It has often been stated that 5D auto-focus is GREATLY inferior to Series 1, that 5D is not suitable for wildlife photography.

A related idea is that Series 1 will do standard precision auto-focus up to f8 and high precision up to f4, while the non-Series 1 bodies require f5.6 and f2.8 respectively. Since 500mm f2.8 lenses are quite a rare breed, this implies a considerable sharpness advantage using Canon 500 f4 and 1D2 vs Canon 500 f4 and 5D.

Since I own this lens, and shot both bodies every single day for 3 months on Yellowstone wildlife, I decided to prove this to myself. I specifically constructed tests to prove that the 5D was inferior in auto-focus and consequently inferior in sharpness.

My first test was high contrast light, high contrast subject, one shot mode with center focus point selected. I photographed a brown Yellowstone sign with white lettering in full sunlight at f4 at about 200 feet. The lens and body were mounted on a Gitzo 1410 aluminum tripod with full Wimberly head. This tripod/head combination weighs 12 pounds and is only slightly inferior in steadiness to a 200 pound concrete block.

I shot a series of auto-focus tests, each time manually focusing the lens at 20 feet, and forcing the lens to auto-focus at the 200’ distance. After trying this many times with each body, I shot the same series again doing manual focus, as a baseline for lens sharpness. I examined all of the center crops side by side in Photoshop at the same time. The result was, even at 100% pixels, there was no difference sharpness from either body, with the lens at f4. I had failed to prove that 5D was inferior.

O.K, back to the drawing board. I selected an extremely low contrast target, on a low contrast, snowy overcast day. I selected an area of weathered brown wood of an old building. I set both bodies to AI servo, with center focus point selected. I realize I was shooting a static subject in a motion auto-focus mode, but I wanted to prove AI servo was much better in Series 1. I shot a similar series of pictures, forcing the bodies to refocus for every single shot. When examined at 100% pixels, the 1D2 pictures were very, very slightly sharper. Let’s say if the 1D2 was 100% sharpness, then the 5D was about 95% as sharp. I don’t think this difference could be seen in a 16”x24” print, but I did not print them; probably should have. O.K. I failed again to prove the 5D was inferior in auto-focus.

The next test was two days of shooting birds in flight at Bosque del Apache, New Mexico. I was using the 500 on the Gitzo, with the Wimberly head loosened for tracking the bird’s flight. AI servo was used, with center point selected and adjacent points enabled. In initial focus acquisition, if the 1D2 locked on in 1 unit of time, the 5D took 1.2 units. If I lost focus by letting the center point drift off of the bird, the 5D took 2 to 2.5 times as long to re-acquire focus. So I saw little difference in focus speed, if I was able to keep the center spot continually on the bird. Sharpness, after both bodies had locked on the bird, seemed to be the same.

I preferred to shoot Bosque in single frame mode, not a 8.5 frames per second, since there seemed to be one split second in time where the bird was closest, and correctly framed. I prefer to time the exact, correct moment, and not just machine gun it. This also has some benefit in editing, since even my slow shooting resulting in too many pictures to edit. The 1D2’s lightning fast frame rate and shorter black-out time was quite noticed and appreciated, but I didn’t see that this resulted in any better pictures. I judge a camera by the result it produces in the final print, and not on what an awesome machine it is. A camera is like a screwdriver or hammer, it’s just a tool on the way to the final result.


I discovered one slight advantage of the 5D, in birds in flight, only at home after the shoot. I tried the fill the frame as much as possible with each body for maximum picture quality. But accurate tight framing is quite difficult with the birds twisting and turning. If the bird perfectly filled the 1D2 frame, the 5D, at the same distance, could be cropped to equal quality, since both bodies have the same sensor pixel density. But a wing clipped off on the 1D2 would ruin the picture, whereas the wider field of the 5D might have captured the entire wing.

I did not save the pictures after this series of tests. Anyone with this super-telephoto and both bodies could try to duplicate the test. I think my auto-focus results, specifically at f4 on a super-telephoto, are quite interesting.

