[Marc Sabatella]

Note that, when I say the APS-C sensor can't match the output of the FF camera in certain extreme conditions, I don't mean "can't match without a lot of extra trouble," I mean can't match plain and simple. The Nikon D3X can shoot at, what is it? ISO 25,000 or something ridiculous like that. On my K20D, I don't have ISO 25,000 as an option, so there's no point talking about what I COULD do if the K20D had an ISO 25000 option.

But it's not like FF magically provides a camera with the ability to go to 25000 and APS-C magically prevents a camera from doing that. You're describing a difference between those cameras that has *nothing* to do with sensor size. Might as well attribute the D3X's faster AF, frame rate, more metering points, or better LCD to the FF sensor too.

And while the Pentax K20D can shoot at ISO 6400, I know from personal comparison with shots from Nikon FF users that the FF image at 6400 is generally going to be considerably better than the K20D image at the same ISO.

Right, because if you're shooting at the same ISO on both camera, then you must also be using the same f-stop, and that means you're using a much *larger* lens on the D3X, as explained in my previous posts above. If you used as large a lens on your K20D, you'd get a lower f-number and hence be able to use a lower ISO than you would on the D3X, and the noise would be the same.

I'm simply not interested in numbers, except as a means to results.

Ditto. And result that the numbers point us to is this: if you want low noise, you need a large diameter lens. The only advantage to FF is if it makes using an appropriately large lens more feasible (eg, there exist 70-200/2.8's but there do not exist 50-135/2's).

[Marc Sabatella]

that's silly. Tv priority isn't applicable to every scenario and lowest possible iso = greatest detail = priority overall.

No one said anything about using Tv mode. Doesn't matter what mode you are in. I'm just pointing out that if you're trying to compare cameras, it's the same *shutter speed* you need to compare at.

If one camera has better noise at ISO 6400 than another, but the other somehow lets you get the same shutter speed *without resorting to ISO 6400*, then there is in effect no difference between the cameras in practice, because you won't bother using ISO 6400 on the camera that can give you the same shutter speed at a lower ISO. That is indeed the whole point.

a pro canon photog last night was telling me how he really disliked having his 1ds mk2 for an ac-dc concert 'cause it's iso1600 is horrible compared to his 5d mk2 which gives him a usable 3200.

And I maintain - and the numbers *prove* - that this difference is solely because he uses a larger diameter lens for a given FOV on his 5D.

[attack11]

.. but there is always a difference. every sensor will have different noise starting at approximately iso400 and subsequently a loss of detail as you get into high isos; usually the ones that offer a cleaner high iso also offer a cleaner low iso.

you also have to take sensor dynamic range into account for detail recovery depending on push.

[stanleyk]

And finally, a good photographer with a K20D will outshoot Uncle Larry the camera collector with his D3X every day of the week, in any kind of light.



Will

Thank you!!

[Haakan]

Let me take a practical example. Say 2 photographers are shooting the same wedding in the same church, one with a K20D and the other guy with a Nikon D3X. Let me think of a very typical situation: subject distance is 30 ft, shutter speed can't go lower than 1/30th sec. To get the same depth of field (roughly 8 ft), the Pentax setting at 50mm would be f/2.0, while the Nikon at 75mm will stop down to f/2.8 to get the same depth of field. What's the result?

Will

This is exactly the type of comparison I was thinking of. So I looked at the D90 and D700 DxOMark curves (I wanted to take away other factors, and these cameras seemed to be very similar except for the sensor format). The difference in SNR curves shows about 1 EV benefit for the D700. But since the ISO for the 75mm f/2.8 would have to be 1 EV higher than for the 50mm f/2.0, the end result should be almost identical (again based on the DxO noise measurements)

Please do not think that I am trying to argue against your findings though, I am just following through on the logic of what I found when looking at these two specific cameras (D90 and D700)

Best regards,
Haakan

[WMBP]

Now, if you're telling you me you can shoot at 200mm, f/4, and 1/60" with your FF camera and get significantly better noise than you can shooting 135mm, f/2.8, and 1/60" on your APS-C camera, then I guess that suggests the "quantum efficiency" of your particular FF camera might be better than that of your particular APS-C camera.

