[WMBP]

I've heard this too, many times, but now understand it to be, at best, misleading. Because reducing the pixel count on a smaller sensor doesn't make it do as well. The high resolution FF cameras of today have smaller pixel sites than the low resolution APS-C cameras of yesterday.

From the Equivalence site on the Web that dosdan in Oz mentioned:

MYTH #8) Larger sensor systems have less noise because they have larger pixels

Larger sensor systems have less apparent noise than smaller sensor systems because they have the ability to collect more total light (by using a larger apparent aperture with a concomitant more shallow DOF, or by using the same apparent aperture and longer shutter speed), not because they have larger pixels. For fully equivalent images where both the DOF and shutter speeds are the same, however, larger sensor systems will collect the same amount of light, and the system with the more efficient sensor will have the least apparent noise.

Now, as I said earlier, one of the things that I've long assumed is that, the more you pay for the camera, the more efficient the sensor. Well, I don't actually think in terms of efficiency very often, but I do assume that the sensor in a $5K camera is super-duper state of the art extra wonderful.

Will

[nostatic]

I'm still wanting the sensor that is asymptotic to zero in size as can use a really small lens and have almost zero noise.

[falconeye]

1. doubling the size of a lens that is pretty big means you end up with a monster.
[...]
2. I keep returning to the lenses that are actually available
[...]
3. bigger photosites and this in turn was supposed to mean less noise. Was I simply mistaken?
[...]
4. Could Olympus produce a 4/3 or micro 4/3 camera whose noise properties rival those of a Nikon D3X - simply by providing larger diameter lenses?

5. Isn't there a reality ceiling here somewhere beyond which it's impossible to go technically?

I nice collection of good questions.

1. Correct. And the cost of long tele lenses roughly increases with physical diameter to the power of 2.5.

2. This is the correct way to look at it. Lens availability. I once checked that for the Pentax system, the DoF/noise advantage of MF vs. FF is little because of a more limited choice of MF lenses.

3. Mistaken, yes. Cf. post #33.
BUT, and this is missing from my post #33, FF at ISO100 performs like APS-C at ISO50 (with a lens big enough) and ISO50 may not be available for APS-C. But D700 starts at ISO200, so equal again

4. Theoretically, yes. Practically, no.
E.g., the Olympus Zuiko ED 35-100/2.0 is almost twice as expensive as the equivalent DA*50-135/2.8 for APS-C which in turn is more expensive than the A 70-210/4.0 for FF ever was.

Which is why I keep saying that the cheapest SLR system was FourThirds in the past, is APS-C now and will be FF in the future. Simply because glass stays expensive while the cost of electronics keeps going down.

5. Yes, there are technical limits for how large a lens can become in practice. For microscopes and mirror-less systems, this limit seems to be about f/0.7. For SLR about f/1.1-f/1.2. But the real point how large a lens can become and still stay sharp.

The best lenses (I know one) have their sharpest f-stop at f/2.8, excellent lenses at f/4 and good lenses at f/5.6. A majority of consumer zooms performs best at f/8. So, even with big money, f/2.8 seems to be the limit for maintaining full resolution.

So, if an f/2.8 lens is already used for FF, then APS-C will most likely not be able to catch up, neither in resolution (at f/2.0) nor in light (at f/2.8).

[Andrew Faires]

In general, Andrew, you've summed up what I've tried to say in, um, my last 37 tiresome posts.

However, there may be a practical point here, at least for some of us. I think all the time about buying a full-frame camera. Unless I win the lottery, I wouldn't abandon my Pentax cameras. Actually, even if I WON the lottery, I wouldn't give up on APS-C. But if I don't, I'm certainly not going to be shooting weddings with a couple of full-frame cameras around my neck. However, what I do think about - and what would be practical for me - is the purchase of one full-frame camera, with perhaps just one good lens in the 50-150 (35mm equivalency) range. I'd use it in church instead of my Pentax 50-135 f/2.8. It's possible to get a used Canon 5D these days at a pretty good price. And Sony now has a very attractive looking full-frame camera for less that $2K.

