Advantages of full frame?

[Jon Ruyle]

Does this mean that if DOF is a goal, and keeping a lens close to its sweetspot is a goal, than a APS-C sensor actually has some advantage here?

I'm sure Daniel will have a more authoratitative answer, but my guess is the reason for the percieved "sweet spot" is that diffraction has not yet set in in. If this is true, then the answer would be no.

[Dallasphotog]

I'll just add a simple note. I loved my EF24-70 F/2.8L USM on the 1DMKII, but I LOVE it on the 5DMKII. The extra bit of wide angle and the additional creative possibilities are pretty dramatic. If you came from film, it's going to remind you why you loved your wide angle lenses.

[Daniel Browning]

It's also true if something other than pixel density is the limiting factor in image quality.

You're right; thanks for the correction.

I was confused by some of the talk when the 50D came out, around seeing diffraction when stopping down. This was not due to the sensor size it was due to the decreased pixel size and I forgot about that.

Yeah, pixel size is a whole new (and big) topic. Under ideal circumstances (no diffraction, perfect focus, no motion blur, etc.), the 50D will have 21% higher resolution than the 40D. Under no circumstances will the 50D ever have *less* resolution; however, in the very worst circumstances (f/45 diffraction, focus way off, severe motion blur, etc.) the increase in resolution will be vanishingly small (0.1%). Every photograph will be between 0.1% better and 21% better. For most people, I think it will be much closer to 21%.

It is in fact good news that an APS-C camera can get the same DOF as a FF without stopping down as much.

I don't subscribe to that idea, personally. The way I see it, there is no advantage to either; they are just equal.

In fact one of the things I have noticed with my lenses is that there is a noticeable increase in sharpness closer to the sweet spot of the lens. For landscapes I am actually better off trading in some of that DOF for increased sharpness by opening things up a little, say to f8, or f11, rather than f22 or higher.

Yes, that is because the effect of diffraction is reduced.

I have not completely thought through this idea yet so I will pose it as a question. Does this mean that if DOF is a goal, and keeping a lens close to its sweetspot is a goal, than a APS-C sensor actually has some advantage here?

I think it's actually a disadvantage. Generally speaking, I think it could be said that the sweet spot varies by format size. One example might be:

• 4/3rds: f/4
• APS-C: f/5
• FF35: f/8
• 6x9: f/18

So if an FF35 lens has aberrations at f/5, but becomes diffraction limited at f/8, that's great for FF35 cameras. But when you put that lens on an APS-C camera, you get aberrations at f/5 or too much diffraction at f/8.

This is a natural consequence of the fact that we have so many decades of lens development for FF35, but only a few lenses built specifically for APS-C only. In cases where there is a lens built specifically for APS-C, it usually tromps the equivalent EF lens (e.g. EF-S 17-55 vs. EF 17-40; EF-S 60mm f/2.8 macro vs EF 50mm f/2.5 macro).

The lenses we do have for APS-C (EF-S) must have aberrations corrected at a higher spatial frequency and a wider f-number than equivalent EF lenses in order to have the same image quality.

One way to think of it is like this: the crop factor must apply to the *MTF chart* as well as focal length and f-number; otherwise the image quality will not be as good as FF35.

Thanks for an interesting thread!

You're very welcome.

I would think...

Thanks for the response, Johan.

For example, assuming we use the 5D1 and 50D as examples; the 50D would have to have the aperture stopped down much farther than the 5D1 would for the same depth-of-field*,

[* At least at the pixel level; maybe it would be different at equal print/output sizes?]

I kindly think you are mistaken, pixel level or not. To get the same angle of view, the focal length must be different between the 50D and 5D1. With a longer focal length and the same f-number, the 5D1 would have a wider iris diameter, which changes the geometry between subject and focal plane to change, which changes depth of field. However, if the f-number is 1.6X narrower, then the iris diameter returns to the same size as the 50D, and the geometric relationship between the subject, iris, and focal plane are again the same, so depth of field is the same. I think you'll find that DOF calculators and other resources will reinforce what I'm saying.

You do bring up an important factor, though, which is the effect of resolution. The 5D1 and 50D only differ in resolution by 2.2 megapixels, which is not enough to have a big effect on depth of field; however, there are times when total resolution can, indeed, affect depth of field.

If the 5D2 at f/9 is compared with the 40D at 5.6, for example, both will have the same depth of field at normal print sizes. However, at very large print sizes, where the 5D2's immense resolution can be utilized, it's possible that the DOF will be thinner. This would require the resolution to be preserved from the lens (aberrations, motion blur, etc.) through processing and display (e.g. 30x20 print size). Of course, if the thinner DOF is not desirable, it's always possible to just blur the 5D2 image so that it only has 10 MP of resolution, and then the DOF will be the same as the 40D again.

