1Ds2 versus 5D

[Jack Flesher]

We should not lose sight of the fact that a lot of photography is about catching the moment. My interest in such hair-splitting differences is mainly so I can make an informed choice whether or not to ignore them. [a href=\"index.php?act=findpost&pid=70115\"][{POST_SNAPBACK}][/a]

I totally agree on both points!

[bjanes]

When doing comparisons, I like to keep the variables to a minimum. However, sharpening will help make the image look better (cosmetics) and I always do some sharpening for a finished print.

I took some test shots today of my neighbour's fence with my 100-400 plus 1.4x extender, with 20D and 5D. There's no sharpening and no interpolation, yet it's quite clear the 20D/560mm at f22 is the sharpest and the 5D at f11 the least sharp. There might be some doubt comparing the 20D at f11 and the 5D at f22. They look so close they are as good as equal as far as I'm concerned, but I give the edge to the 5D at f22.

[attachment=804:attachment]
[a href=\"index.php?act=findpost&pid=70073\"][{POST_SNAPBACK}][/a]

The results of the cameras at the same aperture are not surprising.

I measured the distance from one picket to another on the two images and found that the distance was 325 pixels on the D20 and 321 pixels on the 5D image. Since the D20 image is being viewed at 100%, the above dimension is also 325 pixels on the sensor. Since the 5D image is being veiwed at 127%, the corresponding dimension on the sensor is 253 pixels. The image is in effect upresed for display. The pixel pitch of the D20 is 6.42 microns, so this dimension on the sensor is 2087 microns (325 * 6.42). The image on the 5D is 2072 microns.

The above calculations show that the picture was taken from the same position. Since the same lens was used, the images on the sensor are the same. Since the D20 sensor has higher resolution in lp/mm, it shows greater detail. If Ray moved back with the D20 so that the field of view with both cameras was the same, the results would be different.

[Ray]

If Ray moved back with the D20 so that the field of view with both cameras was the same, the results would be different.
[a href=\"index.php?act=findpost&pid=70184\"][{POST_SNAPBACK}][/a]

Bjanes,
That's an interesting mathematical analysis, so you see, my images do not lie.

If you change perspective, then of course the results will be different. As far as I'm concerned my 20D is a 22mp full frame 35mm sensor that is regrettably cropped.

[bjanes]

Bjanes,
That's an interesting mathematical analysis, so you see, my images do not lie.

If you change perspective, then of course the results will be different.
[a href=\"index.php?act=findpost&pid=70218\"][{POST_SNAPBACK}][/a]

Ray,

Yes, the persepective would changed if you move back with the D20, but  that would make no significant difference, since the image has little depth. What I implied (and perhaps should have said explicitly) is that if the field of view were the same with both cameras, the 5D image would be sharper.

As far as I'm concerned my 20D is a 22mp full frame 35mm sensor that is regrettably cropped.
[a href=\"index.php?act=findpost&pid=70218\"][{POST_SNAPBACK}][/a]

That is a good way to look at it.

[Doug Kerr]

Hi, Gang,

Really interesting thread. Thanks to all who contributed.

[dwdallam]

When doing comparisons, I like to keep the variables to a minimum. However, sharpening will help make the image look better (cosmetics) and I always do some sharpening for a finished print.

I took some test shots today of my neighbour's fence with my 100-400 plus 1.4x extender, with 20D and 5D. There's no sharpening and no interpolation, yet it's quite clear the 20D/560mm at f22 is the sharpest and the 5D at f11 the least sharp. There might be some doubt comparing the 20D at f11 and the 5D at f22. They look so close they are as good as equal as far as I'm concerned, but I give the edge to the 5D at f22.

[attachment=804:attachment]
[a href=\"index.php?act=findpost&pid=70073\"][{POST_SNAPBACK}][/a]

That's a bad test. Lenses are best around f8-f11--Canon lenses.

[Ray]

That's a bad test. Lenses are best around f8-f11--Canon lenses.
[a href=\"index.php?act=findpost&pid=70300\"][{POST_SNAPBACK}][/a]

I suppose your next statement will be, "All lenses are equal."

[bjanes]

That's a bad test. Lenses are best around f8-f11--Canon lenses.
[a href=\"index.php?act=findpost&pid=70300\"][{POST_SNAPBACK}][/a]

I think the test is fine for what it was intended to demonstrate. Under the test conditions, the 20D gives better results than the 5D, because the 20D resolves more lp/mm even though the 5D has better resolution in terms of lp/picture height. I don't think Ray was trying to find the optimum aperture.

In any case, a perfect lens gives maximum resolution wide open. The best Canon lenses are not perfect, but for many I suspect that the optimum aperture is above f/8-f/11. Even my lowly 50 mm f/1.8 Nikkor gives optimum resolution at f/5.6. Beyond that, diffraction takes its toll.

