A Site for Photographers by Photographers

Home > Equipment > Canon > Canon EOS 5D vs. Canon EOS 20D — Full-Frame vs. APS-C sensors

Featured Equipment Deals

Photo.net Prints for Sale! Read More

Photo.net Prints for Sale!

These are just some of the exceptional images you can buy from the photographers on photo.net and have printed by CanvasPop!

Canon EOS 5D vs. Canon EOS 20D — Full-Frame vs. APS-C sensors

by Bob Atkins, 2006

You might think that it would be pretty easy to compare images from a Canon EOS 5D (full-frame 36mm x 24mm sensor) with a Canon EOS 20D (APS-C 22mm x 15mm sensor), but it's actually more difficult than you might think. You need to compare apples to apples, not apples to oranges and that's not quite so easy as it sounds. There are perhaps four different ways that you could use to compare image quality

1 - The first way is that you could use a different lens on each camera, each giving the same field of view. For example a 35/2.0 lens on the EOS 20D and an 50/1.8  lens on the EOS 5D would give you almost the same image in each case. However you'd then be testing the image quality of the two lenses as much as the image quality of the two sensors.

2 - The second way is using a zoom lens and equalize the field of view. So if you set the zoom to 70mm for the Canon 5D you would set it to 112mm for the Canon 20D. However you'd also be testing the image quality at two different focal length settings.

3 - The third way would be to use the same lens on each camera, but crop the EOS 5D image so it showed the same view as the EOS 20D. However if you do that you're testing a cropped full frame image, which isn't really fair.

4 - The fourth way is that you could use the same lens on both cameras, but shooting from different distances. Say a 35mm lens shot at 10ft from a subject with the Canon 5D and at 16ft from the subject with the Canon 20D. This is pretty fair test method since image quality isn't a strong function of focus distance in this range. The practical drawback it's not always possible to do this in the real world. You can't usually back up from 0.5 miles to 0.8 miles when shooting a landscape for example! However when just doing testing, this seems like the most reasonable way to look at differences caused by the sensor.


20d_vs_5d_vignetting.jpg (10618 bytes)

All lenses vignette, especially fast lenses shot wide open. It's usually worse with fast, wideangle lenses and better with slower, telephoto lenses. At apertures of f8 and smaller vignetting should be slight with any decent lens. Vignetting comes from a number of sources. First is the theoretical "cosine to the 4th power" falloff which all wide angle lenses exhibit, but more important is usually physical vignetting of the aperture due to the construction of the lens. A typical 50mm f1.8 lens can show up to 3 stops of vignetting at the corner of a full 35mm frame. Stopping down to about f8 can reduce this to about 1/2 stop. Since vignetting is a strong function of distance from the center of the frame, the "cropped" APS-C sensor shows less vignetting, especially when used with lenses designed for full frame illumination.

Center Resolution -  Test Method 4

Based on test method (4) as described above we can look at the center of images shot with a 50/1.8 lens at f8 from different distances. While the 50/1.8 is a cheap lens, its performance is good, especially when stopped down to f8, so this should be representative of good lenses.

Canon EOS 5D vs. Canon EOS 20D - Full Frame vs. APS-C Sensors

Clearly, I think, you can see that the full frame sensor of the EOS 5D shows more detail than the APS-C sensor of the EOS 20D.

Edge Resolution - Method 4

If we now look at the edge of the frames, we see the following:

EOS 5D vs. EOS 20D - Full Frame vs. APS-C Sensors

The first obvious thing is the vignetting again. The corner of the full frame shot from the Canon EOS 5D with the lens at f1.8 is clearly a lot darker than the center and a lot darker than the corners of the APS-C shot from the EOS 20D. In terms of resolution though, the full frame images are again superior.

Looking at a pretty average (or below) lens, here are a couple of shots from the corner of the frames using an EF 22-55/3.5-5.6 lens at 22mm and f3.5. This is a low cost, plastic mount, lens that was originally designed for use with APS film bodies, though which does have full frame coverage.

Canon EOS 5D vs. EOS 20D - Full Frame vs. APS-C Sensors

As you can see, again the full frame corners are darker due to vignetting. In this case I think the lower image quality at the corners of the full frame shot cancels out any size advantage.