Don, I also have all 3 pieces of equipment you list and I mostly agree with you. Initial focus acquisition between the two bodies is very close, and the 5D really isn't bad for AI servo shooting. I think where the 1 series beats the 5D in this area is the sheer number of focus points. With the 5D it is easier for the subject to slip between focus points and as you noticed it does take a bit longer for the 5D to recover. For that reason I do prefer the 1 series for AI servo stuff, but you are correct that the 5D is quite good. I know you said you prefer to use 1 shot and time your shots, which is a perfectly acceptable approach but the higher frame rate of the 1 series is just another tool in the tool box that can come in handy. An example might be trying to get a certain wing position of a bird in flight, 8.5 fps could make your life a lot easier in that situation. The ability to use a 2x TC and maintain AF along with weathersealing are also nice features for nature/wildlife photographers. I do think we would both agree that both are excellent and highly capable cameras that produce fantastic results. Sure is nice to have both, isn't it?

In all of my Yellowstone shooting, I kept the 1D2 in 8.5 fps mode. That camera is so responsive, that I almost always shot two frames when trying for one. I couldn't get my finger off fast enough. But in Bosque, I was shooting so incredibly much, a shot every 10 seconds, all day long, that I was filling up all of my cards. So I shifted to single frame there.

I tend to prefer center focus point, since one time in Thailand, I had all points enabled, and the camera focused on a beautiful young woman's hat brim, and not further back on her face. That was a shot that couldn't be reshot, so I am more sensitive, in fast grab shots, about where the camera is choosing to auto-focus.

Fair enough. The 8.5 fps does take some getting used to if you want to fire off single frames, I'm kinda conservative in my shooting, a product of the film age I suppose. I use the single point when I'm not tracking something, actually I'm becoming quite fond of manual focus since I put a focusing screen in my 5D, the 1DMKII may get one as well. Then they will be perfectly equal in regards to focus speed

I'm becoming quite fond of manual focus since I put a focusing screen in my 5D

I love my Ee-s screen for use with my Zeiss and Zuiko manual lenses.

5D is decent AF, good one shot AF, decent prediction in AI Servo with central point.

still, in dayly use i find 5D inferior on moving subjects..
first is because of AF points number and positioning. big frame, 9 points on the very center of the frame...taking an in focus subject picture might not be a problem, but when u try to think a little about composition u find that u often need to crop...
subject in focus but crappy frame. for example shooting from the side a running athlete... his face right side of the frame, u want his position in the frame to be in the left half of the frame, little "thirds rule" or something....

also another thing is the FPS.. i shoot dogs as part of my job.... often i find 3 FPS are too little for very fast moving subjects...5-6 FPS is a big difference, 8 is even better...u may shoot 10-12 pictures and if bad timming u may not have at least one peak action...
this is not necessaru a matter of AF but of practical use of a camera...what good a decent AF in this situation if the camera is too slow...

another thing is speed of the shutter even in single shot mode... often, because i need my dog to be placed in a certain area of the frame...for example i want to take a high jump picture.... because AF points are so badly placed i focus on a point where i thing the dog will jump (decoy, a toy or something) keep the focus button pressed and take the shoot when the dog jumps...
if i try to "hunt" the animal this way i fell the lag between the moment i press the button and the moment when image is actually recorded...in fast moving subject YOU need to predict because camera have a lag...
i noticed a very serious difference between a faster camera like 20D...

a 6-8-10 fps camera IS faster in single shot mode, that lag is smaller because camera's mirror mechanism is made to perform faster...

and there are more things like this , things that make the camera and AF system not so great... 5D AF is good, for some circumstances is enough, for others no matter if teoretically could perform, in real world is not due to position of AF points, other mechanical parts of the camera and so on.....

AF ok, the hole tool that camera represent not ideal....

i love my 5D camera, i learned to exploit it and i know it;s limitations but for certain jobs 1D is just at a different level...

If you shoot birds in flight, you want to use predictive focusing and that doesn't work unless you shoot a burst of frames in servo mode. You shoot single shot and you do not get predictive focusing. The predictive focus doesn't kick in until the second frame of the burst.

That is the place where the 1D series shines, so unless you shoot a burst, I don't think the flight comparisons are that meaningful.

- Gene

I found for slower gliders like pelicans and such, the 5D is perfectly fine, but once the action heats up and there's any erratic behaviour or the speed's much higher the 1D II keeper rate is far superior to my 20D and 5D. And the responsiveness of the 1D II is an enormous benefit to me at least.

I agree with what gene is saying to, essentially Don you are using a a series of one shots rather than tracking. Try a to track a bird moving somewhat erratically and use high speed drive and keep the AI servo active throughout the entire sequence.

