5D MkII & 50D, crop camera tele advantage now not true?

[scott_ferris]

Ok,

 

If I was to buy a 5D MkII and a 50D, if I then took a picture of a bird with a 500mm lens from a hide, is it true

that there is no difference in image quality, depth of field etc etc between the 50D's picture and a cropped

version of the 5D MkII if taken with the same lens. If that is so, and I believe it is because although the 50D

gives you more pixels they are not actually adding to the image quality they are just resolving diffracted blur,

then we can't really refer to a multiplying factor anymore.

 

I am not attempting to rehash old terminology arguments I am trying to understand that the "tele advantage" of

using a crop sensor camera is now not true. I do use a crop camera but am looking to upgrade next year to FF, I

don't see that I will lose out by cropping my images and this is a comparatively new phenomenon due to the

increase in pixel numbers.

 

Take care, Scott.

[alan_bryant1]

The 1.6 crop sensor cameras have a sensor that is only about 40% the size of the full frame 5D Mk II sensor. So,

to replicate the 50D view by cropping a 5D Mk II image, you would drop the resolution from 21 mp down to 8.2 mp

-- the same pixel count the 20D and 30D have. That's a very usable resolution, but significantly less than the

50D has.

[scott_ferris]

Yes Alan it is very usable, but a lot of the 50D's pixels are wasted, they are just resolving diffracted blur,

obviously in time somebody will post direct comparisons but until then I suspect we will see very little

difference. Percentage wise the 50D pictures can't be enlarged to the same degree that the 5D MkII's can before

diffraction blurs both images equally.

 

I didn't want to speculate too much that is why I used current examples but I will almost certainly be buying a

1DS MkIV rather than a 1D MkIV, as I see it I still have the cropable "tele advantage" that I get from using a

300 on a 1.3 crop camera and I get to use my 16-35 as it should be.

[alan_bryant1]

Yes - good point - only a high quality lens, used carefully, is going to put the 50D resolution to good use. I would guess the 50D overshoots the resolving power of pretty much any inexpensive zoom.

 

I'm sure the camera resolution will continue to improve, but whether there's any point to it remains to be seen. If a full-frame sensor was built with the 50D pixel size, it would have 38 mp. But would that have much benefit? For most situations, I doubt it.

 

I know I want a 5D Mk II, but it's for the magnificent low-light performance, not the resolution.

[elliot1]

You are pretty much correct.

 

Although the file sizes are obviously note the same ad the MKII/50D, I did a test of the D3 Vs D300 and the

pictures are pretty much identical (cropped D3 at 5.1mp vs full frame of the D300 at 12+mp).

 

http://www.photo.net/photo/7773046

 

(D3 is on the bottom).

 

While more is better, it takes a lot more to get a a lot better.

[mark u]

I think we are getting close. Play with the calculator here:

 

http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

 

Answers will depend on your visual acuity and print size and viewing distance as well as shooting aperture, and can be made to swing either way. They are also affected by the design of the anti-alias filter and the success of the Bayer demosaicing algorithm and other image processing in recovering detail blurred by the AA filter. Note the comment about "are smaller pixels worse" at the bottom of the link. I'd also add that the Raleigh criterion for diffraction is proportional to wavelength of light (i.e. red is more diffuse than blue by ~7:4), and is not an absolute. Astronomers often use the tighter Dawes criterion - but their telescopes are probably rather more accurately manufactured than even a Canon L lens.

[scott_ferris]

Hi Mark :-)

 

It was partly Cambridge in Colour that got me thinking about all this when I stumbled on their website a while ago, I have recommended it to people in several threads.

 

Like I say I don't want to speculate on future models but were the next 1D models to continue the pixel increases then the difference between the cropped 1.3 and a cropped FF have disappeared, So apart from FPS there is no advantage to owning a 1D (MkIV?) to a 1DS (MkIV?), indeed the disadvantage of not having an available Canon wide angle that fits the camera is a major disadvantage.

 

I was surprised by this and thought it was well worth commenting on to a wider audience.

 

Take care all and please keep commenting. Scott.

[scott_ferris]

Thanks Elliot,

 

That is just the kind of example I was expecting to see but hadn't looked hard enough. And that is between a FF and 1.5 crop, pretty interesting I think.

 

Take care, Scott.

[scott_ferris]

Alan,

 

It isn't the lens that is limiting resolution now, though obviously the best lenses need to be used to maximise the images, it is the diffraction, even the best lenses can't work around the diffraction, it is worse at smaller apertures as the Cambridge in Colour site points out, and as it also points out, resolution wise we are pretty much there at normal f stops, f5.6-22, the next generation will take that down to sub 2.8, which is what all my lenses are apart from the 50 f1.4.

 

So it seems that this and the next generation of models really are there, more features will be added in the future but the IQ can't really improve much in the 24x36 sensor size and the IQ limits have been surpassed in the APS-C size.