But this thread may lead me to conclude that full-frame is even less useful to me than I thought.

Will

LOL! It's not that I missed your posts. I kind of skimmed through the last 4 pages

Personally, I think FF offers many practical advantages over aps-c and the cost is now getting to a competitive point. My real reasons for not jumping at the chance are that the advantages of APS-C play more to my style of photography right now and I can't mount my 43mm on any of the FF camera on the market now (and even if I could, I've grown used to the way the 43mm behaves using an aps-c camera).

Like I said, we compose and execute photos based on the gear we have. When I moved from 4x5 to 35mm, I lost some very noticeable advantages but really didn't miss them because I'd get lost in the act of composing with the gear I had at the time.

[Aristophanes]

Perhaps I'm reading it wrong, but most people seem to get those confused, and it was my understanding that's what this thread attempts to correct. People have a very mistaken habit of looking at a picture and thinking, if I had had a full frame camera, I could have taken this same picture and it wouldn't be so underexposed or noisy. However, it just doesn't work that way. If you have a different format camera in your hand, you have to take a slightly different picture. It doesn't just magically take the same settings and let you push the ISO.

The point of this thread, I thought, was to underline the difference between those two view points. The incorrect view point that thinks, "Hey, FF would let me use a higher ISO so it's automatically better for this picture" and the correct view point that thinks, "Hey, it's dark as a mofo in here and even with a high ISO it's slightly underexposed with all the apertures I have available to me, if I had a FF camera, I'd be able to expose this properly".

Everyone go out and get a copy of Bryan Peterson's new book "Understanding Photography: Field Guide". Turn to pages 84-85 where he excoriates the impact of high ISO capabilities in cameras. And he's a poster boy (and stock shooter) for Nikon and Nikon FF.

In this section he rips the impact of high ISO capabilities on composition and creative exposure. If you want snaps at the bar, go right ahead. Knock yourself out with FF and high ISO. That's a recipe for cloned shots with little photographic merit. I have seen (go to Flickr) dozens of high ISO shots in low light, fast glass settings that have awful composition due to DOF limitations imposed by FF high ISO,tech-happy camera geekdom. They look all the same, and always will. (And this is why HD video with its roving eye will have a huge impact on DSLR photography. Within a pan or course of action, the focus and DOF can shift, like our real vision).

For the tech talk here, almost no pro photographer takes high ISO photos. (And very, very few shoot faster than f/2 regularly; one reason why f/2.8 is seen as the pro low large aperture). The standards of composition and exposure to which we aspire are known. As the original post on this thread demonstrated the difference in resolution is negligible between FF and APS-C sensors. It's the glass that counts.

[nostatic]

For the tech talk here, almost no pro photographer takes high ISO photos. (And very, very few shoot faster than f/2 regularly; one reason why f/2.8 is seen as the pro low large aperture). The standards of composition and exposure to which we aspire are known. As the original post on this thread demonstrated the difference in resolution is negligible between FF and APS-C sensors. It's the glass that counts.

Well, not sure if I"m "pro" but I do shoot as part of my day job, and under some conditions I routinely shoot iso1600 and when I go FF will shoot 3200 and 6400. Some of us have no choice but to shoot available light. As for composition, you do the best you can. Try shooting f1.8 in a dimly light container box while a meet-and-greet is going on.

As for the art side of me, the "standards of composition and exposure" are made to be broken and/or ignored

[rawr]

...In this section he rips the impact of high ISO capabilities on composition and creative exposure. If you want snaps at the bar, go right ahead. Knock yourself out with FF and high ISO. That's a recipe for cloned shots with little photographic merit. I have seen (go to Flickr) dozens of high ISO shots in low light, fast glass settings that have awful composition due to DOF limitations imposed by FF high ISO,tech-happy camera geekdom. They look all the same, and always will. [...]