But I'm guessing what's really happening is that you only getting better noise performance by shooting at the same *aperture* on both cameras - that is, shooting 200/2.8. Well, sure. Just as you'd get better noise performance on APS-C if you shot a 135/2, because it would let you cut ISO in half. And in both cases, you'd be greatly increasing the size of the lens, and also greatly decreasing DOF. So if you're comparing shots at the same f-stop on both systems, that's not apples-to-apples - you're slanting the results in favor of the FF system by using a much larger lens on that camera.

OK, this makes some sense. Thanks, Marc.

Except, well, for a couple things. First, I don't think you can ever compare the 2 systems EXACTLY because they are inevitably different in many ways. Only way I can see to do this would be to build a pair of test cameras specifically for the purpose of the test. Me, I'm much less interested in the theory than in the results obtainable with real cameras. As you seem to acknowledge in the first paragraph above, the FF might produce a more noise-free image because its processor is more efficient. If I paid $5000 for a body, I'd damned well want the processing engine to be as good as possible. (But see my final comment below about the future of APS-C.)

Second, I am still not entirely sure I understand this desire for this to be a fair fight. I understand your Geo Metro vs Maserati analogy, except that it doesn't quite seem appropriate. I'd say that it seems like what people here want to do is see if a Metro is as fast as a Maserati by running a race - and telling the Maserati driver he can't use more than 3 cylinders. I mean, in your concert example, why WOULDN'T the FF user stop down to f/2.8 if he had it? He will lose a few inches of depth of field, which, at a hypothetical distance of 30 ft, admittedly is going to be pretty shallow already, but I think that's doable. I'd do it. Or he could just reduce the focal length to 190mm and get back all that lost depth of field and then some. The APS-C camera user, on the other hand, is already wide open at f/2.8 and can't go anywhere - can't open the aperture, can't slow the shutter, and doesn't dare - or at least hesitates - to increase the ISO because he knows the noise will be somewhere between pretty bad and really bad. So he takes the shot he has and reminds himself that his camera is lighter than a full-frame, cost thousands of dollars less, and has built-in shake reduction. I'm talking real-world photography. You do what you need to do within the limits of what you can do.

*

Now, I want to clarify something. It might sound as if I've been the advocate of full-frame cameras. Not so, for two reasons. As I've said, every purchase is a compromise. I'm happy paying a LOT less and getting image quality that's just as good most of the time.

But I have also noticed that the makers of APS-C sensors seem to be making them more and more efficient. I have in fact taken pictures with the K20D at ISO 6400 that were actually pretty acceptable, "noise-wise." I wouldn't be surprised to see this trend continue - and for full-frame cameras to end up being even more of a niche market than they are now. At least I HOPE that's what happens. :-)

Will

[WMBP]

Once again - the only advantage of FF here is that it is generally easier to obtain large diameter lenses for FF than for APS-C. In order to get results on APS-C as good as from a 70-200/2.8 on FF, you'd need a 50-135/2.

You know where I can get one with a K-mount? I would like it to be weather-sealed and have HSM, as well.

Marc, at some point we run into reality here, no? How big a lens would I have to get to be able to compete with my Nikon D3-using friends who shoot routinely at ISO 6400? You're right, I don't give a darn about the ISO, really, except that ideally I would like it to be as low as possible. But when the light gets low enough, I'm screwed, because there isn't a lens available.

If we're simply saying that an APS-C *can* under certain circumstances take an image that's as noise-free as a full-frame, then I apologize to everybody for having wasted their time and mine. Me, I'm interested in the circumstances when the APS-C CANNOT take as good an image.

Will

[Marc Sabatella]

.. but there is always a difference. every sensor will have different noise starting at approximately iso400 and subsequently a loss of detail as you get into high isos; usually the ones that offer a cleaner high iso also offer a cleaner low iso.

