Is Full Frame sharper than APS-C? My answer here

[TrumpetPower!]

More megapixels is always  better = better picture quality,better tonal transitions and less estimation errors and jaggies, easier to correct CA etc .  Here, a discussion, read John Sheehy, Eric Fossum , the father of the active cmos , this subject has been discussed many times , its is funny when people like to tell Eric Fossum how a sensor works.  http://forums.dpreview.com/forums/post/37445608

Is it therefore your assertion that the 18 megapickle Sony WX150 produces better images than the 11 megapickle Canon 1Ds?

b&

[TrumpetPower!]

no, go back to my answer again, I added sensor area because I know that this question was coming, more pixels ia always better and a larger sensor area is always better, it means more pixels

Mikael, you really should use the preview function. Changing your posts, especially after people have replied to them, is rude. It falsely makes other people look like they're arguing against a straw man when all they're doing is responding to something you've already realized is in error -- and it also makes it look like you never made the error in the first place, which is a lie.

If you really must change something after the fact, leave the original, strike it out (EDIT: to clarify, use strikethrough), and add something like, ``EDIT: Sorry, that's a mistrake. I really should have written...''

b&

[TrumpetPower!]

I know that this question was coming, more pixels ia always better and a larger sensor area is always better, it means even more pixels.

As I have already demonstrated, even when both formats have the exact same number of megapickles, the larger format still wins.

The smaller format needs not merely an equal number of pixels, not only an equal pixel density (and therefore more megapickles), but something even more in addition to all that in order to produce the same image quality as a larger format. That could be a more efficient sensor, it could be better optics, it could be even more megapickles, it could be any number of things.

But what sets larger formats apart from smaller ones isn't the number of megapickles but rather the simple geometry of a larger imaging area capturing more information and resulting in a smaller enlargement from capture (sensor / film) size to display size.

Cheers,

b&

[neuroanatomist]

Gee...I don't know...how about getting the actual cameras and putting the same lens on each; changing position to achieve the same FoV; and making sure that all other factors (scene; lighting; aperture; shutter; settings) are equal, thereby isolating the difference you want to measure (sharpness)? 

On second thought...nah...that's just crazy talk 

Yes, that's crazy. Spectacular fail crazy.

Use the same lens on different formats and you get a different field of view. Change position and you get a different perspective in addition to a different field of view. Keep the aperture the same between different formats and you get different depth of field.

I was very specific in what I did and didn't change, and I did it for a reason. My test really was a true apples-to-apples comparison. The shooting position was the same, so the perspective was the same. The shutter speed was the same so the motion blur (not that there was any) was the same. The aperture was different, yes, but it resulted in the same depth of field. The ISO setting was different, yes, but it resulted in the same exposure -- and, unsurprisingly if you know the basics of photography, it resulted in basically the same amount of noise.

I thought at some point in this long, acrimonious thread, AlanF (but maybe someone else) commented that TrumpetPower had failed to isolate the relevant variable.  That's true - way to many differences in this 'test'.

"Spectacular fail crazy."  No, dtaylor is spot on.  I seem to recall that several pages back, TrumpetPower stated that the thing to do would be to use two sensors of the same MP count but different sizes, that's at least barking up the right tree, however, the 'test' didn't do that.

Evaluating sensor sharpness using two different lenses at different apertures? Scaling the resulting images? That's uber-spectacular quintuple-facepalm-type fail. 

TrumpetPower asked what the ideal test would be...first off, not his.    The solution is the one dtaylor suggested - two cameras, one FF, the other APS-C, with the same MP count.  Use the same lens, and change the distance to match the FoV. As for the objections of problems with changing distance resulting in different perspective and the same apertures resulting in different DoF on different formats, there's an obvious solution: a flat target, parallel to the sensor. That means no perspective to be altered (perspective alters the apparent distance relationship between subjects at different distances, but with a flat target the entire subject is at a fixed distance), and the flat target renders DoF moot.

So...two equal MP sensors of the same generation, a flat target framed identically, same lens for both (preferably a sharp one), same aperture.  Where, oh where, could one find such a test? How about here.