For the tech talk here, almost no pro photographer takes high ISO photos.

I think you are being too casually dismissive of high-ISO. You could say exactly the same stuff of low-ISO or anything in-between. The ISO of a shot isn't the determining factor of bad photography.

And yes, even flickr has some excellent pro and other photographers who practically only do high ISO (TGKW - Flickr: TGKW's Photostream - is one good example) with excellent results.

And as to 'no pro photographers take high ISO photos', I think literally thousands of wedding and sports photographers, not to mention the photojournalists and paparazzi you see on the news every night covering major events with huge zoom lenses in dimly lit environments, would disagree with you.

[Aristophanes]

I think you are being too casually dismissive of high-ISO. You could say exactly the same stuff of low-ISO or anything in-between. The ISO of a shot isn't the determining factor of bad photography.

And yes, even flickr has some excellent pro and other photographers who practically only do high ISO (TGKW - Flickr: TGKW's Photostream - is one good example) with excellent results.

And as to 'no pro photographers take high ISO photos', I think literally thousands of wedding and sports photographers, not to mention the photojournalists and paparazzi you see on the news every night covering major events with huge zoom lenses in dimly lit environments, would disagree with you.

First, low ISO from 100-200 closely matches our natural vision. That's why the scale starts where it does.

The vast majority of sports photos I have seen are lower ISO, max 800, and only then indoors. By far 200-400 ISO is most common.

Same for wedding photos, although it depends on the look. The noiseless clarity required of wedding shots usually trumps everything else. Noise Ninja = wedding guys.

Photojournalism, auteur, and street photography use high ISO, but guess what, go to the photostream you quote above (or any tabloid) and tell me if every shot isn't more or less the same qualitatively. Not content, but image design. They are beautiful shots on that stream, with the vast majority being heavily processed portraits with very similar backgrounds, vignetting, a certain "dirty" look, grainy to stay within the artistic parameters, etc. In other words, a single style. Proves my point. High ISO reliance is stylistically limiting.

That's Peterson's point. Over-reliance on high ISO will take you down a pre-determined creative path. You'll get the same look time after time. And you'll make a point of locking into a single thematic stream. If your great at that and can make bread and butter doing so, excellent. But that's rare and exceptional, and in the end, a one style wonder. Stylistically, that photostream is no different than a thousand macros one sees of the same flowers at the same time of the year. In the end, it's just a flower (face) shot the same way time after time.

The catch is, too much of this discussion is trying to use ISO to push the limits relatively, as if higher ISO won't affect aperture and shutter speed. If those are not taken into account, all the high ISO does is limit your control, not extend it. By design, it cannot be otherwise.

[rawr]

I think having access to clean high ISO simply extends the photographers scope and expands the range of variables - shutter, aperture, DOF etc - they can manipulate to produce their work. It technically expands, rather than limits, the photographers toolbox. It opens up more parameters to work with, not less, I would have thought.

Sure over-reliance on it can give your work a 'signature' style and maybe even end up a stylistic ghetto.

But so too may over reliance on flash, filters, shooping, noir, fisheye, ultrawide, vignetting or any other tool, trick, device (Lomography LOL), or even any location or subject (portariture, street, motorsports).

That's a problem of poverty of imagination that is not unique to the subject at hand.

[pingflood]

I think having access to clean high ISO simply extends the photographers scope and expands the range of variables - shutter, aperture, DOF etc - they can manipulate to produce their work. It technically expands, rather than limits, the photographers toolbox. It opens up more parameters to work with, not less, I would have thought.

I agree with you. It's fairly common on here (and surely elsewhere) that when you discuss advantages of another system, people fall all over themselves trying to argue that whatever their system is lacking is really unnecessary/bad/a crutch/limiting/not real photography blah blah yada yada.

Sour grapes, in other words.