Not sure of your point here, so another very simple example:

John has a FF camera, shoots a 200/2.8, and gets a shutter speed of 1/60" at ISO 3200.

Mary has an APS-C camera using the same sensor technology, shoots a 135/2 (same diameter lens), and gets a shutter speed of 1/60" at ISO 1600.

The math says, noise (and everything else) will be the *same*. So are you disputing the math, or saying there is something artificial about the comparison? As far as I can tell, it's *exactly* the relevant comparison.

If you feel Mary should be using a 135/2.8 instead of a 135/2 - presumably so she'll be forced to use ISO 3200 as well - then indeed, noise will be worse for her. But my point will remain - the advantage of FF will have been made possible only because you let John use a lens that is *lot* larger than the lens you let Mary use. And John will paid for this in decreased DOF as well. Let Mary have the 135/2 - same diameter and same DOF as John's 200/2.8 - and the difference in noise goes away too.

On the other hand, if you let Mary use the 135/2 but force her to use ISO 3200 for some reason, then she'll again have more noise - but she'll also have twice as fast a shutter speed. John could also trade shutter speed for noise by turning up his ISO to 6400 - again giving him the same shutter speed as Mary, *and the same noise*.

So in my comparison, FF provides no advantage whatsoever, unless it is that John will probably have an easier time getting a 200/2.8 for his FF camera than Mary will have getting getting a 135/2 for her APS-C camera.

you also have to take sensor dynamic range into account for detail recovery depending on push.

Right, but if you read the referenced article, or the entire contents of this thread, you see that resolution and DR works out the same too once you make sure you are comparing apples to apples: same FOV, same shutter speed, same lens diameter (which works out also provide same DOF), etc.

[WMBP]

Ditto. And result that the numbers point us to is this: if you want low noise, you need a large diameter lens.

By larger diameter, I take it you mean physically larger, right? In other words, you could have two lenses, both 28mm primes, both max f/2.8, but one of them could be larger than the other?

The only advantage to FF is if it makes using an appropriately large lens more feasible (eg, there exist 70-200/2.8's but there do not exist 50-135/2's).

Um, that's my point. Except that I'd get rid of "more" in front of "feasible." The FF cameras make feasible something that simply isn't feasible on a Pentax dslr. If it's because we don't have the big lenses, well, okay, I'll yield on that point, so long as somebody on the other side will acknowledge that lack of lenses has consequences to what we can actually do with Pentax DSLRs.

Will

[Class A]

Like speaker wires in high end audio ...

I like my Monster M2.4s bi-wire. Which ones do you use?

[Marc Sabatella]

Marc, at some point we run into reality here, no? How big a lens would I have to get to be able to compete with my Nikon D3-using friends who shoot routinely at ISO 6400?

You need one *the same diameter as theirs*. You're right, the reality is, you probably won't be able to find one in many cases. So the way it works out is this: *IF* you're OK with using huge lenses in order to reduce noise, then FF will probably make it easier to get those lenses. However, *if* you prefer shooting relatively small systems, then hoping that someone will someday make a small FF camera that magically gives you better noise performance is a pipe dream. the only way you'll reap that potential for better performance is by getting a much larger lens.

Personally, I have no interest in shooting concerts with anything physically larger than a 135/2.8. If a FF camera forces me to use a 200/4 just to get *parity*, and a 200/2.8 in order to get a stop advantage, I'm not impressed.

If we're simply saying that an APS-C *can* under certain circumstances take an image that's as noise-free as a full-frame, then I apologize to everybody for having wasted their time and mine.