What's the conclusion?  Pretty much this:

...all other factors being equal...and I've always found the same thing: FF is sharper out of camera, but the range is small enough to be closed by bumping up sharpening in camera or post. When the sensors are otherwise comparable of course.

Is this a perfect test?  No - there's no such thing with the tools available. The AA filters are different (the 18 MP 7D fares much worse in the above TDP comparison than the T4i, for example, likely due to a stronger AA filter), the firmware is different, etc.  It's a test at one distance per camera, at some distances with some targets the Nyquist sampling issues raised by AlanF are a determining factor. It's a test at ISO 100 - at ISO 3200, the results would be quite different. 

Bottom line - the differences of opinion in this thread far exceed the real differences in sharpness between sensor formats, and the wagging tongues in this thread are sharper than my Wüsthof cutlery.      

Over and out, for me.

[TrumpetPower!]

The solution is the one dtaylor suggested - two cameras, one FF, the other APS-C, with the same MP count.  Use the same lens, and change the distance to match the FoV. As for the objections of problems with changing distance resulting in different perspective and the same apertures resulting in different DoF on different formats, there's an obvious solution: a flat target, parallel to the sensor. That means no perspective to be altered (perspective alters the apparent distance relationship between subjects at different distances, but with a flat target the entire subject is at a fixed distance), and the flat target renders DoF moot.

You might have a point...if we were comparing photocopiers.

But you, of all people, should know that the results of, say, three portraits, one made with APS-C, one with 135, and one with 645, all with a 50mm lens at the same aperture and "zoomed" with the feet, will be so radically different that any photographer who suggested they were somehow interchangeable would be fired on the spot for gross incompetence and laughed off the set. Try that as a sports photographer and you'll be ejected from the venue by security. Try it as a landscape photographer and you'll fall off the cliff to your death.

I cannot imagine a more pointless, unrealistic test. If your goal is to demonstrate a passion for creating the ultimate rig for imaging newspapers taped to brick walls, congratulations, you've succeeded. But if your goal was to demonstrate an understanding of actual photography as practiced by actual photographers....

Again, the test I performed resulted in two images with the same number of pixels in addition to the same perspective and depth of field and background magnification and out-of-focus blur and everything else a photographer cares about. The only variable was format size, and we saw that sharpness is very closely related to format size and not at all related to the number of pixels in the image. All your test demonstrates is that, if you ignore everything that actually defines what an image of anything other than a test chart looks like, you can get two systems with certain similar specifications to make similar images of a test chart and nothing else.

b&

[neuroanatomist]

I cannot imagine a more pointless, unrealistic test. If your goal is to demonstrate a passion for creating the ultimate rig for imaging newspapers taped to brick walls, congratulations, you've succeeded. But if your goal was to demonstrate an understanding of actual photography as practiced by actual photographers....

Again, the test I performed resulted in two images with the same number of pixels in addition to the same perspective and depth of field and background magnification and out-of-focus blur and everything else a photographer cares about. The only variable was format size, and we saw that sharpness is very closely related to format size and not at all related to the number of pixels in the image. All your test demonstrates is that, if you ignore everything that actually defines what an image of anything other than a test chart looks like, you can get two systems with certain similar specifications to make similar images of a test chart and nothing else.

Therein lies your problem. Your test attempted to compare 'the same picture'.  You might have a point...if that were the goal. But we're not asking about 'everything that defines an image', we're asking about only one thing.  Let's revisit the title of this thread: Is FF sharper than APS-C?  Not 'does FF take a better picture'.  Yes, photographers care about 'better pictures' but that's not what's being asked here.  Only sharpness.  So in this case, taking pictures of a newspaper taped to a brick wall would be a far more appropriate test of the question than yours.

"The only variable was format size..."  Really?  Do you understand what a variable is?  Do you know the difference between a dependent variable and an independent variable, and how to tell when a dependent variable becomes a confounding variable?  You used different lenses to take the different pictures. You processed the images differently.  We call those confounding dependent variables.  Your test was not designed to answer the question, and therefore was flawed for the purpose at hand. 

