50D and AI-servo

Hey guys,
A while ago I asked about 50D AI-servo tracking performance compared to 1D MKII N and 40D and the response was that the 50D was better than both so I got a 50D to use with my 400 f/5.6. Since then I have done some BIF field tests, I always use the center AF point only, trying to keep it on the bird and pumping the shutter button whenever there is foliage to force the camera refocus on the bird. Honestly I did not see an increase in keeper ratio compared to the 40D:( for about 1000 shots I took. There are still shots where focus is fully blown and also the ones that focus is slightly back or front,
First I tried a stationary subject to verify there is no consistent front/back focus with the lens (which has been good on my old 40D and older 20D)
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/1.jpg

here are OOF/soft focus examples, you can see where the AF point was
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/6.jpg
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/8.jpg
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/2%20.jpg
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/3.jpg
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/4.jpg

the 2nd issue is that it looks like the 400 f/5.6 is no longer capable of providing the pixel resolution needed for 50D's dense sensor, even the shots that seem to in perfect focus lack the per-pixel crispness that I used to get with my 40D (of course when it was in focus:D) here are two examples

http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/7.jpg
http://www.stanford.edu/~ahazeghi/Photos/birds/50D%20AF%20test/5.jpg

So given that most of the samples that are posted here are small and don't tell how sharp the 100% is, I was wondering if what I am getting is normal and if so is the 50D really better than 1D MKII N in terms of AI-servo :confused:?
Maybe it performs better with the 500 f/4 that most of you guys have (I am jealous:p) but many still have high regards for 400 f/5.6 in sharpness department.

Thanks!

P.S. for comparisson here is a shot I got from my D700 and 300 f/4 + 1.4X TC @ f/6.3 (almost wide open) at ISO 500 so you can see what I mean by crispness (100% crop below)

http://www.stanford.edu/~ahazeghi/Ph...ead%20crop.jpg

Not sure what to say Arash. I don't have focus problems with my 50D and any of my lenses including the 400 f/5.6 and the 500 f/4. The 50D has much better servo AF than my 40D did IMO. Perhaps you could tell us what kind of camera settings you use for flight (type of AF, shutter speed targets, etc). Is this a new lens for you? If not, I assume it gives you sharp flight images with your 40D.

Arash, I have a 50D and find my results similar to yours with similar types of image. Here is my rationalisation of the issues....

The 1D2n has a resolution of 3504x2336 pixels and a sensor size of 28.7x19.1mm. That equates to a linear pixel density of 3504/28.7 = 122 pixels per mm.

The 50D has a resolution of 4752x3168 pixels and a sensor size of 22.3x14.9mm. That equates to a linear pixel density of 4752/22.3 = 213 pixels per mm.

With those figures in mind, when you compare a file from each camera at 100% you are basically looking at the 50D image with a magnification factor of 213/122 = 1.74X greater. To put that another way, if you were to crop the image from a 1D2n to match the physical sensor dimensions of the 50D sensor you would end up with an image of 3504/28.7*22.3 = 2723 pixels on the long side. Viewed on a monitor with 96DPI resolution you would end up with a virtual image of just 2723/96 = 28" on the longest side from the 1D2n. A 50D image would yield 4752/96 = 49.5". (49.5/28 = 1.75) That is a huge enlargement.

Whichever way you look at it, studying a 50D image at 100% on a monitor from a few inches away is asking a lot of the camera, the lens, the photographer, the light and the subject. Magnification levels of that order are going to make a significant difference when it comes to judging lens sharpness, focus accuracy, camera shake, subject blur and DOF.

Think about the guideline for shutter speed when hand holding - for 35mm frame size it's 1/focal length which is 1/400 for the 400mm. On a 1.6X crop body the guieline is modified to 1/(400x1.6) = 1/640. But that guideline is, I think, suitable for viewing a "normal" print size of maybe 10"x8". So if viewing the image at a virtual size of around 50" on a computer monitor you'd actually want a shutter speed of 1/(640x5) = 1/3200. 1/1600 is pushing your luck, I think, for camera shake and subject movement.

I don't know what the 1D2n is like for per pixel noise but I do know that even at 400 ISO the pixel level noise from the 50D can be significant. Certainly it is more obvious than the 40D. Whether you tackle the noise issue with software or leave it, the image will be robbed of some detail.