Actually, I don't consider this time wasted on my part. I came into this thread with only a fuzzy understanding of this. Thanks to Hakaan's thought experiments, Falk's clear posts, my spending a little more time reading the Equivalence essay, and now trying to paraphrase all this stuff myself, I'm coming away with a *MUCH* improved understanding of the situation. In particular, what I just wrote about the idea of a K-7-sized FF camera being a "pipe dream" has just now for the first time hit home. Until now, I've doubted such a thing were likely to ever happen, but assumed that *if* it did, I might somehow be able to benefit from it without buying larger lenses. I can now clearly see that even if someone comes out with a K-7-sized FF body, I'm absolutely going to need bigger lenses to take advantage of it. Meaning I'd be just as well off simply getting myself an A*135/1.8 for my K200D.

I'm interested in the circumstances when the APS-C CANNOT take as good an image.

I'm interested too, but given that the only times this is true is when using a larger diameter lens on FF than exists for APS-C at the same FOV, and I have no intention of using lenses that large, I'm satisfied with the answer here.

The only advantage to FF for me would be the larger viewfinder, and I gather that's doable even on APS-C.

[WMBP]

Not sure of your point here, so another very simple example:

Actually I don't find these examples as simple as I'd like. :-)

John has a FF camera, shoots a 200/2.8, and gets a shutter speed of 1/60" at ISO 3200.

Mary has an APS-C camera using the same sensor technology, shoots a 135/2 (same diameter lens), and gets a shutter speed of 1/60" at ISO 1600.

The math says, noise (and everything else) will be the *same*. So are you disputing the math, or saying there is something artificial about the comparison? As far as I can tell, it's *exactly* the relevant comparison.

If you feel Mary should be using a 135/2.8 instead of a 135/2 - presumably so she'll be forced to use ISO 3200 as well - then indeed, noise will be worse for her. But my point will remain - the advantage of FF will have been made possible only because you let John use a lens that is *lot* larger than the lens you let Mary use. And John will paid for this in decreased DOF as well. Let Mary have the 135/2 - same diameter and same DOF as John's 200/2.8 - and the difference in noise goes away too.

On the other hand, if you let Mary use the 135/2 but force her to use ISO 3200 for some reason, then she'll again have more noise - but she'll also have twice as fast a shutter speed. John could also trade shutter speed for noise by turning up his ISO to 6400 - again giving him the same shutter speed as Mary, *and the same noise*.

So in my comparison, FF provides no advantage whatsoever, unless it is that John will probably have an easier time getting a 200/2.8 for his FF camera than Mary will have getting getting a 135/2 for her APS-C camera.

Marc, I just want to say that I've read the above five times now and I think I see the technical point. I'm still struggling with the practical point, though.

Will

[Marc Sabatella]

By larger diameter, I take it you mean physically larger, right? In other words, you could have two lenses, both 28mm primes, both max f/2.8, but one of them could be larger than the other?

If they are the same diameter, then they have the same maximum f-stop. After all, f-stop is simply focal length / diameter. OK,. it might be possible through optical trickery of one sort or another to get around this (that's how constant aperture zooms, I gather), but in effect, what I mean is one is a 28/2 and the other is a 28/2.8, and the 28/2 *will* have a larger diameter than the 28/2.8.

The FF cameras make feasible something that simply isn't feasible on a Pentax dslr. If it's because we don't have the big lenses, well, okay, I'll yield on that point, so long as somebody on the other side will acknowledge that lack of lenses has consequences to what we can actually do with Pentax DSLRs.

And I'll yield on that point if those on the other side acknowledge that needing a bigger lens in order to accomplish something means it can't do what I can currently do (which is to say, shoot a relatively small camera system). I've said recently in another thread, without fully appreciating the truth of what I was saying, that you cannot "have it all" - you can *either* have the small camera system (body+lens) in your hand, *or* the lower noise camera system in your hand.

[WMBP]

Actually, I don't consider this time wasted on my part. I came into this thread with only a fuzzy understanding of this. Thanks to Hakaan's thought experiments, Falk's clear posts, my spending a little more time reading the Equivalence essay, and now trying to paraphrase all this stuff myself, I'm coming away with a *MUCH* improved understanding of the situation.