"...the test I performed resulted in two images with the same number of pixels in addition to the same perspective and depth of field and background magnification and out-of-focus blur..." Guess what?  So did the TDP comparison I linked. Same number of pixels (but achieved without scaling), same effective perspective (indeterminate because the image has no depth), same effective depth of field (essentially none, as the entire subject is within it), same background magnification (none, no background), same OOF blur (none, everything in focus).  Unlike your test, the TDP ISO 12233 crop holds everything constant.  It's no accident they're called that - ISO 12233 is the "standard for measuring the resolution of electronic still imaging," and that's what we're talking about here...not 'better picture', sharpness.

An analogy: It's as if I asked, "What is the sodium content of french fries vs. carrot sticks?" Your test is trying to answer a question like 'which one tastes better?''  Yes, that's an important question for a person who wants a snack, but that's not what's being asked.  I'm looking for a chemical analysis of Na⁺ content, and you are conducting a blind taste test.  Interesting and informative, perhaps, but a complete fail as far as answering my question.

If you want to test the effect of sensor format on sharpness, you must hold everything else constant - especially the lens and the post processing. You varied both.  It's not that your test was invalid per se, but rather that it was not designed to address the question being asked.  Sharpness is just one component of picture quality, but again - that's the one we want to test.

[TrumpetPower!]

Let's revisit the title of this thread: Is FF sharper than APS-C?  Not 'does FF take a better picture'.  Yes, photographers care about 'better pictures' but that's not what's being asked here.  Only sharpness.  So in this case, taking pictures of a newspaper taped to a brick wall would be a far more appropriate test of the question than yours.

Again, such would be a valid complaint amongst photocopier repair techs.

But, ask any working photographer if it makes any sense to compare sharpness of two photographs of a real-world scene with dramatically different compositions, perspectives, depths of field, amounts of background blur, and all the rest, and they'll think you're either completely nuts or one of those crazy Internet measurebators who only takes pictures of newspapers taped to brick walls -- as if there's a difference.

In the context of a photography discussion, the question most reasonably should be considered to be, "Does 135 make sharper photographs than APS-C?" Not, "Does 135 make sharper photocopies than APS-C?"

Your test was not designed to answer the question, and therefore was flawed for the purpose at hand.

Physician, heal thyself.

I'll even concede the point that 135 and APS-C make comparable photocopy platforms.

Now, which system produces sharper photographs?

Do you know the difference between a dependent variable and an independent variable, and how to tell when a dependent variable becomes a confounding variable?  You used different lenses to take the different pictures.

Yes, I do know what a confounding variable is. And the fact that you led off with the different lenses demonstrates that you don't.

The hypothesis is that image sharpness is independent of sensor format, and that two formats with the same total pixels will produce images of the same sharpness. The experimental results falsified that hypothesis, even -- nay, especially though the choice of lenses is known to have biased the results in the opposite direction.

If your hypothesis is that John and Sue can both run as fast, and you then make Sue but not John carry a 25 pound sack of potatoes around the course, and if Sue still manages to easily beat John, then you know that Sue is definitely the faster runner. You don't know how much faster a runner she really is, but you definitely know that she's a faster runner.

Had John won, you wouldn't know if he won because he's a faster runner or if it's because Sue was saddled with all those potatoes. Had they tied, you'd know that Sue was the faster runner, and that the difference in performance is equal to a sack of potatoes.

But neither scenario is what happened. What we found is that, even when we weigh down Sue with a sack of potatoes, she still crosses the finish line well before John.

And now you understand how confounding variables work, and why my choice of lenses makes the point more emphatic.

...unless, of course, it is your position that the 135 SF is actually sharper than the TS-E 24 II, but I rather doubt you'd be capable of such cluelessness.

Now, go through all of my other variables, and you'll find the same applies: they're tilting the scales, yes, but in favor of the smaller format. In addition to the sack of potatoes in the form of a "meh" lens compared with a legendary one, the larger format also got weighed down with even more sacks of potatoes in the form of aperture (f/4 v f/32), ISO (100 v 6,400), and post-processing (SOC v unsharpened resampling).

Again, if the small format had turned out sharper or comparable, we would have a hard time making a conclusion. But that's not what happened. Even after we loaded down the larger format with all those sacks of potatoes, it still crossed the finish line first.

