you understand the situation pretty well.


There is a simple way to think of all of this. It is common to say that, eg, a 100mm lens on a 1.6fovcf camera acts like a 160mm lens on a full frame camera. Ie, you multiply focal length by fovcf. What people more rarely point out is that you should also multiply f/ number by fovcf. For example, an f/2 lens on a 1.6fovcf camera acts like an f/3.2 lens on a full frame camera. Or, to put another way, an f/4 lens on a 5DII acts like an f/2.5 lens on a 50D. Cropping does not affect aperture, but it makes lenses act longer and slower.


This is true for both depth of field and exposure considerations. That is, you'll get (slightly) narrower dof and more background blur with the f/4 lens on the full frame than with the f/2.8 lens on the 50D, and, for a given amount of noisiness, you'll get (slightly) faster exposures with an f/4 lens on a full frame body than a f/2.8 lens on a 50D.


Of course I'm not saying you can shoot iso 400 on both cameras, f/2.8 on 50D and f/4 on 5DII and have a faster shutter speed on the 5DII. Obviously at a given iso, the f/4 lens will have a slower shutter speed no matter how big the sensor is. What I'm saying is that because the 5DII sensor is 2.5x as large, iso 1000 on the 5D II should be about as noisy as iso 400 with the 50D. So you can shoot f/4 at iso 1000 on the 5DII and have a faster shutter speed with about the same noise as f/2.8 at iso 400 on the 50D. (This is all just another way of saying what you said above... I think you have it exactly right).



A 17-55 on a 50d is effectively a 27-88mm from a full frame perspective - DoF not considered.


Right. But you can say more. A 17-55 f/2.8 on a 50d is like a 27-88mm f/4.5 on a 5DII (2.8 times 1.6 is about 4.5), and this includes DOF and exposure considerations.



In Bryan's review of the 17-55 he says it is arguably the most hand holdable lens Canon makes, this excludes camera body and its' ISO perforamnce.


Exactly right. The 24-105 f/4 on the 5D II is slightly more hand holdable than even the 17-55 f/2.8 on the 50D (assuming the IS's are comparable). But since the 17-55 is incapbable of illuminating a full frame sensor, I personally would call the 24-105 a more hand-holdable lens.



In my opinion, the charts easily show the difference between sensor noise is at least 1 stop of improvement


I agree with you here as well.

[waltgary]

A "stop" is a factor of 2. (So a "stop", or even a "stop of light" is actually unitless)


Aha. It's amazing what one word (in this instance - factor) will do for my understanding of a subject. I've been thinking of it from the wrong perspective, so Jon's reply helps. Now, while I don't pretend to comprehend everything, I will continue to study this subject. It's like the compliment I once got from a boss/business owner, "I love hiring people smarter than me. It's so useful". gary

[Aaron K]

This discussion, as well as my own experience moving from a Rebel XT to a 40D to a 5D Mark II, only helps to reinforce my opinion that the maximum (well, minimum...) f-number of a given lens is becoming important largely in the area of composition, instead of actual light-gathering ability, at least for typical walk-around lenses and 'average' situations. Of course, 'average' situations are going to vary wildly by photographer, and there are definitely times when popping on an f/1.4 lens can make a huge difference. But for most walk-arounds (and the 70-200 zooms), the extra stop of light gathering ability from f/4 to f/2.8 just doesn't seem to be a deciding factor for most situations, given the abilities of the Canon's recent cameras. In my case, at least, I gladly accepted the IS, and smaller size of the 24-105 f/4L over the 24-70 f/2.8. Similarly, my 70-200 f/4L IS has been much nicer to carry around than the f/2.8 I used for several days.I guess my point is this: as cameras become better at squeezing the most out of every photon they let through the door, our purchasing decisions can become a bit more focused and a bit less concerned with trade-offs.

[Daniel Browning]

Excellent explanation, Jon. Lots of good points, like this one:



you should also multiply f/ number by fovcf.

[SupraSonic]

If you want to know better put both oo your camera on a tripod set to mannual mode, make sure your meter is enable.Then you know the relative effect of ISO vs Apature.