As well as the 50D I also have a 30D, 40D and 1D3. In my experience, to gain value from the extra pixel density the 50D offers, you really do need all your ducks in a row (light, focus, panning etc.). I only have a 100-400 lens and I frequently find myself wishing for greater reach, speed and sharpness. I too find myself disappointed with many of my results from the 50D due to the exact issues that your sample images demonstrate. Where the 50D does shine, and completely outclasses the other cameras, is in situations where you are lacking reach and you do have good light and the opportunity to nail focus with zero camera shake and zero subject movement.

That said, when viewing images at equivalent final sizes at modest reproduction sizes, rather than pixel peeping, then there probably isn't that much to choose between them.

As for lens sharpness and the 50D, Even with my 100-400 + Kenko 1.4X I gain significant benefit from the 50D vs my 40D and 1D3 in good light and with a static target, tripod, MLU etc.. If in doubt, try a moon shot with each body and see which camera yields the greatest detail. In my own tests my 50D absolutely blows away my 40D and 1D3. The results are, unsurprisingly, very similar when shooting a static target (CD) in good light at around 15m in my back garden. The 50D trounces my other cameras for pulling out detail. Fine printed text that I can read in the 50D file is just mush with the other cameras.

But when things are moving fast (BIF) I find it very hard to gain meaningful benefit from the 50D's extra pixels. There are just too many factors (noise, shake, blur, imperfect focus) that conspire to negate their value. If you're not pixel peeping and are filling the frame and reproducing at more conventional image sizes then I think the trade off tend to balance each other out. As the per pixel detail gets squashed - e.g. fitting each image into, say 900x600 for online viewing - so does the per pixel softness and the per pixel noise. In that sense the 50D is surely no worse than the 40D at delivering fine looking images. Just don't expect to crop massively unless you are incredibly skilled/lucky and have great light.

Of course, that's just my experience and opinion. Somebody who is actually good at bird photography, unlike me, might well have other ideas.

I have the 40D and the 1DMKII (original, just before the 1DMKII N)... I also own the 500mm f/4, with my equipment the MKII auto focus is fater than the 40D with the same settings.
But noise performance the 40D is much better, most of my attempts to BIF are with the 40D, unless there is plenty light.

My .02, Raul

Quote:

---

Originally Posted by Doug Brown

Not sure what to say Arash. I don't have focus problems with my 50D and any of my lenses including the 400 f/5.6 and the 500 f/4. The 50D has much better servo AF than my 40D did IMO. Perhaps you could tell us what kind of camera settings you use for flight (type of AF, shutter speed targets, etc). Is this a new lens for you? If not, I assume it gives you sharp flight images with your 40D.

---

Hi Doug,
If you are getting good results with 50D and 400 f/5.6 then it must be good enough. I use the center point only, you can see the shutter speeds in the examples above, most are above 1/1600 sec Do you use a tripod or handhold the 400 f/5.6? I have had the 400 for three years now, it did give sharp in flight images with the 40D but AF on 40D was a hit and miss. Can I ask what is the usual keeper ratio you get from 50D and 400 f/5.6? Last week I took about 1000 shots and only about 15% were critically sharp.

Thanks

Tim, I agree with you completely, the pixel pitch on 50D is very tight and thus it hits the diffraction limit sooner than the 1D series or the 40D, so I had expected pixel level sharpness to reduce when moving from 40D and I was also aware of increased noise especially in the shadows, it definitely needs NR. So yes IMO noise, diffraction, aberrations from the gap-less micro lens array due to small pixel pitch all in less than ideal light conditions and moving subjects kill the so called "reach" of the 50D in practice unless you are on tripod at f/8 and ISO 200 with perfect light, basically when you are using the combo as a telescope.

I also agree that by the time you down-sample the 50D image to 1500 pixels or lower the imperfections will vanish but then what was the point of 15 mpixels to start with, if at the end of the day you will be reducing it to 1500 pixels. The main reason I got this camera was not the "reach" but the AF, I was somewhat frustrated with my 40D AF (IQ and reach was good enough for me but not the AF) and I wanted a camera with better AF. So I inquired about AF performance between 50D and 1D MKII (N) and the general consensus was 50D was better so I got it. I also have some Nikon gear and IMO AF on the D700 with 300 f/4 and 1.4XTC was a head and shoulder above the XXDs, but I can't afford a Nikon 500 f/4 now which is awfully expensive.