So, my test can't tell you how much sharper the larger format is, but it most emphatically does tell you that the larger format actually is sharper.

(Even if they're equally lousy photocopiers.)

Cheers,

b&

[neuroanatomist]

...my test can't tell you how much sharper the larger format is, but it most emphatically does tell you that the larger format actually is sharper.

So do the ISO 12233 crop comparisons. That just goes to show that even a poorly designed experiment which includes both intentional biases (different lenses shot at different apertures) and unequal data manipulation (image scaling), can yield a correct result...

[TrumpetPower!]

...my test can't tell you how much sharper the larger format is, but it most emphatically does tell you that the larger format actually is sharper.

So do the ISO 12233 crop comparisons. That just goes to show that even a poorly designed experiment which includes both intentional biases (different lenses shot at different apertures) and unequal data manipulation (image scaling), can yield a correct result...

Once again, you're comparing photocopiers, not cameras.

The comparison between a Sony WX150 and a 1Ds would most emphatically not be with a lens of the same focal length on each. The Sony lens has a focal length of about 4.5mm - 45mm. I've never even heard of a 4.5mm lens for EF mount.

No, you'd set the Sony at the ends and middle of its range, and either use a superzoom on the Canon or grab a couple / few lenses with comparable fields of view.

Now, do you agree that that's how you'd go about comparing the images produced by those two cameras? Or would you shoot the Canon (somehow) at 4.5mm with the lens pressed up against the test chartt?

In other words, are you a photographer, or are you a photocopyist?

b&

[neuroanatomist]

The comparison between a Sony WX150 and a 1Ds would most emphatically not be with a lens of the same focal length on each. The Sony lens has a focal length of about 4.5mm - 45mm. I've never even heard of a 4.5mm lens for EF mount.

I will repeat the topic at hand, one more time, please read carefully: Is Full Frame sharper than APS-C?  Does the Sony camera that you keep bringing up have an APS-C sensor in it?  If it does, we could discuss how to compare that camera's sensor to a Canon full frame sensor.  If it doesn't, such a test would be tangential, irrelevant, and yet another test that fails to answer the question being asked. Exactly like your test…

To be honest, this thread is starting to smell a lot like another recent thread, with the part of 'intransigent, insult-delivering antagonist' played by a different actor, only the insults are more frequent and rude and there's apparently not even the excuse of a language barrier in your case.  Accordingly, given that my normal limit is one such 'debate' per week, I'm done.   If it would help your ego to view this as a forfeit on my part, and you want to infer that means your inappropriate 'test' was somehow a properly relevant one, you go right ahead. 

干杯 (or gānbēi, if you can't read Mandarin - pinyin is easier to Google  )

[AlanF]

sorry for the brain fart!

Apologies accepted!

[TrumpetPower!]

The comparison between a Sony WX150 and a 1Ds would most emphatically not be with a lens of the same focal length on each. The Sony lens has a focal length of about 4.5mm - 45mm. I've never even heard of a 4.5mm lens for EF mount.

I will repeat the topic at hand, one more time, please read carefully: Is Full Frame sharper than APS-C?  Does the Sony camera that you keep bringing up have an APS-C sensor in it?  If it does, we could discuss how to compare that camera's sensor to a Canon full frame sensor.  If it doesn't, such a test would be tangential, irrelevant, and yet another test that fails to answer the question being asked. Exactly like your test…

Obviously, the Sony has an even smaller sensor still.

You really should be aware of this but obviously aren't because you're belaboring the point, but, when comparing a specific case with marginal differences, it is often very helpful to compare the general case with greater differences.

I would hope that you'd agree that an hypothetical 20 megapickle 1/2.3" camera would be nowhere near as sharp as an equally-hypothetical 20 megapickle 8x10 digital view camera.

I would also hope that you'd agree that, when comparing those two cameras, you would not compare them both with the same lens -- that is, with a 4mm lens on the small format camera as well as a 4mm lens on the view camera. Rather, you'd use a 4mm lens on the small camera and a 180mm lens on the large camera. Otherwise, to maintain the same magnification of the subject, the subject would probably have to be a negative distance from the front of the large format camera's sensor.