[Ray]

I don't think Ray was trying to find the optimum aperture.

[a href=\"index.php?act=findpost&pid=70318\"][{POST_SNAPBACK}][/a]

Actually, Bill, that was part of the test. I tested all apertures down to the minimum of f57 with 1.4x extender. The testing is not yet complete. The next stage will be Norman Koren line charts and fine newspaper print from a distance of around 10 metres.

I'm getting the impression that the reason why many people who try using this combination of 100-400 with extender, and who find results are often of dubious value, is because they are clinging to the assumption that all 35mm lenses are best at f8 and f11. There's no doubt that the 100-400 at 400mm is best at f8 and f11, but even with a perfect teleconverter that introduces no aberrations of its own, that would translate to optimum performance at f11 and f16 with the 560mm lens.

Since a teleconverter is far from being a perfect lens and must unavoidably introduce further aberrations to the system, the sweet spot appears to be f22 for this combination, with both the 5D and 20D.

[dwdallam]

It's not going to mean very much to you if a complete stranger on this site tells you that he regularly makes 20x30 prints from 20D images and that they are perfectly satisfactory, is it? Peoples' standards vary. Jack Flesher probably wouln't have a bar of such a print   .
[a href=\"index.php?act=findpost&pid=70007\"][{POST_SNAPBACK}][/a]

You're right Ray, but I'm not asking what a 20x30 looks like with a 20D. I'm asking what a 20D 20x30 looks like compared to a 5D 20x30. If the majority of people say there is a pretty good difference, then that is going to mean a lot to me.

[dwdallam]

I suppose your next statement will be, "All lenses are equal."
[a href=\"index.php?act=findpost&pid=70311\"][{POST_SNAPBACK}][/a]

haha, no, but all lenses perform differently at different aperture settings and from what I read, differnt bodies as well. But if you your methodology takes all of that into consideration, then you must be correct.

[dwdallam]

I think the test is fine for what it was intended to demonstrate. Under the test conditions, the 20D gives better results than the 5D, because the 20D resolves more lp/mm even though the 5D has better resolution in terms of lp/picture height. I don't think Ray was trying to find the optimum aperture.

In any case, a perfect lens gives maximum resolution wide open. The best Canon lenses are not perfect, but for many I suspect that the optimum aperture is above f/8-f/11. Even my lowly 50 mm f/1.8 Nikkor gives optimum resolution at f/5.6. Beyond that, diffraction takes its toll.
[{POST_SNAPBACK}][/a]

Here is a defraction calculator and essay using the 20D. You can input the camera and all the other aspects that affect apeture to get the read out. Let us know what you think:
[a href=\"http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm]http://www.cambridgeincolour.com/tutorials...photography.htm[/url]

Of course I have to say I don't understand how this relates to the lens, unless this applies to ALL lenses on a  20D (Or 5D) etc. It seems that the 5D has a much greater range of apertures before it becomes defracted than does the 20D, which is around F8 to F13. But anyway, use the calculator and see what you get.

[jani]

Since a teleconverter is far from being a perfect lens and must unavoidably introduce further aberrations to the system, the sweet spot appears to be f22 for this combination, with both the 5D and 20D.

For what it's worth; this matches my experience exactly, too. I tested* a friend's 100-400 mm last fall, both with his 5D and with my 20D, and my 1.4x converter. I was briefly stunned.

But, interestingly enough, the 100-400 in question without the converter also seemed to be better at f/16 than at f/13 and f/11.

* I won't claim that these tests were the epitome of scientific studies ...

[bjanes]

Actually, Bill, that was part of the test. I tested all apertures down to the minimum of f57 with 1.4x extender. The testing is not yet complete. The next stage will be Norman Koren line charts and fine newspaper print from a distance of around 10 metres.

I'm getting the impression that the reason why many people who try using this combination of 100-400 with extender, and who find results are often of dubious value, is because they are clinging to the assumption that all 35mm lenses are best at f8 and f11. There's no doubt that the 100-400 at 400mm is best at f8 and f11, but even with a perfect teleconverter that introduces no aberrations of its own, that would translate to optimum performance at f11 and f16 with the 560mm lens.

Since a teleconverter is far from being a perfect lens and must unavoidably introduce further aberrations to the system, the sweet spot appears to be f22 for this combination, with both the 5D and 20D.
[a href=\"index.php?act=findpost&pid=70321\"][{POST_SNAPBACK}][/a]

Ray,

That raises another question. A 1.4x extender loses 1 f/stop. In the old days before lenses had electronic linkages, you had to make the translation yourself manually. If the lens was set at f/8, the effective aperture with the teleconverter was f/11 and you exposed for f/11. With my Nikon teleconverter, the translation is done electronically. The camera shows the aperture as f/11, but the diaphragm blades in the lens are at f/8 position and the image presented to the teleconverter has f/8 resolution characteristics, which are then degraded to a certain extent by the teleconverter.