Examples:
High ISO not so good => High ISO is bad because you'll end up always using it and limiting your creativity

AF not so good => Real photographers (and always "pros") just manual focus. Real photographers buy D3s and 1Ds just to disable the one thing the camera does better than other cameras.

FPS not so high => You will be forced to "machine gun" and lose all ability to compose and frame. Also, if you know what you're doing, you can ALWAYS predict the action and make that one single perfect shot.

[kittykat46]

Photography is a multi-faceted experience. Theres no one feature which trumps others.

Just to give another aspect of the High-ISO thingy.
I occasionally use a Canon 50D, which looks better at ISO3200 than the K-7 (not sure if its really that much better - it looks smooth but smeared away a lot of detail).

Especially when shooting with a non-IS lens, I found I NEEDED the 50D's high ISO setting just to get the shutter speed high enough to shoot handheld.

I would not hesitate to use the DA 70mm at 1/15 secs. handheld on a K-7, for example, but on a Canon 50D I would need at least 1/60s on a 70mm focal length, otherwise I'd get a blurred picture. That's easily a 2-stop advantage which the K-7 has with on-board SR.

[WMBP]

I spent several hours last night reading the Joseph James article "Equivalence," which was referenced by dosdan early in this thread.

Equivalence

Got through the whole thing, although I really need to read it two or three (or five or six) more times to absorb it all. Much of it is, frankly, a bit challenging - but it's really interesting and the parts that I feel I truly understand make great sense (finally). It's a terrific article, really terrific, and I recommend it to anybody who found this thread interesting (and who hasn't read it already). At the end of the (very long) article there are lots of references, including references to some threads over at dpreview.com that were similar to this thread, and I found some of that additional reading very helpful.

I want to thank Haakan for starting this thread, Falconeye and everybody else who pushed it forward, and especially Marc Sabatella for hanging in there with me while I was struggling.

*

What's the upshot for me?

Well, first, I still think I was right that full-frame cameras, in practice, have advantages in low light. But I see now that the reason for this is NOT the size of the sensor or even the size of the photosites. As a practical matter, I still wish I could afford a Nikon D3X to complement my Pentax gear. But I certainly have a better understanding of the differences between the Pentax and the Nikon now.

And I will say that this improved (and improving) understanding has made me feel a little more confident about the APS-C platform. It's possible that the future lies with full-frame. But I don't think that's at all certain yet, in fact, if I had to place a bet, I think I'd bet that APS-C will get better and better. The popularity of the Nikon D9, the critical reception of the Pentax K20D and now the K-7, and the excitement already generated by the forthcoming Canon 7D, all suggest this to me. An awful lot (perhaps most) of the pro wedding photographers I know are using top-of-the-line APS-C cameras (Nikon D90 or D300, or Canon 50D) rather than full-frame.

Finally and perhaps most importantly, I just have a better understanding of certain technical issues. No, this better understanding doesn't make much of a difference to my real-life shooting. As I said, I shoot with the camera and lenses that I have and I don't spend any time at all thinking about how much better a job I could do if I had a different camera in my hands. Still, I'm happy to have learned some really interesting stuff here about how my camera works and what it does and does not have in common with other cameras.

Anyway, thanks again for a vigorous discussion, thanks especially for Marc Sabatella for staying with me from post to post, thanks to everybody in the middle of the thread, and again, thanks to Haakan for starting it.


Will

[Mister Guy]

I especially want to thank everyone involved for keeping their frustration in check and actually coming to a meaningful discussion! I think discussions like this help people to realize that it really IS the real world performance that makes one camera better than another.

The practical application Will is looking for is one of money, not of camera quality. It's simply pointing out that a photographer doesn't automatically get an ISO advantage with a full frame camera without the fast lenses to go with it. A full frame camera simply makes fast lenses faster, out of necessity.