Can we agree on that? Yes? Please?

Let's continue the thought experiment.

Rather than compare 1/2.3" and 8x10, let's compare 4/3 and 645. Here, you might use a 24mm lens on 4/3 and a 100mm lens on 645. 24mm is a normal lens on 4/3, but it's a 180° fisheye on 645.

Still with me?

Just to recap: when comparing 1/2.3" and 8x10, it's probably physically impossible to use the same focal length lens at the same subject magnification on both. You can physically do such a comparison between 4/3 and 645, but the results are nowhere near similar, even at a first glance by a non-photographer.

If the subject were, "Is 8x10 sharper than 1/2.3"?" or if it were, "Is 4/3 sharper than 645?" would you consider your protocol valid? Hell, would it even occur to you to use such a protocol? I would really hope not.

So, now, we finally move to a comparison of APS-C and 135.

Can you offer a rational explanation as to why an experimental protocol that's physically impossible in one setting and laughably absurd in another would suddenly become valid in a third setting merely because the variables are close enough that you can fudge them to kinda sorta fit if you pick just the right type of never-happens test?

Again, my test simulates what actual photographers actually do when actually making photographs. You pick your subject, figure out the ideal perspective and composition, pick a lens with a field of view that suitably frames it, pick an aperture that creates a certain depth of field, pick a shutter speed that captures motion suitably, leaving you with hoping that the resultant ISO is within your camera's acceptable range, else you start making compromises to the other variables. Once you've got all that set, if you were going to compare your results with a different camera, you'd want the same composition, the same framing, the same depth of field, and the same shutter speed if at all possible. If you can achieve all that and the ISO is in the camera's range, you've got your apples-to-apples comparison. Swap the two cameras and the only difference is in the image quality inherent in the two systems. (If not, the necessary compromises are a mark against the camera that can't keep up. We haven't even mentioned DoF; my experiment was at the almost-not-overlapping extremes of available DoF between the two formats -- you could get much shallower DoF out of the large format and much deeper DoF out of the smaller format.)

TL/DR: again, your protocol is valid for photocopiers. Mine is valid for photographers.

Cheers,

b&

[bdunbar79]

You're still not answering the question.  I read your posts and it's like, which is sharper, the 5D Mark III with the 35 f/2 lens, or the 7D with the 16-35L II lens set at 22mm.  Who cares?  That doesn't answer which sensor is sharper. 

[TrumpetPower!]

You're still not answering the question.  I read your posts and it's like, which is sharper, the 5D Mark III with the 35 f/2 lens, or the 7D with the 16-35L II lens set at 22mm.  Who cares?  That doesn't answer which sensor is sharper.

Are we comparing silicon chips set up to poorly mimic photocopiers, or are we comparing camera systems used to make photographs?

See, i was under the impression that this is a gathering place for photographers, and that photographers would know that one matches lens focal length to format. Thus, the understanding would be that a comparison between APS-C and 135 wouldn't be of one isolated component in the camera, but rather of the whole system -- a true comparison of 135 format with APS-C format. And, as such, you'd compare an APS-C camera with a normal lens to a 135 camera with a normal lens. And, as all photographers should know, a normal lens is defined as one with a focal length roughly the same as the diameter of the imaging area -- 43mm for 135, and 27mm for APS-C. (And 7mm for 1/2.5", 300mm for 8x10, and so on.) What you wouldn't do is compare APS-C with a short telephoto lens (50mm) to 135 with a normal lens (also 50mm). Even if you could physically mount the same lens to both cameras. You certainly wouldn't compare 1/2.5" with a long telephoto lens (50mm) against APS-C with a short telephoto lens (50mm) against 135 with a normal lens (50mm) against a physically impossible 270° fisheye lens (50mm) on 8x10.

But, no. Apparently this is a gathering place for some very confused photocopy techs who think that "photographic sharpness" has something to do with how well you can adapt one isolated component of a camera to serve as a woefully inadequate photocopy machine. My mistrake....

b&

[RLPhoto]

14 pages later, how hard is it to believe that FF is sharper than APS-C?