Basically, DOF will be affected by anything that affects total resolution; not in absolute terms, but relative to the viewer's acceptibly sharp CoC.

While I'm on a tangent, I might as well discuss reproduction magnification. To make a 12x18 print, the full frame sensor is enlarged by a factor of 12.7x. To make the same size print, an APS-C sensor must be enlarged by 20.3x. It is magnified much more because it is smaller. Notice that the difference between the two is the same as the crop factor (1.6X). It is that reproduction magnification that causes f-number to scale with sensor size for depth of field, diffraction, and a variety of other things.

Daniel Browning said: Instead, it should have the same iris diameter
(focal length divided by f-number) in order to have the same angle of
view, perspective, and depth of field.


I understand all those terms, but not that statement; would you mind explaining what you mean a little more simply?

It's my pleasure.

For other readers, let me prefix the explanation by saying that iris diameter means the same thing as pupil opening, physical aperture, absolute opening, "hole in space", and other similar synonyms. It does not mean f-number (which is "relative" aperture). The iris diameter is the focal length divided by f-number. For example, 50mm f/1.2 has an iris diameter of 41.7mm (50/1.2=41.7).

Let me try phrasing it in two additional ways:

• For a given field of view, focus distance, and bellows factor, the lens with the widest iris diameter will project the thinnest DOF, no matter what the focal length, f-number, or sensor size.
• Similarly, if any two lenses have the same iris diameter, the DOF will be the same, no matter what the focal length, f-number, or sensor size. Given the same field of view, focus distance, and bellows factor.

The iris diameter of 50mm f/2.8 is 18mm. The iris diameter of 80mm f/4.5 is 18mm. When 50mm f/2.8 on APS-C is compared with 80mm f/4.5 on FF35, one finds that they have the exact same perspective, angle of view, depth of field, and diffraction.

In short: Apply the crop factor to f-number and everything evens out.

This article is why I purchased a used 5D even though I have a 40D. http://www.kenrockwell.com/tech/full...-advantage.htm

He addresses quite a few assumptions.

Ken's site has some good content, but it's mixed in with a lot of misinformation, myths, and misconceptions. I did a quick review of that article; please oblige a brief response to a few of his points:

Bigger pixels collect more light (photons). More photons means more clean image compared to the relatively constant amount of electrons making noise. (Yes, I'm simplifying, but it's why bigger sensors are quieter and cleaner.)

Fatter pixels catch more light, so the larger camera doesn't have to amplify them so much for the same ISO sensitivity. This means that even though we wind up with the same exposure, noise and crud aren't amplified as much, leaving us with a much cleaner images without having to cheat with wimpy noise reduction filtration. There is simply less noise.

This is a myth that I busted on April 25th:

Myth busted: smaller pixels have more noise, less dynamic range

...all cameras smooth over the noise in flat areas to varying amounts where it's most visible...

That's only true for JPEG images, not raw.

the larger format camera sees the subtle difference between the white window frames and the light tan house

This is due to the difference in the spectral response of the color filter array and the JPEG processing, not anything caused by the format size difference.

Aside from those three myths, the information appears to be correct.

Thanks for the discussion, everyone.

[Keith B]

I'll just add a simple note. I loved my EF24-70 F/2.8L USM on the 1DMKII, but I LOVE it on the 5DMKII. The extra bit of wide angle and the additional creative possibilities are pretty dramatic. If you came from film, it's going to remind you why you loved your wide angle lenses.

I own a 24 1.4 L prime that is a great lens. It was on my 40D 90% of the time. It was essentially a 38mm loved it but it wasn't quite wide enough all the time. So I figured it was time to go full frame (enter 5D mk2) and started shooting 24 and it seemed either too wide or not wide enough. So I finally ponied up and bought the 16-35 2.8L II and man have I found my spot(s). Between this lens and the 5Dmk2 my photography has become what I had always wanted it to.


The 16-35 is the best, most pratical (for me) and fun lens I have ever owned. The wide with a well thought out Deutsch Angle and images become so dramatic.


I LOVE this lens and FF.

[Chuck Lee]

This article is why I purchased a used 5D even though I have a 40D. http://www.kenrockwell.com/tech/full...-advantage.htm

He addresses quite a few assumptions.

Ken's site has some good content, but it's mixed in with a lot of misinformation, myths, and misconceptions. I did a quick review of that article; please oblige a brief response to a few of his points:

Bigger pixels collect more light (photons). More photons means more clean image compared to the relatively constant amount of electrons making noise. (Yes, I'm simplifying, but it's why bigger sensors are quieter and cleaner.)

Fatter pixels catch more light, so the larger camera doesn't have to amplify them so much for the same ISO sensitivity. This means that even though we wind up with the same exposure, noise and crud aren't amplified as much, leaving us with a much cleaner images without having to cheat with wimpy noise reduction filtration. There is simply less noise.