Center Resolution - Test Method 3

Here's a series of shots using test method 3, i.e. with the same lens at the same aperture and at the same distance from the subject:

Canon EOS 5D vs. EOS 20D - Full Frame vs. APS-C Sensors

As you can see, in this case the 20D image shows more detail. This situation (same lens form same distance) is the situation that nature photographers often find themselves in. They're using their longest lens (typically a 500mm or 600mm, maybe with a multiplier on it) and they're as close to their subject as they can get. They might use a longer lens if they had one, but they don't! Under these circumstances, where you'd have to crop one or both images to the same field of view, results from an APS-C sensor with a higher pixel density (such as the EOS 20D) can give better results than those from a full frame camera with a lower pixel density (such as the EOS 5D)

Real World Shots - Test Method 2

Here's a real world image comparison. shot using test method (2), i.e. a zoom lens at different focal lengths on the Canon 5D and Canon 20D. This would probably be the practical situation if you had both a 20D and a 5D. You'd probably use the same zoom, but at different settings. I tried to equalize image quality by shooting at f11 and shooting at the wide end of the 70-300/4-5.6 IS USM, where image quality is high and doesn't change a lot. In addition to comparing image quality in terms of resolution, you can compare EOS 5D vs. EOS 20D noise levels since these images were shot at ISO 3200. First, here's the whole image:

EOS 5D vs. EOS 20D - Full Frame vs. APS-C Sensors

The area shown outlined in red is shown as a 100% crop in the two images below:

EOS 5D vs. EOS 20D - Full Frame vs. APS-C Sensors

And finally here's the same two images, but with the EOS 20D shot upsized by about 125% (using bicubic interpolation) so that the subject is the same size in both shots:

EOS 5D vs. EOS 20D - Full Frame vs. APS-C Sensors

The superiority of the Canon EOS 5D image is clear. Now they were shot at different zoom settings, but since they were also both shot at f11 and the zoom is pretty good in this focal length range, I think what you see in the images reflects the difference between a full frame 12.7MP sensor and an 8.2MP sensor rather than any difference in lens quality.


It's a win, but not a total "slam dunk" for the full frame sensor, at least as far as the EOS 5D vs. EOS 20D goes. With good lenses the results from the Canon EOS 5D are better, though with noticeably higher vignetting when lenses are used wide open. Of course you can mostly compensate for vignetting in software, but still you may lose something. With low cost lenses shot wide open, the full frame advantage disappears, and in situations where you have your longest lens in use on an APS-C camera, switching to a lower pixel density full frame camera and cropping will lead to lower resolution images.

However there's no denying that under the right conditions, the images from a full frame camera like the Canon EOS 5D can beat those from an APS-C camera like the Canon EOS 20D, as the real world examples above show. Given the budget, I'd own both a 5D and a 20D and I'd use each in the situation for which it was best suited. Since I do a lot of nature work, the higher pixel density of the EOS 20D is useful for me. If I didn't shoot extreme telephotos, I'd certainly pick the Canon EOS 5D over the EOS 20D, especially if I had a good set of lenses to go with it. Good here doesn't mean all "L" series either. It's clear that the better "consumer" grade lenses can also take advantage of the higher performance of the full frame Canon EOS 5D.

Canon has made a significant step in DSLR technology by producing a full frame camera which is selling for under $3000. They seem to be on a path which may end up - in maybe 3 to 5 years time - with a full frame "prosumer" DSLR sellng perhaps in the $1500 region. However currently, and for the forseeable future, the EOS 5D is second only to the $7100 Canon EOS 1Ds Mk II in terms of image quality and is, I think, the best bang for the buck out there if you're looking for the highest image quality from a DSLR.

Where to Buy

Purchasing through the following links helps to support photo.net.

Featured Merchants
Search for similar products on PriceGrabber
Search for similar products on PriceGrabber
Search for similar products on PriceGrabber

Related Reviews

Canon 20D review
Canon 5D review
All Canon reviews

Test and images are © 2006 Bob Atkins ( www.BobAtkins.com) All Rights Reserved

Readers' Comments

Add a comment

Alfie Wang , February 07, 2006; 03:14 P.M.

Wow, 5D looks bad relative to the 20D... good thing Nikon didn't jump to full frame :)

Bob Atkins , February 07, 2006; 04:30 P.M.

There's absolutely no justification for assuming that anything like:

1/r = 1/r(sensor) + 1/r(lens)

Anything based on that premise is just speculation.

It's pretty much true that digital resolution is mostly sensor dependant unless you have the lens stopped down so far (say f16 for APS-C) that diffraction dominates.

Resolution is also a poor criterion for image quality since what realy counts is the MTF at the critical viewing spatial frequency. What counts most in sharpness for viewing 8x10 or 11 x14 prints from a couple of feet is the system MTF between about 5 cycles/mm and 25 cycles/mm. Larger sensors have an advantage here since they are working on a higher part of the lens' MTF curve.

