Point of diminishing returns with diffraction on 5D and 50D

So I'm working on figuring out which of these to buy, a used 5D or a new 50D, and one of the reviews I was reading indicated that images taken with the 50D were OK at f8 but by f11 were distinctly softer, because of diffraction.

Can anyone provide info, or lead me to info, on where the 5D stands on a similar diffraction scale?

One of the reasons I'm considering the 5D is because it's full frame, and that has some very nice landscape photography possibilities the 50D doesn't offer (unless I also buy the 10-22, which is not in my budget.)

But if the 5D doesn't score well on the diffraction meter compared with the 50D, that will affect my decision somewhat.

This is a topic that got me concerned too. I saw it mentioned in a couple places (maybe fredmiranda's and also in the dpreview review I think). However, some respondents argued that diffraction is lens dependent and independent of the sensor and they still don't appear to agree on it. Seems to me the only way to find out is hopefully some agreeable fellow on this board with access to both a 5D and 50D will do a comparison with an L class lens at various f-stops and report back with some results.

With respect to the landscape lens I am fortunately at a point where I need to buy either the 10-22 or 17-40 anyway so the body is my big ??? right now. The 50D offers so many technological advances over the old 5D but most reviews are comparing it to the 40D so its hard to tell--Larry

Thanks, Larry. While it's my understanding that different lenses will soften at different points on the aperture scale due to diffraction, it's also my understanding that it is at least to some degree sensor-dependent. One review says the more pixels you pack on, the worse the diffraction becomes, which is one reason I'm interested in comparing the full-frame 12 Mp 5D and the 1.6 crop sensor 15 Mp 50D. If that's true, my guess is diffraction is less a problem with the 5D. If I can't stop down my 17-40 on either body past f8 without softening, it's just not much use to me when I want do do landscape work.

EDIT: Ah, I found it. On The Digital Picture review of the 50D, it says the point of softening for the 50D is f7.6, where the 5D is f13.2. Which makes the 5D the best and the 50D the worst in that category among Canon cameras, according to the review. The review of the 50D can be found here: http://www.the-digital-picture.com/Reviews/Canon-EOS-50D-Digital-SLR-Camera-Review.aspx

Now, review info aside, in the real world how much will this matter at f16 to f22?

Anyone?

I too have thought about going with a used 5D, or perhaps even a new one. But at nearly double the price of the 50D for a new-in-box 5D, I think the more modern-technology 50D is my best bet. Don't forget, I'm coming from a 10D. Generally, since I hold on to camera gear for a long time, I prefer to buy new.

No doubt the 5D/5D Mkll is the best bet for certain kinds of photography, such as landscape work where stopping way down is the norm. The 50D would be better at wildlife photography with its faster frame rate and 1.6 sensor, where we usually shoot wide open anyway.

For those of us who dabble in all kinds of photography and cannot afford both, it becomes a tougher decision.

"Now, review info aside, in the real world how much will this matter at f16 to f22? "

I'm betting less than lens quality and/or shooting technique. I think that for the vast majority of "stationary subject" photography, f/8 to f/16 is the sweet spot anyway - all factors considered. YMMV

Yeah, I hear you, Jim.

Primarily, I'm looking to get better high iso peformance over my 30D bodies for indoor wedding work with fast lenses, and that could seriously make it a tossup between the used 5D and new 50D. And I can end up with a new 50D using a current battery grip for a few hundred less than buying a used 5D/grip, which makes it seem like the 50D would get the nod.

But throwing in the full frame capabilities of the 5D, particularly for landscapes, changes the equation somewhat. If I stay with a 1.6 crop sensor lineup, I'll have to spring for the Canon 10-22 or the Tokina/Tamron/Sigma equivalent.

I may stay with 1.6 crop cameras, dump the 17-40 and pick up a new Tokina ATX 12-24, which seems to get the best reviews among the non-Canon ultrawides. That'll still allow me to do all my wide angle wedding posed work but will open things up to a full-frame focal length equivalent.

It's supposed to be simpler than this. Budgets suck.

Steve, I too have the 17-40/4L, which of course ends up being the equivalent of a 28-70 on a 1.6 crop camera. The 17-40 isn't too shabby to start with and then using just the center of the glass makes it even better. I have no plans on getting rid of the 17-40 but may look for something wider to supplement it. Good thing about the 17-40 is if you do end up with a 5D, it will be a good lens for that camera.

Yes, budgets are no fun!

Steve - Diffraction is only caused by the lens aperture, and is not caused or changed by anything else including the sensor. Diffraction can only be seen at the sensor, though, and how much it can be seen is based on the pixel size.