This is a myth that I busted on April 25th:

Myth busted: smaller pixels have more noise, less dynamic range

...all cameras smooth over the noise in flat areas to varying amounts where it's most visible...

That's only true for JPEG images, not raw.

the larger format camera sees the subtle difference between the white window frames and the light tan house

This is due to the difference in the spectral response of the color filter array and the JPEG processing, not anything caused by the format size difference.

Aside from those three myths, the information appears to be correct.

More myth info here: http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm


Wikipedia, under "pixel density" sensor section: "Contrary to the popular myth, a high pixel density does not increase noise, decreasing the sensor size does." I think this agrees with your position Daniel.


I'm not sure where all this is going but it is quite a conversation.


All I know is that I am very impressed with the high ISO output of the 5D.In my "subjective observation"the40D generates much more noise at the same ISO. Both are very nice cameras that serve me well. It is also good to note that ISOs are not equivalent between the Canon D series of cameras. The 20D, 30Dand 5D have by default more sensitive ISO bias than the newer 40D and 50D and 5DMII. I think dpreview illustrated this. There's mention of it in several reviews, I know it's in the 40D review. 5D ISO 800 acts more like 1000 yet still seems to produce less/equal noiseas new5DMII at 800.


Now, to me, the most important part of the sensor noise equation has to do with the artificially controlled 1/3 ISO steps on the D series cameras. Obviously this has been shown many times that the ISO sensitivities at 1/3 steps between major ISO sensitivities of 50,100,200,400...etc are generated on a post proccessed level by adding or subtracting numerical values from the RAW output. This is why on a 40D ISO 250 is just as noisy as ISO 400. ISO 320 is better than ISO 250 and about as good as ISO 200.


One of the site's that I find way too deep for anyone (except you Daniel)is http://www.pages.drexel.edu/~par24/rawhistogram/CanonRawScaling/CanonRawScaling.html


Canon 5D signal/noise vs. ISO: http://www.pages.drexel.edu/~par24/rawhistogram/5DTest/5DTest.htmlSo, I use 50,100,200,400,1600,3200. on this body


Canon 40D signal/noise vs. ISO: http://www.pages.drexel.edu/~par24/rawhistogram/40DTest/40DTest.htmlSo, I use 160,320,640,800,1250,1600,3200 on this body.


Works very well. I also have a tendency at 640-3200 to use +.3ev just to be able to comp down in post. That would mean my 3200 is actually more like 2400. I very rarely comp down and under expose. That would only help to amplify the output noise level.


For me Daniel it's more of a layman/blue collar approach to all of the technical stuff guys like you, Rockwell, Drexel, etc. talk about. All I need to know is what are the practical settingsand techniques to getthe cleanest images possible. So, keep on splaining........


Chuck

[cian3307]

Wow. I have been enlightened, some excellent posts above, very informative. thanks everyone. I have been thinking about upgrading to FF at some point but am now thinking the 50D would suit my needs better. I work mainly at the 2 extremes - macro with my 100 f2.8 USM and telephoto with my 300L f4 with 1.4TC. I always try to work in the 100-400 ISO range so noise isn't really a decider.

[David Selby]

http://learnslr.com/EOSForum/index.php?topic=183.0

[adam]

Wow, lots more info here...thanks! It's kinda what I was afraid of...full-frame would help half of my photographic interests and hurt the other half...


I'm wondering if, when my time comes, instead of springing for a 5DII, I should keep my 40D and pick up a used 5D...kind of a "second-best of both worlds" strategy. And it would cost less money, so I'd be able to replace my crop glass (a Tokina 12-24 and a Canon 60 macro) with better full-frame versions...

[JJphoto]

a good artical about full fram and crop body from popular photography. link http://www.popphoto.com/Features/The-Full-Frame-Decision

[Daniel Browning]

a good artical about full fram and crop body from popular photography. link http://www.popphoto.com/Features/The...Frame-Decision

Thanks for the link. I remember reading this when I got it in the mail. Please don't be offended by my criticism of it, but I think the Pop Photo article has very poor treatment of the subject; even Ken Rockwell did a better job.

With more loosely packed pixels, manufacturers can make the pixels bigger, and that’s the case here: A D700 pixel is about 8.5 microns across, while a D300 pixel is about 5.5 microns. Bigger pixels can catch more light, which can make for images that have less noise and greater dynamic range.

I debunked this myth here:

Myth busted: smaller pixels have more noise, less dynamic range, etc.

...for most serious amateurs, and many pros, there is no compelling reason for going to a full-frame DSLR.

There are indeed many reasons, already outlined in this thread, but not mentioned in the article.