See http://www.bobatkins.com/photography/technical/mtf/mtf1.html

See also http://www.bobatkins.com/photography/digital/size_matters.html

Ben Boule , February 07, 2006; 05:01 P.M.

It is interesting to note that on both cameras it is really hard to shoot a quality real world image which is ruined by vignetting/light fall off. (Not counting putting the wrong filters on, etc..)

Shooting white sheets under studio conditions really exaggerates the issue IME. Different light setups in the real world make it much harder to see.

This test would indicate that the 50/1.8 lens is a problem even on the 20D, in reality it works fine @ f/1.8 on both APS & 35mm for appropriate uses, and even if your favorite type of shooting is white sheets stopping down to f/2 or f/2.8 drastically reduces the fall off.

ISTR the only image I have shot which actually showed fall-off in the real world on my 5D was a hand held shot @ f/1.4 at dusk. It was underexposed and poorly focused because it was handheld, so the fall-off was interesting but not real meaningful. The correct thing to do was shoot on a tripod to solve the exposure issue, which eliminated the fall off.

Brian Y , February 07, 2006; 08:26 P.M.

Thanks for sharing this.. very interesting. Another good comparison between FF and a higher pixel count DX sensor was done by Bjorn. NOTE ALL: please ignore the camera brands in the test and look at the sensor size and image results. http://www.naturfotograf.com/D2X_rev06.html#top_page

Wayne Campbell , February 07, 2006; 08:52 P.M.

"Clearly, I think, you can see that the full frame sensor of the EOS 5D shows more detail than the APS-C sensor of the EOS 20D." ------it's not so clear to me. In fact, it looks in many cases that the APS-C sensor is sharper than the full frame. i.e., Center resolution Test 4.

Ben Rubinstein - Manchester UK , February 08, 2006; 07:37 A.M.

I may be being thick here but having used the 20D and now shooting extensively with the 5D, these tests, (not including the corner vignetting issue), all seem to be showing that the 5D has more resolution than the 20D, which is something we know. There is far less here to do with crop vs FF, the only way to fairly test that would be with two cameras of equal resolution would it not?

Remco Jan Woldhuis , February 08, 2006; 09:14 A.M.

If you see these results I can't justify the price difference. Allthough the 5D is relatively affordable it is still much more expensive than a 20D (and with similar build quality). (you can buy a few 20D's for 1 5D). The picture quality is only slightly better in the best cases.

John Souleles , February 08, 2006; 11:00 A.M.

After reading the article, I think another conclusion would be that an investment in a 5D doesn't stop at the body. It must include an upgrade to the best glass to take advantage of the sensors size and quality. Should Canon release a FF sensor with the same density of pixels that the APS-C sensor has, we might see the unrivalled and undisputed quality "jump" we are all expecting. However, again, the quality bottle-neck will rest with the lenses.

Perhaps, this is a silly question but, I wonder about the dynamic range of the sensor/lens combinations. Are they the same ? Is this irrelevant Bob?

Cheers John

Per Kylberg , February 08, 2006; 03:15 P.M.

This is a very strange comparison: An "old" 8 MP aps-size sensor camera agaist a newly developed 13 MP "full size" camera. If the later wasn?t better it would be a catastroph for Canon! The 20D does pretty well I think. It would have been more intelligent though, if you want to analyse difference between APS-size and "full frame", to choose a APS camera with the same amount of pixels as the 5D......... And which would that be?

Ben Rubinstein - Manchester UK , February 09, 2006; 10:51 A.M.

Comparing a D2X against the 5D, both at 5D so there is no noise issue, would have been a closer 'Fullframe vs. APS-C sensors' comparison would it not? Though then again you run into the difference between lenses, etc.

Aaron Buckley , February 09, 2006; 11:45 A.M.

Well great... Now I want a Full Frame Nikon even more...

Tim Bosley , February 09, 2006; 01:57 P.M.

Good read as usual, but something did not sound quite right:

"2 - The second way is using a zoom lens and equalize the field of view. So if you set the zoom to 70mm for the EOS 5D you would set it to 112mm for the EOS 20D. However you'd also be testing the image quality at two different focal length settings."

Should that not be 70mm for 20D and 112mm for 5D?

Martin Datzinger , February 09, 2006; 06:52 P.M.

Sadly, there's still a lot of religious believe in the advantage of the APS-C format and obsession in resolution and corner performance. At the same time, people upgrading to cameras such as a D2X run into problems getting sharp pictures out of their cameras because

  1. the resolution is so high,
  2. the glass and/or sensor is not fast enough to provide fast enough shutter speeds,
  3. improper handholding technique is used,
  4. no or insufficient tripods are used or
  5. there is no other image stibilisation device.