Frequently when someone mentions that "diffraction is independent of the sensor" all they mean is that the lens aperture will produce the same amount of diffraction no matter what the sensor is. But whether or not it's apparent in your images is sensor dependent. The CiC Tutorial on diffraction is pretty good as a reference.

Now here's the thing....diffraction effects set on sooner with smaller pixels, so you'll start to lose resolution due to diffraction sooner on a 50D than a 30D. However, you've got a lot more pixels to begin with on the 50D. So in terms of resolving power, the 50D probably wins even when diffraction is figured in (although it's hard to calculate and is somewhat image dependent). People tend to get really freaked out by diffraction effects, but that's at 100% on their screen. What they fail to consider is that 100% on a 50D image is WAY WAY more blown up than a 30D image at 100%. The same print size would be a better comparison for most people, but people love 100% screen views. (Similar arguments could be made for noise, in that the 50D is worse at a pixel level than, say, the 40D, but that's at 100%. On more equal footing, they are, apparently - I don't own either - much more equivalent.)

Most people shouldn't worry much about diffraction, except to test their setup at very small apertures which apertures aren't acceptable to them. (Point of reference, I have no problem shooting at f/16 on my Oly E-520, with its small pixels. But I'm not usually trying to print 40x30 or anything.)

That being said, the 5D is 12.7 MP and the 50D is 15.1 MP. That's not a huge difference in resolution so I'd make your choice based more on the noise characteristics for a given print size, and your particular uses for the camera. From a purely lens/IQ point of view, I'd probably get a 5D. If you consider all the bells and whistles more important, I'd get a 50D.

Thanks, David. I hoped you'd show up. I didn't know all the tech info on diffraction.

I'm also not a pixel peeper. Can't find a photographic application in my real life for it, and my real-world comparison also is based on prints, not computer screens.

Here's a question for you. Let's say I want to knock down the size of the jpeg I'm shooting in the 50D. And I'm assuming it'll still use all the pixels on the sensor for capture, because how wouldn't it? So, will going from large to medium jpeg, for example, in any way trim digital noise? For weddings, for example, I could easily use a medium jpeg from a 50D (in low-light high-iso situations) to get what I need if that would decrease noise even further.

Jim, because of budget constraints, I'd really have to sell the 17-40 in exchange for an ultrawide if I stay with 1.6 crop sensors. I have a very sharp copy of the 17-40, and it would be nice to hold onto it in case I can afford full frame down the line, but not likely.

Steve - Down-sampling to a smaller resolution file can definitely help clean-up noise due to some pixel averaging to "smooth out" values (one of the reasons web JPGs can look so good). I don't know how it's actually implemented in the camera, but I'd guess it would be similar to what you could do yourself : Take a noisy raw file and save it as a large JPG. The downsample and create a smaller JPG. Now clean-up both however you find appropriate, and see if the smaller one is acceptable/better from a noise vs. resolution point of view.

Thanks again, David. I'd thought so but wanted to hear it from someone who knows a lot more about electrons and software than I do.

David, boy that interactive calculator there is very helpful in trying to determine how much effect diffraction will have for a given sensor size, number of pixels, print size, and viewing distance. Plus their discussion of real world effects puts this all into perscpective. Still don't know which way I'll go, but I plan to make up a table of results for all these interactions--Larry

I think David gave the perfect answer to the question. Before deciding to upgrade from the 400D to the 50D, I was a bit worried about the noise level of that incredibly dense sensor. I was quite happy with the image quality of the 400D at 100 Iso, where this camera is virtually visible noise free thanks to the reduced density. Fortunately I had the opportunity to get the camera in my hands before buy it, and decided to make some print just to compare the two IQs. I did a very simple test, I shot the same picture with the same lens with the two cameras. Then I reduced the size of the 50D file to match the size of the 400D, and finally I printed exactly the same 100% crop of the images on a A4 inkjet printer. Just after the test, I had clear in my mind that the 50D was an amazing photo tool, and decided to buy it right away.
The point is really very simple: if you investigate at 100% an image made with a 50D and one made with a 400D (or 40D, 30D, etc.) you will probably notice more noise and/or soften edges by diffraction, if that's the case. But that's an unfair comparison, as it is just as you were checking two prints, one with a 10x loupe, and the other with a 15x loupe.
Put the two pictures on a par, and you'll clearly see who wins.


Paolo

It's been 2 months since I upgraded from a 20d to a 50d...Love it. Budget wasn't a consideration for me and I only use L glass. Since 95% of my work is done from tripods and rarely above an ISO of 200, I really haven't seen any value in the higher ISO's, but as one who routinely enlarges my images to 20x30 to hang on walls, I've got to say that for me, the 15mp was what became important. I will probably buy the new 5dmkII in the very new future and will use it for some portait work and underwater work, but not for my macro or other nature work, I love the crop factor.