I shot the first half of the first day in AI servo at 8.5 fps. It was only when my cards were filling up alarmingly fast, that I switched to single shot. I could have switched to some filled cards, but they contained some dangerously close grizzly pics that were far more valuable to me that anything I could have shot at Bosque. The griz pics were loaded on my computer, but I insisted on keeping them on the card, too.

Wild Yellowstone grizzly, full frame, no cropping, no extender, 60 feet away

60 ft away and I'm pissing myself.

One-Shot mode is certainly not the way to highlight differences between a 1-series camera and any other, unless you are shooting at f/4. Most cameras are as good as each other in One Shot with only some variation in AF computing speed.

In the Canon world the 5D centre AF is the next best thing to a 1-series in AI Servo mode because of the assist AF sensors that help with tracking - if you bother to enable them. Many people use a single AF sensor without any expansion and they either miss a lot of shots or they are exceptionally gifted at precise aiming on a moving target. I can see why you'd do it with a 9-point system but not with a 15- or 45-point AF system.

Another 1-series advantage apart from the speed of AF processing is the high AF sensor density that allows a selectable amount of AF sensor expansion (to 1, 7, 13 or 45 points) around any sensor, not just the centre one, without necessarily having to select full auto AF selection as on a 9-point system. That gives a useful compromise between AF selectivity and AF tracking while allowing greater scope for selecting an AF sensor that suits your chosen composition. The difference between it and the 5D (or others) becomes greater once you are not using the centre AF sensor.

I'm a little surprised that the centre AF sensor of the 1D2 did not offer better focus accuracy than the centre AF sensor of the 5D with the f/4 lens. Perhaps the edge(s) that the 1D2 was focusing on were not horizontal with respect to the camera, as it is only in that direction that the 1D2 AF sensor offers greater focus precision. Alternatively, if the shooting aperture was sufficiently small then the extra DOF would obscure any focus precision differences.


By the way, with the 1D2 you can slow down the maximum frame rate a little bit (e.g. to 8 or 7 or 6 fps) if you want or need a compromise between shooting too much when you don't want to and shooting fast when you do. One of the personal functions controls it. One day Canon will discover and implement a great feature that allows quick and easy deletion of the last burst of shots, with a custom setting that lets you specify how many of that burst will be kept (e.g. none, first or last one, first or last two, etc.). Sports shooters would love that.

- Alan

monochrome wrote:
60 ft away and I'm pissing myself.

wimp

Alan321 wrote:

By the way, with the 1D2 you can slow down the maximum frame rate a little bit (e.g. to 8 or 7 or 6 fps) if you want or need a compromise between shooting too much when you don't want to and shooting fast when you do. One of the personal functions controls it. One day Canon will discover and implement a great feature that allows quick and easy deletion of the last burst of shots, with a custom setting that lets you specify how many of that burst will be kept (e.g. none, first or last one, first or last two, etc.). Sports shooters would love that.

- Alan

Or speed up the minimum continous frame rate from 3 to say 5.

Yep, that'll work too, but the slow speed cannot be increased by much. 5fps may be its upper limit. If you only want a slight decrease in speed from 8.5 fps then you will have to adjust the high speed setting rather than the slow speed. At least we can do it on a 1-series; the 5D, 40D , etc. cannot be tweaked like this. That's part of what we get for all of those extra thousands of dollars

- Alan

GeneO wrote:
If you shoot birds in flight, you want to use predictive focusing and that doesn't work unless you shoot a burst of frames in servo mode. You shoot single shot and you do not get predictive focusing. The predictive focus doesn't kick in until the second frame of the burst.

That is the place where the 1D series shines, so unless you shoot a burst, I don't think the flight comparisons are that meaningful.

- Gene

You can always use back * button to focus without tripping the shutter. That way you can take only 1 shot and that first shot in focus while in AI servo mode.

I have C. Fn. 4 -3 and with that I use * button to track (center point only) and when I have it tracking and nice in frame, I pull trigger with shutter button. I will take couple of shots, sometime 1 or 2 (mostly 2) even with 8.5fps.

I also keep home button (next to left) set to 45 points for BIFs against clear sky.

My thoughts are that AF in AI servo on my 1dmk2 works much much better than my 30d when using 500mm f4.