I ask myself how hundreds of thousands of photographers could take pictures in the 35mm film days if corner quality/light falloff was such a major obstacle back then. I guess they used a device called aperture to help themselves deal with that problem back then. Or they didn't take a lot of pictures of brick walls or empty skies. As for the myth of corners being especially problematic on digital FF because of light rays not hitting at a near perpendicular angle - I don't buy this Nikon/Olympus marketing baloney. Just look at the 18-70 DX lens. Full open it shows severe vignetting and corner blur/CA in the same way that lets say the 28-105/3.5-4.5 Nikkor did back in film days. In contrary - there is software like DxO, that can make corner performance on digital FF better than it ever was on film. Saying "We throw away more than 50% of your lenses image real estate and take away lots of your fast primes (e.g. 28/1.4) magic because we can't make FF sensors at a reasonable price. But hey - our lenses are useless in the corners anyway!" sounds very much like "It's not a bug, it's a feature" to me. And please let us not forget that an APS-C sensor with the same resolution as a FF sensor is 1.5x more demanding on (center) resolution of the lens.

The argument that the very best lenses have to be chosen for full frame to show full advantage is rather questionable IMO. As soon as one puts subject separation via selective focussing above corner performance and takes noise levels into account it becomes obvious, that "full frame glass" is even cheaper than equivalent "APS-C glass" at comparable FoV. Just compare the prices of a 24-105/4 L IS with a 17-55/2.8 DX, a 50/1.8 with a 35/1.4, a 85/1.4 with 135/2.0 or a 17-40/4L with a 12-24/4 DX (the latter should be a 2.8 lens to make this valid). And for lenses used on FF sensors like 24/1.4, 28/1.4, 35/1.4, 50/1.4, 24(28)-70/2.8 or 16(17)-35/2.8 there are simply no equivalents in terms of selective focussing and low light noise levels at the same FoV in the APS-C world. I really wouldn't want to be someone having spent a fortune on these lenses and having used them on 35mm film just to realise that these devices are not allowed to do what they where intended for on an APS-C sensor.

I hated that cropped sensor stuff when it was invented because FF sensors where too expensive back then, and I still hate being bound to it with the brand I use (yes there is a 35mm F mount digital camera, but how usable is that in the real world?). But maybe people like me are the minority of non-pixel-peepers.

Kind regards, Martin

Martin Datzinger , February 09, 2006; 07:05 P.M.

BTW, of course it can be said that being able to use lenses the way they where intended to be used, selective focussing and low noise are the only three reasons for FF sensors. But I think they are far more important than a nominal 22% linear resolution increase as in case of the 5D vs. 20D. I really wonder why this hasn't been mentioned in the above review.

John Souleles , February 10, 2006; 02:01 P.M.

"BTW, of course it can be said that being able to use lenses the way they where intended to be used, selective focussing and low noise are the only three reasons for FF sensors. But I think they are far more important than a nominal 22% linear resolution increase as in case of the 5D vs. 20D"

Ok,...., for pro's,..., quite frankly IMHO, those of us who do not see noise control as a major issue,(with software control)..for amatures/consumers the 5D makes little sense,..., I believe what the review suggests is that while pro's can justify the 5D body given the resolution, the lens issue/crop factor debacle, and the low sensor noise, We self professed garden variety non-income based amatures have a more dificult time justifying the cost.

I think part of the frustration, as it is with all of these discussions,is that we should all probably just go out and take more picutres, and perfect a wonderful craft,...., but this is also a forum for technology and it's appropriate use and merits.

I suppose I could buy a 5D and mate it with a Phoenix/Cosina 28-105mm f/2.8-3.8 & exclusively shoot at f8.

Bottom line 20D is a prousmer product, a 5d is a back-up body to a EOS series 1. Different people use them.

Ilkka Nissila , February 10, 2006; 04:13 P.M.

It's obvious that larger sensors make creating sharp images with good tonality much easier. This is nothing new. The only problem is that it's expensive, and manufacturers are reluctant to bring out products which lose money. Kodak made really nice and relatively inexpensive back for 645 cameras. Gone, not profitable enough. Contax and Kodak made FF 35 mm DSLRs - gone, not profitable enough. Canon is the only one that can make money out of reasonably priced FF DSLRs. The next step up from the 5D is such a big step in cost that you really need to _need_ it to pay for it.

So that is the situation: only one company sees larger-than APS sensors as profitable below four-figure prices. If you need one, buy it from them, it's as simple as that.