Well, I'm hanging in here (much to everybody's chagrin, no doubt) because I'm TRYING to understand what's being said. And I'm doing that because I too feel that I would like to be able to grasp all this better. And if the end result is that I feel even better about having my Pentax cameras and less envious of my full-frame friends, well, that's gravy.

But I'm still struggling with a conflict between the theory and math, on the one hand, and the reality I know on the other. I know that full-frame bodies are bigger, and that they're bigger in part because they have somewhat bigger sensors in them. I'm actually quite content with the size of the K20 + grip and while I would probably go to FF if I had $20K in spare change lying around, I don't, so I won't, and even if I did, I would not expect my photos to be much or any better most of the time.

But the lenses? This Nikon 50 f/1.4 doesn't look any bigger than my Pentax 50 f/1.4:

Nikon D3 Lenses

Is your point that this Nikon 70-200 f/2.8 is technically comparable to the Pentax 50-135 f/2.8, but that the Nikon WEIGHS 5 lbs while the Pentax weighs only 3 lbs? For what it's worth, the Nikon weighs only a little more than 3 lbs if you take off the tripod collar. The lenses are comparable in width - the Nikon's 3.4 inches, the Pentax 3 inches. The Nikon is of course LONGER (8.5 inches vs 5 inches) but them's the breaks with longer focal lengths, I guess.

Anyway, so here we have a Nikon lens, specs similar to the Pentax lens, both f/2.8. I'm pretty sure I'd get better results from the Nikon lens mounted on a D3 than I get with the Pentax lens.

The problem for me is that the D3 costs $5K and this Nikon 70-200 VR lens costs almost $2K.

It seems to me that the difference in the SIZE of the lenses is less of an issue if you tend to shoot things that are closer....

Will

[Marc Sabatella]

Second, I am still not entirely sure I understand this desire for this to be a fair fight. I understand your Geo Metro vs Maserati analogy, except that it doesn't quite seem appropriate. I'd say that it seems like what people here want to do is see if a Metro is as fast as a Maserati by running a race - and telling the Maserati driver he can't use more than 3 cylinders.

True, the analogy doesn't work in this direction very well. Because if it were to be true, it would be a simple matter to outfit the Metro in a manner similarly to the Maserati and thereby match it in performance, when in reality, it isn't.

So here's an analogy that works in that direction: let's run a Ford Taurus SRE - a hypothetical "Special Racing Edition" that has been fitted with a special high performance engine - against an unmodified Ford Focus.

The Taurus SRE wins, but not because it's a Taurus - it wins because of the fancy engine it's been outfitted with. We could, in theory, fit the same engine into the Focus, and it should do as well.

So buying a Taurus only provides an advantage over the Escort if you outfit the Taurus with a fancy engine but don't do so for the Escort.

I mean, in your concert example, why WOULDN'T the FF user stop down to f/2.8 if he had it?

??? Which specific lens are you referring to? I was never considering a case where an FF user has something faster than f/2.8 available to even have the option of "stopping down" to f/2.8. If you're envisioning a new example, in which the FF user has an f/2 lens, then indeed, the APS-C user needs an f/1.4 lens at the corresponding shorter focal length to get the same diameter lens and the same performance in terms of DOF, noise, and everything else. But since I never mentioned any f/2 lenses for FF, I think you might still be confused about something - not sure what?

The APS-C camera user, on the other hand, is already wide open at f/2.8 and can't go anywhere

If you're talking about an APS-C user with an f/2.8 lens, then the equivalent lens for FF (same diameter, hence same DOF and same IQ potential) is f/4. You can't stop an f/4 lens down to f2.8 - it's already too small for f/2.8. You need an f/2.8 or better lens to do f/2.8, and at the focal length required to get the same FOV on FF, that's going to be much larger than the f/2.8 lens the APS-C user is using. So again, the way you gave the FF user the advantage was by giving him a larger lens, not by giving him the FF camera. Give the APS-C user a lens the same diameter - which would be f/2 at his correspondingly shorter focal length for that FOV - and he'll get the same results. Same noise, same DOF, etc.