The other practical upshot is again reinforcing which case is the 90% group of users, and which is the 10%. The 90% of sunny day, flash, landscape, etc etc users aren't going to instantly see improvement with a full frame camera compared to a similarly featured APS-C. This discussion helps remind people what that 10% case really is: the instances where it really is simply too dark for proper exposure with your current best lens and current APS-C camera. If you can't buy a faster lens, then your only option is to buy a better performing camera. I think Haakan has done a great job of demonstrating exactly WHY and under what conditions the combination of a fast lens and FF camera shows improvement.

However, there is a SIGNIFICANT subjective argument ready to happen about the quality of the noise in the first place. That's the whole noise versus detail argument, and it belongs in a different thread. Some of the Nikon results are certainly cleaner, but I'd still personally reject them as throwing too much away. In my personal opinion, noise (or what USED to be grain) does a better job of conveying the feel of the environment better than a completely sterile one.

I'll link to an example, since it's not completely on topic, but I don't think I'd like this image at all without the tiny bit of noisey grain that emphasizes that yes, it really was that dark in there.

[palmor]

This thread reminds me in some ways of the "crop factor" threads that used to pop up.

Anyway a couple of things jump out at me. First when we are talking about high ISO shooting we really are talking about what the cameras can do at their limits. When you need to shoot your K20 at ISO3200 what does the output look like? If we need to shoot some imaginary FF camera at ISO3200 (or more) what does the output look like.

I don't think that it really matters what the "apples to apples" comparisons are like when you try to keep all the variables in check between the two formats. What matters is how a camera performs at it's limit AND what those limits are.

I'll throw out another real world example... I shoot a lot of dog agility and this will be the 1st indoor season for me. I'm going to pick up a 135/f2 lens for the season because to get close to the shutter speed I want I need to have a very fast lens. On my cropped camera (canon 40d) I will be shooting at its limits.. probably ISO1600 and wide open at f/2 which will get me the bare min shutter speed I need to freeze dogs in mid flight (and a lot of noise that I'll have to PP out). If I had some full frame camera I would still be using the 135/f2 lens but I would probably be able to bump the ISO to 3200 and increase my shutter speed to something above the bare min and still have less noise.... and even if I had the same level of noise I bet I'd have more keepers with the higher shutter speed.

So while maybe the noise is the same between cropped/ff sensors when you artificially put all these variables together there are situations in the real world where it doesn't make any sense.

A great post anyway as I do enjoy the occasional theoretical post

John

[Marc Sabatella]

I know *I* learned a ton by participating in this discussion, so I'm glad there are others who have been sticking it out too. My sense now is that Falk is absolutely right when he observes that the "Equivalence" article is way too wordy to explain what should be a relatively simply concept, and I do hope to decides to write up his shorter version.

Meanwhile, I've been doing some more thinking, and here's my own "executive summary" that cuts through virtually all of the math and gets to the heart of the matter in a relatively straightforward way.

- When comparing identically-framed shots taken in identical lighting with identical exposures (same aperture, same shutter speed, same ISO) the FF camera *will indeed* have less noise. That's what everyone knows already, and why some find it hard to reconcile this discussion with their experience or what is "common knowledge". So relax, no one is challenging the truth of this statement. We're simply asking you to question the value of performing the comparison in this particular way. Instead, we're asking you to consider what else is different about those two camera systems aside from the size of the sensor and how this has affected the results.

- In order to get those identically framed and exposed shots, you needed a longer and physically wider larger diameter lens on the FF camera. That is, if you shot APS-C with a 135/2.8, you need a 200/2.8 to get the same frming and exposure. OK, sorry, you need the tiniest bit of math to get that, but really, we're just talking crop factor here. We all know that you a 200mm lens on FF to get the same FOV as a 135mm lens on APS-C, right? And math or no math, hopefully we can all accept that a 200/2.8 is going to be both longer and have a physical larger diameter than a 135/2.8.

- Not only is the 200/2.8 a physically much larger lens than the 135/2.8, but you'll also get shallower DOF with the 200/2.8 on FF compared to the 135/2.8 on APS-C.