People who think small sensors are sufficient are just ignoring the fact that needs in art are highly individual and there are many people for whom these sensors are not sufficient. Are we enough to generate a big enough market for larger sensors? It doesn't seem to be the case. We can always pay $40000 for a nice Phase One back, you know, if DX is not enough. Stop complaining and buy the product which matches your needs.

Bob Atkins , February 10, 2006; 04:48 P.M.

It would have been more intelligent though, if you want to analyze difference between APS-size and "full frame", to choose a APS camera with the same amount of pixels as the 5D......... And which would that be?

Please, you provide me with the cameras, and I'll do the tests. Better still, you do the tests, write the article and I'll be more than happy to publish it here.

If Nikon send me a D2x and a lens or two I'd be happy to test it against a Canon EOS 5D - if you can also persuade Canon to send me a 5D with similar lenses at the same time.

I think a 20D vs. 5D comparison is much more useful (and practical) for the majority of readers. Personally I don't care how good the D2x is because I'm not switching to Nikon! I'd assume Nikon users feel the same way and they wouldn't be switching to a 5D even if it were "better".

If you're trying to decide whether to buy a 5D or D2x, you're in the minority I'm afraid.

Doug Peck , February 10, 2006; 07:03 P.M.

Thanks for this review; no big surprises, I think, but useful comparisons. I think a key factor was hit by a previous commenter who said the FF sensor in the 5D lets you use lenses as they were intended. For example, for portraiture I want around 100mm for individuals and 50mm for groups. It's hard to do this in prime lenses with a 1.6 crop factor, so I use a 24-70 zoom. Also the justification for FF becomes stronger for landscape or other photographers who like to shoot lots of wide angle.

Ilkka Nissila , February 10, 2006; 10:22 P.M.

For landscapes FF doesn't seem appropriate, just check the corners of Canon's own example taken with the 5D+17-40 well stopped down. A 12-24 + D2X seems the way to go for that, or a medium/large format system

Umit D , February 11, 2006; 05:44 A.M.

Just compare the prices of a 24-105/4 L IS with a 17-55/2.8 DX, a 50/1.8 with a 35/1.4, a 85/1.4 with 135/2.0 or a 17-40/4L with a 12-24/4 DX (the latter should be a 2.8 lens to make this valid). And for lenses used on FF sensors like 24/1.4, 28/1.4, 35/1.4, 50/1.4, 24(28)-70/2.8 or 16(17)-35/2.8 there are simply no equivalents.. - Martin Datzinger

That's quite a bit demagogy. I mean hand picking the examples that favour your point of view and omitting others. How about 50mm/1.4 on aps vs 85mm/1.4 on ff? 200/2.8 vs 300/2.8? 300/4 vs 400/4? 18-200 VR vs 28-300 IS? On telephoto and zoom side (range) aps format has the advantage, on the wide angle there are more options for full frame, but there is also the corner issues, even Canon's own sample demonstrates this fact. If you check the Canon 1dsMk2 (FF) and Nikon D2x (aps) comparision at www.dpreview.com you can see there are corner issues even in moderate focal lenghts like 70mm which show up as more color fringing.

By the way even your hand picked examples are misleading, 85/1.4 lens doesn't become a 135/2 lens on aps format, it still is 1.4; 17-40/4L vs 12-24/4 is a valid one though you say latter should be 2.8 for a valid comparision (!); and you compare 24-105/4 with 17-55/2.8 which are of different apertures.

Umit D , February 11, 2006; 05:57 A.M.

As for the myth of corners being especially problematic on digital FF because of light rays not hitting at a near perpendicular angle - I don't buy this Nikon/Olympus marketing baloney.

I do, because real world evidence support this and I can imagine why it happens, a light ray coming at an angle affects more pixels than one coming perpencilularly to the sensor, this is simple geometry, question is why this is more an issue with digital than film. I guess for three things, first film is quantized in smaller particles than pixels, second film grains are in random alignment and in multiple layers where pixels are strictly arranged in a pattern, third sensor capture is interpolated to form an image, that is, a pixel in the actual image is formed by the help of neighboring pixels, which means any bad data recorded in a pixel propagates to pixels nearby.

Digital capture with existing technology has its well known aberrations like stairstepping and moire compared to film, poor response to tilted light rays can be (and seems to be) one of them as well. Btw I am neither professional nor enthusiast on this matter and would appreciate opinions of optics or solid state electronics experts.

Martin Datzinger , February 11, 2006; 08:32 A.M.

I mean hand picking the examples that favour your point of view and omitting others.