 

Take care, Scott.

[alan_bryant1]

If diffraction were the entire story, lenses would be sharpest wide open. But anyone who has tested their lenses is aware that it ain't so.

 

Diffraction is an insurmountable theoretical barrier that has the advantage of being easy to calculate, and fixed for pretty much all lens designs. But there are other barriers preventing perfect sharpness on the wide-open end of the f-stop spectrum, which vary widely by lens design and are difficult to calculate, but nonetheless are equally insurmountable in the end. The balance between the two is why most lenses achieve their best sharpness near the middle of their f-stop range.

 

So, in practice, I'd have to say it *is* usually the lens that limits resolution now, at least on the latest camera models. Unless you have a good lens used at optimal aperture, the lens is going to be the weakest link, and not necessarily because of diffraction.

[mendel_leisk]

Just cribbing info from The-Digital-Picture site:

 

Canon EOS 50D pixel size: 4.7µm

 

Canon EOS 5D Mark II pixel size: 6.4µm

 

Canon EOS 5D pixel size: 8.2µm (just for reference)

 

Cropping a 5D II picture down by the 1.6 crop factor would only yield the horizontal and vertical pixels of a 50D *if* it's

pixel size was 4.7µm. It's a long way from that.

 

The 50D image size is: 4752 x 3168 (pixels)

 

A 5D Mark II picture cropped down by 1.6 would be: 5616/1.6 x 3744/1.6 = 3510 x 2340

[alan_bryant1]

....Yes, and 3,510 x 2,340 = 8,213,400 = 8.2 mp. Like the first post said ;-)

[scott_ferris]

Hi Mendal,

 

That still equates to 8.2 MP which is what Alan said first off, that is still a very workable MP count,

particularly if you are a Nikon user, or like me a 1D MkI user with 4.2MP total. The other point is pixel size is

not the whole story re dimensions, there are gaps and micro lenses etc, but the numbers all come out around the

same, basically a 1.6 crop yields 40% of a FF, a 1.3 crop yields 77% of a FF.

 

Alan,

 

I do take your point about lens limitations, too often people who don't use the equipment resort to quoting

physics and figures, in my case I do own good lenses but don't use good technique so I know I will be the

limiting factor, it just interested me that now the sensors are very much not the weakest link. I have one

beautiful photo of a surfer, as the wave breaks behind him the sun lights up the spray, the drops of water

registered as one pixel each and so there is a beautifully coloured mosaic effect, very fine but obviously made

up of squares not drops, I like the picture, but it is totally unsaleable and did show up my aging cameras

limitations. I am looking forward to that not being a factor in my next camera.

 

The theory isn't totally without merit though, lenses don't have to be at their sharpest opened up, and, as you

point out we all know that. It is just that diffraction as an influencing factor becomes less apparent as you

open up. There is a sweet spot, as we have all known for years, the spot where the diffraction is not obvious but

the lens is stopped down to take away the resolution limits wide open.

 

Take care all, Scott.

[zafar1]

I ran 40D and 5D head-to-head last year on the same question of crop factor. The results are here.

http://www.flickr.com/photos/zafar1/sets/72157603827705006/

 

I did not find any real advantage of 40D's crop factor vs. 5D (accept for that the object appears closer in viewfinder).

[scott_ferris]

Thanks Zafar,

 

If I can get the frame rate I am used to in full frame I am sold. All I ever wanted in a digital camera was to replace my 1VHS, looks like I am getting remarkably close, I have gotten so used to my crop factor though that I was wondering if there was still an advantage to be had, it seems there isn't.

 

Take care, Scott.

[eric merrill]

I ran tests with a 5D + 300/4 + 1.4x vs 40D + 300/4 taking pictures of subjects about 30 yards away.

 

In my tests, the 40D was better. In fact, the XTi + 300/4 beat the 5D + 300/4 + 1.4x, as well. Used a

tripod. Mirror lockup. Remote release. Autofocus center point.

 

I'm guessing the difference is the 1.4x teleconverter. Straight up, the 5D images are slightly better than

the 40D ones and a lot better than those from the XTi.

 

Since Zafar's results were the opposite for him, I think this illustrates the importance of making sure

exactly what you are testing is what you want to test. That's why I tested shooting objects at a

distance. That's what I cared about in my test. Closeups with macros may tilt the scale differently.

 

 

Eric

[eric merrill]

p.s. And I also repeated my tests handheld using IS. The results were the same. Since I'd rarely use

mirror lockup taking pictures of birds, I also wanted a more real world approach, as well.

[scott_ferris]

Hi Eric,

 

Yes I agree you need to know what you are asking! But did you try taking the same picture without the 1.4 and just cropping the 5D file? I believe that would give you a result that shows little difference and even less if you used the newer cameras, as you point out in your test the 1.4 degrades the image more than the increased object size gives you.