So, I don't know what to do, keep the 50D or give it up and sell my lens and get something like a 200-400 f/4 VR. Looks with current Canon system anything less than a "good" 1DMKIII is not perfect in AF for flight shots.


Here's one of the above samples downsized for web posting, it looks good but the original is not so great
http://www.stanford.edu/~ahazeghi/Ph...9236812-lg.jpg

Quote:

---

Originally Posted by Tim Dodd

Arash, I have a 50D and find my results similar to yours with similar types of image. Here is my rationalisation of the issues....

The 1D2n has a resolution of 3504x2336 pixels and a sensor size of 28.7x19.1mm. That equates to a linear pixel density of 3504/28.7 = 122 pixels per mm.

The 50D has a resolution of 4752x3168 pixels and a sensor size of 22.3x14.9mm. That equates to a linear pixel density of 4752/22.3 = 213 pixels per mm.

With those figures in mind, when you compare a file from each camera at 100% you are basically looking at the 50D image with a magnification factor of 213/122 = 1.74X greater. To put that another way, if you were to crop the image from a 1D2n to match the physical sensor dimensions of the 50D sensor you would end up with an image of 3504/28.7*22.3 = 2723 pixels on the long side. Viewed on a monitor with 96DPI resolution you would end up with a virtual image of just 2723/96 = 28" on the longest side from the 1D2n. A 50D image would yield 4752/96 = 49.5". (49.5/28 = 1.75) That is a huge enlargement.

Whichever way you look at it, studying a 50D image at 100% on a monitor from a few inches away is asking a lot of the camera, the lens, the photographer, the light and the subject. Magnification levels of that order are going to make a significant difference when it comes to judging lens sharpness, focus accuracy, camera shake, subject blur and DOF.

Think about the guideline for shutter speed when hand holding - for 35mm frame size it's 1/focal length which is 1/400 for the 400mm. On a 1.6X crop body the guieline is modified to 1/(400x1.6) = 1/640. But that guideline is, I think, suitable for viewing a "normal" print size of maybe 10"x8". So if viewing the image at a virtual size of around 50" on a computer monitor you'd actually want a shutter speed of 1/(640x5) = 1/3200. 1/1600 is pushing your luck, I think, for camera shake and subject movement.

I don't know what the 1D2n is like for per pixel noise but I do know that even at 400 ISO the pixel level noise from the 50D can be significant. Certainly it is more obvious than the 40D. Whether you tackle the noise issue with software or leave it, the image will be robbed of some detail.

As well as the 50D I also have a 30D, 40D and 1D3. In my experience, to gain value from the extra pixel density the 50D offers, you really do need all your ducks in a row (light, focus, panning etc.). I only have a 100-400 lens and I frequently find myself wishing for greater reach, speed and sharpness. I too find myself disappointed with many of my results from the 50D due to the exact issues that your sample images demonstrate. Where the 50D does shine, and completely outclasses the other cameras, is in situations where you are lacking reach and you do have good light and the opportunity to nail focus with zero camera shake and zero subject movement.

That said, when viewing images at equivalent final sizes at modest reproduction sizes, rather than pixel peeping, then there probably isn't that much to choose between them.

As for lens sharpness and the 50D, Even with my 100-400 + Kenko 1.4X I gain significant benefit from the 50D vs my 40D and 1D3 in good light and with a static target, tripod, MLU etc.. If in doubt, try a moon shot with each body and see which camera yields the greatest detail. In my own tests my 50D absolutely blows away my 40D and 1D3. The results are, unsurprisingly, very similar when shooting a static target (CD) in good light at around 15m in my back garden. The 50D trounces my other cameras for pulling out detail. Fine printed text that I can read in the 50D file is just mush with the other cameras.

But when things are moving fast (BIF) I find it very hard to gain meaningful benefit from the 50D's extra pixels. There are just too many factors (noise, shake, blur, imperfect focus) that conspire to negate their value. If you're not pixel peeping and are filling the frame and reproducing at more conventional image sizes then I think the trade off tend to balance each other out. As the per pixel detail gets squashed - e.g. fitting each image into, say 900x600 for online viewing - so does the per pixel softness and the per pixel noise. In that sense the 50D is surely no worse than the 40D at delivering fine looking images. Just don't expect to crop massively unless you are incredibly skilled/lucky and have great light.