As I said I've compared lenses with the same eqivalent FoV, minimum DoF and allowing for comparable low light noise levels*. On that basis my comparison is valid. See: DOFMaster for a comparison of DoF. You will see that a lens with a 1.6 times shorter focal length needs (even more than) a 1 stop wider aperture to accomplish the same level of subject separation. At a distance of 3 meters, a 50/1.4 (EUR335 in Austria) full open gives 19 cm DoF on a 20D, a 85/1.8 (EUR366) gives 13cm full open on a 5D and is thus better in terms of image separation.

*) I am certainly aware that under the same conditions the Canon 20D generates about the same amount of noise as a 5D when the output of both is seen at 100%. However, printing images from both at the same size will make the 5D's noise less visible because noise frequency is raised. Or the other way around, stronger noise reduction can be applied while retaining the same amount of image details. In theory, sensor locations of twice the area give 3db or 1/2 stop better S/N ratio. See this paper at page 3.

Umit D , February 11, 2006; 08:56 A.M.

DOF argument is a pretty weak one, because higher DOF of aps format can be advantegous depending on the situation you shoot, in my opinion having the more DOF at lower light is a greater advantage. If we extrapolate from what you say medium format must be cheaper than 35mm. Is it?

Martin Datzinger , February 11, 2006; 09:54 A.M.

Higher DoF is always an option on 35mm via stopping down and cranking up ISO giving about the same practical results as what can be done with a smaller format. But with current lenses, limited DoF can't be created so easily on wide or normal FoV with formats smaller or bigger than 35mm. I guess a 24mm/1.4 DX Nikkor for EUR400 or 35mm/1.0 DX for EUR350 won't happen in near future, in the same way that there are no new 80/2.0 or 50/2.8 medium format lenses available at these price levels. Subject separation is the 35mm format's party piece. If one doesn't need it and noise at ISO 800 and above is irrelevant, APS-C, 4/3 or compact digicams are just fine and of course much cheaper and less bulk, too. Kind regards, Martin

Umit D , February 11, 2006; 10:54 A.M.

Enough subject separation? D70, 85/1.4 AF-D

Martin, as your portfolio suggests you seem to be a dedicated portrait shooter and I conclude you therefore put so much emphasis on DOF but even for portraits so precise "DOF management" is an overkill, if you are after DOF manipulations (as Keith Carter does) forget it in 35mm, you should use a tilting camera (he uses a Hasselblad flexbody).

Low light shooting, handholding necessity and action stopping are other application areas for fast lenses and usually these have higher priorty than strict DOF control. If cranking up ISO was always a viable option then all fast lenses would go extinct in these application areas but they are not, similarly stopping down the lens AND cranking up iso for high DOF isn't always desirable.

Have you shot with a real fast lens (even in aps format) and managed to get things in focus at portrait distances? I have great difficulty doing that, it is almost impossible for me without AF.

Martin Datzinger , February 11, 2006; 11:56 A.M.

50/1.4 full open on D70 at ISO 1600 with a single 100W lamp in the room. Cat and camera handheld, image cropped to 1/3 :)

Umit, thanks for pointing me towards Keith Carter. Tilt lenses would leave be both bonkers and bankrupt, I guess :)

Narrow DoF at facial portrait distances isn't what I am missing on my D70. My 85/1.8, even stopped down to 5.6 does plenty in this regard. I'm craving street photography with fast wideangles such as this. But hey, maybe I'm getting a used F100 body and B&W film equipment for that application.

Kind regards, Martin

Ben Anderson , February 11, 2006; 05:23 P.M.

Yaron Kidron said that on the 20D f/11 may be approaching diffraction limits, I thought that diffraction was a function of the lens aperture explicitly, and the sensor has no bearing on it at all? That's certainly what I'd expect from my schoolboy physics.

Yaron Kidron , February 12, 2006; 10:05 P.M.

Ben, Large format shooters can shoot at f/45 without being diffraction limited. The smaller your emulsion, the "faster" diffraction limited you become.

Ilkka Nissila , February 12, 2006; 11:57 P.M.

The blurring due to diffraction is constant for a given aperture and lens FL. However, if you have a small sensor, you need to magnify the image more in post-processing and the effects of diffraction are also magnified.

The only reason large format photographers can shoot at f/45 and get good results is because 1) the lenses aren't very sharp at wider apertures so they're not diffraction limited at (say) f/5.6, 2) the area of film is large, hence post-exposure magnification is low.

Ben Anderson , February 13, 2006; 02:42 P.M.

OK, that makes perfect sense. Thanks for enlightening me.

Dan Zimmerman , February 16, 2006; 11:04 A.M.

"...does FF digital produce more vignetting than film?