 

That then follows on to am I better using a 1.4 extender or cropping a picture taken without it. It can go either way at the moment depending on MP lens used and the quality of the extender, all quite informative I think.

 

Take care, Scott.

[ronhartman]

Eric, Interesting test. I have a 300mm f/4 on order to use on a 5d and have been wondering what way to go if I want more telephoto.

 

Can you speculate how cropping the 5d/300mm/no converter would compare to the full image of the 40d/300mm?

[jamie_robertson2]

Just to pipe in, there is one factor to consider that is often missed when debating cropping FF images:

 

In a real world situation, especially when photographing distant subjects such as birds and animals, using the cameras AF produces effectively sharp images when used correctly. However, the next time you have a long zoom lens attached (100-400 is a good example), try focusing on something at the 100mm setting... then, without refocusing, zoom the lens to 400mm and see if the subject is tack sharp in focus. More than often it isn't. That's because the camera's AF only had the view through a 100mm focal length to base its calculations on. When you zoom right in at 400mm you can see just how far out those calculations really were. So, when you take a photo that's perfectly in focus on a FF camera they always look great. But... if it's a distant subject photographed at a large aperture (i.e. shallow depth of field) then chances are it will not always be sharp if the image is cropped to the same dimensions as a 1.6x crop cameras angle of view.

 

For closer static subjects it isn't a problem... just zoom the lens to its maximum (400mm in this example) then take the shot at 100mm. That way, it's perfectly in focus. For wildlife and moving subjects you don't have that luxury.

 

That's why I'll miss the 1.6x crop factor for wildlife when my 5D2 arrives.

[mendel_leisk]

Alan and Scott, aren't we comparing the 5DII and the *50D*?

[scott_ferris]

Hi Jamie,

 

I think I disagree with you on that one, the focusing point changes due to lens design limitations not because the camera can't do it. If you do it the other way round and focus at the 400 then zoom out you normally need to refocus but it is not as noticeable because the object is smaller. Lens design is so precise that lens temperature makes a measurable amount of difference to where the focus is accurate, that is why lots of good lenses can focus past infinity, zoom position has always had an effect on focus position too. Some lenses are much more consistent through their zoom range but few do actually keep perfect focus throughout.

 

Take care, Scott.

[scott_ferris]

Hi Mendal,

 

That was my original question yes, but the truth is I am more interested in the next generation of 1 series cameras, my reasoning is that it won't be worse than the current generation. If the difference between a 5D MkII and a 1.6 crop 50D is not there anymore then that should hold up even better between the 1DS MkIV and the 1.3 crop 1D MkIV.

 

If that is so then I will get a FF and crop my tele pictures so I can get my wide angle. One real problem for the 1.3 crop is that the EF-S lenses don't fit so I would need to go generic to get a true ultra wide on the 1D MkIV and that would mean another lens when I already own a very good Canon ultra wide zoom.

 

Anyway thanks for the input, Scott.

[eric merrill]

Scott:

 

The 1.4 wins out over up sizing the file, at least on the 300/4 IS.

 

So to summarize my results:

 

1. 40D and XTi

2. 5D + 1.4x

3. 5D + upsize

 

 

Eric

[denny_wells]

I was going to chime into this conversation with the cambridge color diffraction limit link (posted by MarkU). Note

that the link includes sample images from a 20D displaying the diffraction limits at various apertures. Pretty obvious

differences to me. My take from their description - we are fast approaching a limit where pixel sizes will be reduced

to a point where small aperture, large DOF photography becomes problematic due the basic physics of light and

diffraction.

 

I was also going to chime in with the digital picture tables of pixel sizes (posted by Mendel). Interesting note there:

5D Mk II pixel size = 20d pixel size = 6.4µm. The 20D diffraction limits are already displayed in the cambridge color

piece.

 

Taken together, here's my take relative to the original question of (paraphrased) is the crop sensor better at resolving

detail in the middle of a long lens:

 

There are more pixels in the middle of the image on a 50D than a 5D Mk II - simple math with the pixel size.

However, those "more pixels" may not produce better / sharper images. There can be many reasons: shooting

conditions, hardware or firmware differences of each camera and their processes of handling the information from the

pixels, etc. Beyond any general difference in the cameras, there will be a difference in the limits of physics and

diffraction. The crop sensors with their smaller pixels may out-resolve the FF larger pixels at wide apertures, but as

apertures are reduced, the sharpness of the images will be reduced more quickly in the smaller pixels of the crop

sensor. At some point in the megapixel race, pixel sizes will become so small that diffractions limits are reached

even with the fastest lenses. Of course, all of these limits may only be noticeable when pixel-peeping.

 

Fun theoretical stuff - but unlikely to affect the quality of my images much. Lighting, composition, basic camera

settings, white balance, holding the camera still, choosing compelling subjects - these will affect my images much

more than any effects I would get by upgrading my 20D to a 50D (unlikely) or a 5DII (likely - I desperately miss my

wide angles).