Of course, that's just my experience and opinion. Somebody who is actually good at bird photography, unlike me, might well have other ideas.

---

Quote:

---

Originally Posted by arash_hazeghi

I also agree that by the time you down-sample the 50D image to 1500 pixels or lower the imperfections will vanish but then what was the point of 15 mpixels to start with, if at the end of the day you will be reducing it to 1500 pixels.

---

I think there are probably three main justifications for 15MP in the 50D....

1. Marketing;
2. Marketing;
3. There are photographic disciplines other than birds and BIF and it is possible to shoot with controlled lighting with incredibly sharp glass, and patience, and benefit from, not only the pixel density, but also the focusing aids such as AF microadjustment, high resolution display, high resolution preview image, very accurate Live AF, which might be slow but is very precise. Think macro, studio work, art reproduction, landscapes, architectural etc..

I do like to understand my equipment as much as I can, in order to know which tool to use and how to get the most from it. With that in mind I thought I'd reshoot my comparisons at differing ISOs with my 40D, 50D and 1D3. There is no doubt in my mind that under the right conditions those extra pixels are not wasted. Here is the test scenario....

I set up a static target consisting of a CD and some little furry animals to give some detail that could be used for accurate focusing and also to judge the level of detail resolved and image sharpness. The cameras were set on a Manfrotto 055MF3 tripod and 468MGRC2 head with only the first set of leg extensions extended. The camera was focused using Live View and (very very careful) manual focus to make the writing on the CD as sharp/contrasty as possible. The lens was my 85/1.8 set to f/5.6 to obtain a high degree of optical sharpness, hide minor focus error within the DOF and ensure diffraction was not a risk to the results. The camera was tethered and focusing assessment and shutter release were performed from the laptop. The subject was set ~5m from the sensor (measured roughly). This gave a reasonably sized subject, within the frame, to represent small birds shot at some distance. e.g. for a 400mm lens this scales to a distance of 23.5m.

Images were shot raw and processed through DPP with standard picture style and sharpening = 3. NR was left at the auto defaults that DPP now chooses for various bodies and ISO settings.

Here is the overall scene before the cameras, as captured by the 1D3 at 100 ISO....

http://lh3.ggpht.com/_4_R8TkwT74w/Si..._Full_0001.JPG

Here is the scene as captured by the 50D at 3200 ISO....

http://lh4.ggpht.com/_4_R8TkwT74w/Si..._Full_0006.JPG

The different field of view is obvious and there is a change in colour tone (maybe the light changed a bit) but in terms of sharpness/detail/noise I'm struggling to see a massive difference between these two extremes of the IQ scale. Full frame images, resized to 800x533 can be found for other ISOs and cameras at the online album here - http://picasaweb.google.com/EezyTige...TestFullFrames. To be honest there isn't anything very exciting to see, since the massive downsizing obliterates the subtle differences, but I included all the images for completeness.

I'm always cautious about the value of 100% crops. After all, we're here to create images, not pixels, but in this test I believe the difference in sensor performance is best demonstrated by viewing at 100%. Here are three 100% crops at 100 ISO from the 1D3, 40D and 50D.

1D3 100% crop....
http://lh3.ggpht.com/_4_R8TkwT74w/Si...op100_0001.JPG

40D 100% crop....
http://lh3.ggpht.com/_4_R8TkwT74w/Si...op100_0001.JPG

50D 100% crop....
http://lh5.ggpht.com/_4_R8TkwT74w/Si...op100_0001.JPG

This is where the 50D shines. It may not be easy to read all the programme information on the CD but it is just about possible. Certainly the programme name and episode name can be read. The finer print below is a real challenge. The 40D allows the programme name to be read with reasonable ease, but not the episode name. With the 1D3 image I think it is fair to say that the programme names cannot be distinguished at all. If you look at the muzzle detail on the monkey and the fur on its right arm there is detail visible there from the 50D that does not come through with the other cameras.

I have uploaded all the 100% crops at all ISOs to an album here - http://picasaweb.google.com/EezyTige...onTest100Crops. EXIF is in the images and details can be viewed by clicking onto the thumbnail to show an image and then clicking "more info" over on the right of the screen.