- So the bottom line here is that you got less noise on FF, but that was made possible because you shot with a physically larger lens and accepted shallower DOF.

- The big leap one needs to make is this: you could have *also* got less noise (and shallower DOF) on APS-C by using a physically larger lens on that camera. In particular, with a 135/2, you would have been able to shoot at half the ISO and still get the same shutter speed as in the original comparison. You'd also get shallower DOF, of course.

- You'd need to work through to math to prove whether or not the noise levels and DOF are in fact *exactly* the same as they were on the FF camera in the original comparison. Maybe they are, maybe they aren't. But clearly, using a lens with a physically wider diameter on the APS-C camera allows it to reduce noise (and DOF) - *just as it did for FF*.

Now, assuming we accept all this, the ramifications for long focal lengths such as this are pretty clear: in order to less noise, you really do need a physically larger lens. If you're currently shooting a 135/2.8 on APS-C, then you can reduce noise (and DOF) either by getting an FF camera *and a physically larger 200/2.8 to use with it*, or you can reduce noise (and DOF) by getting a physically larger 135/2 *for the camera you have now*.

So the issue really becomes one of which of the two physically larger lenses is cheaper, more widely available, etc. And I suspect the details differ for pretty much each FOV you might be interested in. Scoring a 135/2 for APS-C and scoring an FF camera and 200/2.8 are *both* pretty expensive propositions. For the "normal" view, though, you'd need a 33/1 to match the 50/1.4 on FF, and we all know that is simply *not* going to happen - FF will provide a clear win. On the other hand, you might end up having a hard time beating an APS-C + 77/1.8 combo by going to FF + 115/2.5.

Zooms are kind of a wild card here. Almost regardless of focal length, the fastest zooms available are f/2.8, and those are pretty much the standard for people considered with high ISO performance in the first place. If you're assuming you're going to use f/2.8 zooms either way, well, then using them FF is the clear winner in term of noise, because you'd actually need an f/2 zoom to match tht for APS-C. Your 70-200/2.8 will be much larger than your 50-135/2.8 (and DOF less), but that will be the price you pay for better noise control. In the normal range, the design of the zoom is such that you can get a 28-75/2.8 for FF that is no larger than the 16-50/2.8 zoom for APS-C, so you do end up with an even more clear win for FF - the only price you pay is DOF.

Kit lenses illustrate this tradeoff too. A basic 28-80/3.5-5.6 lens for FF is indeed the "larger" version of the 18-55/3.5-5.6 needed to get the noise advantage, but again, zoom lens design being what it is, there isn't necessarily much of a size difference in practice. On the other hand, an f/2.4-4 standard zoom for APS-C would match the FF kit. Suggesting that Sigma is really onto something with their 17-70 - that's the equivalent of a 28-105/4-6.7 for FF. Again, though, zoom design being what it is, the 17-70 for APS-C is actually a *larger* lens than the typical 28-105.

Anyhow, when you start looking at the specific lenses that would be involved in "equivalence" comparisons, you realize that *in practice*, FF really does have more or less the advantage we all know it does, at least when using the lenses we are most likely to be using. It's still interesting to consider the fact that we could *in theory* get the same advantage for APS-C with larger lenses. And that longer focal length ranges, getting better noise performance for a given FOV involves much larger lenses no matter which way you go.

What I take from this is that the main advantage of FF comes to those who favor zooms or shorter focal primes, because those are the lenses for which you *can't* easily get an "equivalent" for APS-C. Those favoring telephoto primes may find it more of a wash - you need to get much bigger and more expensive lenses to improve performance either way.

And me? Well, since I've known for a long time that my main concern is concert photography and that my preference is for small short-medium telephoto primes and that and I would *not* want to be having to shoot with a 70-200/2.8 in order to beat the performance I get now, it turns out I am in the particular niche for whom APS-C is actually the winner!