I've just gone through the equivalent Nikon decision. Since I am mostly a night shooter this time of year, high ISO performance is very important. I took a close look at the Nikon D700, and it's performance is excellent. I could certainly afford to buy one, but there was a snag. My current lenses are geared for DX. To really justify buying a D700 I'd need to have the FX lenses in place, and that would be an additional $3,000. I could sell my FX lenses to offset that, but I'd still be putting out many thousands of dollars for D700 + lens. I would gain roughly one stop of usuable ISO over a D300. I already have state of art DX lenses: Tokina 11-16mm f2.8, Nikon 17-55mm f2.8, Nikon 70-200mm f2.8 VR, TC-17E, plus Nikon 80-400mm VR and Nikon 28mm f3.5 PC. I decided to buy the D300 since I already had the lenses to fully support it. I know the D700 will be dropping in price, and will look at one again next summer. I'm in no hurry as I've learned that patience is well rewarded when it comes to DSLR bodies. The lenses I will have when I get a D700 will be: Nikon 14-24mm f2.8, Nikon 24-70mm f2.8, Nikon 70-200mm f2.8, TC-17E, and Nikon 24mm PCE. That plus my Gitzo tripod and 14 flash units/radio triggers will be a very capable camera system. Really, it's looking like the D300 is meeting my current needs.

One thing I'll add to the discussion of DX ultrawides is the performance of the Tokina 11-16mm f2.8. It is excellent. It is better than the perfermance of the Tokina 12-24 f4 or even the Nikon 12-24mm f4. For starters, there is less distortion. It's my favorite lens. The f2.8 combined with usuable ISO 3200 now gives me the capability to photo at night using ambient light in a way I couldn't before. I suppose my point is that my own strategy has been to focus on the entire SYSTEM, rather than dumping a ton of money on just a camera body. That would be pointless if I didn't have the state of art lenses to support it.


Kent in SD

Kent, good point about the lenses - something many of us don't fully consider. Thanks as well for your input on the Tokina 11-16/2.8 - there's a canon version too!

While there is no Canon equiv. of the Tokina 11-16mm f2.8, the lens does come in Canon mount. It's such a good lens that it's tough to find it in stock. It's worth the look though. It's the only non-Nikon I'm using at the moment. One thing that Canon users have that Nikon doesn't are some of the pro quality f4 lenses, such as 70-200mm f4 IS. They are a stop slower, but the image quality is excellent. The price is lower than the f2.8 lenses too. For me, I still need the f2.8 capability though.


Kent in SD

Someone mentioned the small difference in resolution of the two cameras and while the actual pixel count isn't much different, the raw resolving power of the two sensors is. The 50D, by virtue of its smaller pixels will have greater overall resolving power than the 5D; that is, the ability of the sensor to resolve fine detail will be better. The question becomes does the better resolving power get outweighed by the increase in noise inherent in packing more pixels into a smaller space and things like diffraction showing up sooner. According to DPReview, test results aren't that much different between the two. High ISO noise performance seems to be not as good as the 40D (which isn't as good as the 5D) so it seems like Canon may have actually found the limit for pixel density with current technology.

Put simply,
5D resolution drops some at f11, more at f16
1DsIII/5DII resolution drops some above f8 and gets worse from there.

-Diffraction is better than OOF.
-50D is a denser sensor but as a crop sensor it uses a shorter lens for the same shot compared to full frame.
- most hyperfocal calculators use old COC numbers that are too large for dense digital and large prints. Add a stop or two. But yes, this increases diffraction.
- many photographers seem to be focusing too distant and stopping down wide lenses too much. With an ultrawide the correct focal point for a shot that goes to infinity may be less than ten feet.

As far as the camera choice, I think it's as simple as 5D for landscape and 50D for critters. The cost in glass is a lot more for critters and full frame. Certainly the 50D can shoot superb landscapes. It has a much better LCD than the 5D.

I think your choice of camera body needs to be based on things other than diffraction effects.

If you shoot with alot of short - normal primes, the 5D would be a better choice because it will align the intended FOV of Canon's prime lens selection. There are no wide/fast primes for a crop body.

If you shoot wide angle with zooms, there are now appropriate/equivalent zooms for both the crop and full frame bodies. I have both the 10-22 and 17-40L and the lenses are almost identical in image quality, tone, contrast, etc...

If you shoot long lenses and always need more reach, then the higher pixel density of the 50D will be a better fit.

If you shoot daylight sports, then the AF and faster frame rate of the 50D will be a better fit.

If you shoot indoor/night sports the better high ISO performance of the 5D will be better.

Generally wildlife is better suited to the crop body and landscape is better suited to the full frame body, but these are generalizations.

That's exactly right, Mel. It's a much tougher choice for people who will only own one camera body and are photographic generalists.