Not to be dense or anything, but how would a 24X36 digital sensor produce any more vignetting than a 24X36 frame of film?

Nicolas TREGOURES , February 16, 2006; 01:07 P.M.

"Not to be dense or anything, but how would a 24X36 digital sensor produce any more vignetting than a 24X36 frame of film? "

Rays of light coming out from the lens reaching the side part of the sensor are not perfectly perpendicular to the sensor (simple geometrical optics). Some people claim that digital sensors are less sensitive to non perpendicular light rays than to perpendicular light rays (which happen in the center of the sensor). According to them this induces more vignetting on a 24x36 digital sensor than on a 24x36 film since film is not very sensitive to this effect.

I don't know who is right but it should no be very difficult to find out.

You can take a look at


and also look at this surrrealist thread



Scott Eaton , February 20, 2006; 09:55 P.M.

I conclude you therefore put so much emphasis on DOF but even for portraits so precise "DOF management" is an overkill, if you are after DOF manipulations

At least with FF you have some control with DOF vs APS where a you essentially have to guess because the pathetic viewfinders are so small and dimm.

Ilkka Nissila , February 24, 2006; 12:47 A.M.

You don't have to guess anything. With the D200 you can zoom to see what seems to be every pixel captured. So you can check on-site how sharp each part of the picture is just like with PS. And the viewfinder shows sharpness and depth of field quite well too.

Terence Mahoney , February 26, 2006; 12:14 P.M.

Bob: I quite agree that your first method is the only one with a glimmer of validity, and in those examples whilst there is clearly a cleaner rendition of black from the 5D and thus an apparent increase in contrast, quite correctable in post-processing, the differences in resolution are quite minor and these crops are extreme to say the least. In the Method 4 had you properly set the zoom to begin with it should not have been necessary to upsize and interpolate those crops, so you've in fact introduced an unnecessary variable.

From the results of your testing one can clearly see that there is certainly not, to the rational mind, a performance advantage in the 5D commensurate with the doubling (considering now the price of the 30D)or trebling (eventual pricing of the 20D now that it's been obsoleted)of the cost. In fact to the consciencious photographer the vignetting and corner-performance at the limits of the full-frame image circle in un-cropped photographs far overshadow the minusucule resolution differences noticeable only at extraordinary magnification. Therefore it distills down once again to pure ego whether or not one chooses to turn a blind eye to the obvious fact that the law of diminishing returns is in full force here.

Mike Dixon , March 07, 2006; 12:20 P.M.

Terence, which lenses would give me the same control over DOF, usefulness for handheld shooting in low-light, and angle of coverage on a 20D or 30D as the 50/f1.4, 28/f1.8, and 20/f1.8 do on a 5D?

Bob Atkins , March 09, 2006; 10:53 P.M.

One could ask which lens on the 5D will give me the same angle of coverage with the same IS system and speed that I get with a 600/4L IS on an EOS 20D.

Ilkka Nissila , March 13, 2006; 10:36 A.M.

Well, you could mount a Nikon 800mm f/5.6 (you'll be probably shooting at max aperture anyway) with an adapter and use one stop higher ISO setting (if the sensor and lens technologies were equal, this would give roughly the same results).

Marc Montocchio , May 02, 2006; 10:53 P.M.

Application, application, application, so many of the above posts assume pro's from shooting the APS format in long lenses, even medium applications. I have worked as a pro underwater photographer and digital formats drove me mad. My need for my workhorse 16mm fisheye, 14mmm 2.8 and 20mm fixed focal wides just doubled the size of my travel case, film and digital. Nikon owed pro underwater photography since start of Nikonos 60 years ago, digital changed that. Canon full frame products have caused big underwater shooters and housing manufactures to start making top end Canon housing products. Now Nikon loyalists are stuck with big bucks in Nikon lenses and wide's that just got a whole lot narrower. When it comes to underwater wide angle photography, APS bodies need to box in the same weight division as full frame sensors. I would like to see a test underwater with the 5D and Canon 14mm or 15mm against the D2X and only DX lens they have with the same angle of view in the APS format. Although I have been a Nikon guy since I started, I just had all my lenses and Kodak Pro 14n SLR stolen. I'm left with a D1X, F5, 2 F4's, F100, a pile of underwater housings and more Nikonos gear than most retal houses, I am selling everything thing I have left and starting again. My choices are D2X or D5, because of the full frame I'm switching to Canon after 20 happy years with Nikon. Apples to apples, application and output are everything.

Julian Li , June 01, 2006; 07:11 P.M.


Gregory L-S Minneapolis , June 10, 2006; 03:08 P.M.


It looks as if this thread may be out of steam, but I still want to add my two cents.