How does your camera operate? When you say f/22 is that the effective aperture of lens+teleconverter or is it the actual setting of the lens only?

Bill

[PetterStahre]

Back from lunch I just stumbled upon this thread and couldn't stop reading. A lot of interesting information and opinions. Thank you all!

Since I own both bodies I decided to do a (very) quick test. I don't know if you feel it helps the discussion, but here it is...

I took a studio shot of a winebottle, with a dull design but with small print dots visible. (Probably one of the dullest photos I've taken!  

I used a tripod and mirror lockup. I shot in manual mode 1/200 s, but using flash meaning the effective speed was 1/2030 s (approx. burn time of the flashhead I used). All light in the image comes from the flash.

For once I used auto focus and aligned the cameras vertically and horizontally so that they spot measured focus from the same place (the right "7" in the "BIN 707").

I used the EF100/2.8 macro at f5.6 since it's suposed to be sharpest in center at that aperture. (Ref: http://www.photozone.de/8Reviews/lenses/ca...00_28/index.htm )

I shot in RAW and developed with RAW Developer, since that is the software I prefer when I want to extract details. Still I turned sharpening AND anti-noise off. Developed to AdobeRGB and converted to sRGB later.

I measured the white balance from the white paper in the background. RAW Developer reported a difference of 50 Kelvin. (Sensor differences?)

Here are tumbnails of the 5D (left) and 1DsII image I took:
 )

Cheers,
Petter

[dwdallam]

me personally I've made up my mind... I wouldn't reinvest in the 1DsII *purely* on resolution basis. (But I would reinvest in it, for other reasons  )

Cheers,
Petter
[a href=\"index.php?act=findpost&pid=70345\"][{POST_SNAPBACK}][/a]

That's an intiuging test, thanks.

Anotehr concern of mine is that it is not necessarily sharpening that deliniates which camera is producing details, but if the information is there in the first place, which has also to do with a lens. I've read that, for example, landscape shots, the 5D outperforms teh 20D in distant detail, such as fiolage--and no amount of processing is going to make it match the 5D. On the other hand, for close ups, like you show here, the difference, while still evident, is not as large as detail in distant foilage.

It would be nice to see what the difference is between the 1DsII and the 5D at distance shooting landscape and foilage.

[skid00skid00]

Petter,

Very interesting crops.  It looks like there's alot of lens aberation in the 5D shot, though.

Could you optimize that in ACR's lens tab?

Thanks!

[Ray]

But, interestingly enough, the 100-400 in question without the converter also seemed to be better at f/16 than at f/13 and f/11.

* I won't claim that these tests were the epitome of scientific studies ...
[a href=\"index.php?act=findpost&pid=70339\"][{POST_SNAPBACK}][/a]

Sounds like there might be some misfocussing going on. It's surprising how critical focussing can be with a long lens. When comparing camera bodies with the lens fixed to the tripod, I always remove the body from the lens, leaving the focussing the same. However, comparing 400mm with 560mm might require refocussing. I'm not sure about this, but that's what I've been doing, and occasionally I find that focussing is then perceptibly different.

[Ray]

With my Nikon teleconverter, the translation is done electronically. The camera shows the aperture as f/11, but the diaphragm blades in the lens are at f/8 position and the image presented to the teleconverter has f/8 resolution characteristics, which are then degraded to a certain extent by the teleconverter.

How does your camera operate? When you say f/22 is that the effective aperture of lens+teleconverter or is it the actual setting of the lens only?


[a href=\"index.php?act=findpost&pid=70342\"][{POST_SNAPBACK}][/a]

Bill,
That's how my camera operates also. With 1.4x extender attached, the maximum aperture is f8 and there's no way of setting it to f5.6. I'm assuming that the main lens is physically at f5.6 and the presense of the extender effectively closes the aperture to f8. Likewise, when I set the aperture to f22 electronically, I assume the aperture blades on the main lens are at f16.

[Ray]

I also realize I should have looked up an even more detailed object to shoot, but for me personally I've made up my mind... I wouldn't reinvest in the 1DsII *purely* on resolution basis. (But I would reinvest in it, for other reasons  )

[a href=\"index.php?act=findpost&pid=70345\"][{POST_SNAPBACK}][/a]

Petter,
These results seem very consistent with other reports I've seen on the net. A more detailed target, especially a line test chart, would probably have revealed a slightly greater difference (in favour of the 1Ds2), but for most folks it's real world images that count and there's hardly any difference worth bothering about, between these 2 cameras.