Most of the guts of digital camera body are printed circuits - microelectronic devices - and as such will be orders of magnitude more capable and less inexpensive in a few years. That is obviously not true for lenses.

For all but the very few of us who have many tens of thousands of doors to spend in camera equipment, an extra $1500 spent on an L-lens we don't yet own would add far more sharpness to our pictures than another 4 megapixels. The capabilities of a new lens might even add whole new opportunities to our photography, something that the 5D will certainly not do. I lust for the 180 mm f3.5 macro and the 85 mm f1.2.

It's also important to remember that megapixels are spread out over a square, ie a TWO dimensional, surface. Thus the improved resolution, even aside from the sophicated and interesting considerations of Bob Atkins, are not (12/8) = 1.5 times more sharp, but Square Root (12/8) or about 1.25 times more sharp. And this is what his tests really showed: the 12 megapixel sensor is noticeably, but not dramatically sharper. But what a price difference. Whew!

And now I'm going to be really radical: BODIES ARE DISPOSABLE, BUT GLASS LASTS.

In 5 years of microelectronic progress, I'm going to want a new body and I'll probably be able to convince my wife in 7-10 years. There was a Doonesbury cartoon a few years ago with Mike sitting at a new computer that was "so fast it makes my fillings hurt." In the last panel he sees a notice on the back, "Best if used by August 15."

Predictable obsolesence is especially acute with ALL of Canon's current bodies: they don't have any way of cleaning the sensor. All camera manufacturers are going to have to solve that one in the next 2-3 years to keep Olympus from stealing market share. And who can predict what new functionality will be in the cameras of 5-10 years from now. Autofocus seemed like the impossible dream 25 years ago

Okay, so if camera bodies are obsolete the moment you open the box, then "build quality", at least within the group of prosumer - pro cameras, doesn't seem relevent (unless you are a pro doing very rugged work). In fact, for most of us, whether or not we have the camera with us in the first place is going to be the biggest determinant of the "quality" of our photos. So why buy even the 20D? Why not the Rebel 350 XT? Same sensor. Two fewer focusing points. Fractionally slower continuous shooting modes. I don't know what all else. But it's not very significant while the difference in weight and size is immediately apparent when you heft the two cameras.

I bought the XT about 6 months ago and couldn't be happier. I upgraded the glass to the 17-85 mm f4-5.6 (Ok. Ok. I know this is still a mid-quality lens but it's fairly light and has higher optical quality and more versatility then the standard 18-55 mm). I have used it to take some pictures that make me very happy and I see some beautiful postings on photo.net made with the XT.

So, I think I'll want a new body every few years and and I'm not going to put big bucks there, but I do want the best lenses I can afford and plan on using them for a lifetime.

Rich Morgan , September 24, 2006; 01:31 P.M.

Olympus increasing its marketshare? IMO, Olympus will be lucky to survive.

rick byerly , February 08, 2007; 12:17 P.M.

was surprised to see that there were no comments about the no crop 35mm issue which has been a thorn in my side as i use a canon rebel from 2004- coming from someone who sells 8 by 10 and 16 by 20 prints i thought the 5d would be a definite favorite over xt.

anyone see a good reason to upgrade from my rebel to the new xt? or 5d?

i focus on nature mostly- macro, some landscape and just started some stock work




Ruud van Gaal , May 02, 2007; 06:09 A.M.

Interesting read, going on for a long time. :)
I mostly do portraits, now have a 350D (RebelXT?) and am looking at the 5D as an upgrade. Mostly because of crop size; I like small DoF's and this sensor combined with fast lenses would give me that, I would say.

Funny is that I often intentionally vignet my photographs. :) It just seems to add some more intimacy/abstraction to the picture.

What do you say; would the 5D give noticable small-DoF improvements compared to the (1.6 crop) 350D?

Dolphinity Photography

Rob IJsselstein , August 09, 2009; 12:57 P.M.

The article was very useful. I am now considering buying a new camera. FF or APS? Nowadays most cameras give around 12 megapixels. That makes comparing between FF and APS surely different. Suppose I want to make a picture of a scene, and frame it the same with the two cameras; so with the FF I set my zoom on 75 mm, and my APS on 50 mm to get the same crop. Then the image will be recorded by the same amount of pixels, presumably also with the same total amount of light. What would be the difference? I can think of more noise for the smaller sensor. What else? Suppose I don't use a tripod; will my vibrating arm result in more or less blurr in either of the two cases? Generally a 50 mm setting is easier to hold still than a 75 mm one. Does anybody have answers to these questions?

Add a comment

Notify me of comments