Finally, since we should be more concerned about our subject/scene/image than individual pixels I have also created an album where I have cropped just to include all the subjects, quite ignoring how many pixels that picked up, and then resized that crop to 800x533 in all cases. For each image this did result in downsizing to some degree, but very little for the 1D3. This is probably closest to a real world type scenario when having to crop aggressively and figuring out which camera will give you the best result.

I shall have to make a new post to include "crop to subject" sample images from the three cameras, as I've hit my 5 image limit in this post.

Here are the 100 ISO "crop to subject" examples....

1D3 crop to subject, 100 ISO....
http://lh5.ggpht.com/_4_R8TkwT74w/Si..._Crop_0001.JPG

40D crop to subject, 100 ISO....
http://lh4.ggpht.com/_4_R8TkwT74w/Si..._Crop_0001.JPG

50D crop to subject, 100 ISO....
http://lh5.ggpht.com/_4_R8TkwT74w/Si..._Crop_0001.JPG

To my eyes the only example here that is usably sharp/detailed is that from the 50D. This image has not been edited/finessed so maybe some more appeal can be added, but I think compared to the other results the 50D image is the only one worth pursuing. The album with all "crop to subject" images is here - http://picasaweb.google.com/EezyTige...tCropToSubject.

So, when all is said and done, I think there is a purpose to the 50D's extra pixels, and when it is possible to focus precisely and eliminate camera shake and subject blur the benefit is there to be had. However, if focus is missed just a little, or there is any smearing of the image on the sensor during capture, due to movement somewhere, then the pixels seem to have no benefit. Thus, for perched and walking/wading/nesting birds and a tripod mounted camera/lens I think the benefits can be realised. For BIF, unless accompanied by great skill and a little luck, I'm not sure the benefits can be realised fully if at all.

Putting the sensor to one side and returning to AI Servo performance, which was the original topic for this thread, I have no doubt that the 50D is every bit as capable as the 40D, if not more so, but the sensor may more readily reveal shortcomings in technique or fortune that are not exposed by the 40D or especially the 1D2n.

As for AF performance of the 50D vs the 1D3 (I don't have and never have had a 1D2n), I'll take the 1D3 thank you.

Hi Tim,
Thanks for the nice comparisons, I agree that under controlled conditions in bright light and low ISO, with a stopped down 85mm 50D can pull more detail from a particular scene from a given distance. However many of the flight shots or even stationary birds are in less than ideal light conditions with shadows that I need to lift, I usually set my camera to ISO 400-800 because my lens is f/5.6 and now with high pixel density looks like I need to stop down to f/7.1 or beyond to get the kind of sharpness that I like, so for me sensor has no advantage in practice. I wish Canon had gone for better AF as opposed to megapixels. 40D AF is not bad but I wanted something better. I wish there was a way to systematically test and compare AF between cameras, I usually go by keeper ratio for in flight and landing shots and I didn't see any difference between 40D and 50D but it might be my technique or hand-holding. I have never had the pleasure of trying 1DMKIII but I have little doubt that it is better. I was initially looking at getting a used MKIII but with all the AF fiasco I changed my mind so MKII N and 50D were my only choices.
BTW, with the 50 why are you still keeping the 40D?

Quote:

---

Originally Posted by arash_hazeghi

I wish Canon had gone for better AF as opposed to megapixels. 40D AF is not bad but I wanted something better.

---

So did I. All I needed was a 10MP sensor with pro AF, a light handed AA filter and terrific high ISO performance. When the 50D was first announced, based on the spec sheet, I was dead set against it , but as time passed and end user reports came in I took a closer look and decided to buy one after all.

The reason I bought one was not for birding especially, but for photography in general and weddings in particular. I had been shooting with a 30D and 40D and the differing button layout was driving me nuts. I bought the 50D as an upgrade/replacement for my 30D and to use alongside my 40D, with the 30D relegated to the spares cupboard. I could have bought a second 40D but I thought that if I was buying anyway I may as well go for the latest and greatest and at least get some new features/performance rather than simply an identical model. I do favour a two body setup whenever I will need more than one lens and I find the xxD cameras ideal for travel/vacation photography due to the compact size and light weight of the bodies and lenses compared to the heavyweight 1 series and EF glass.

I do find it somewhat ironic that the camera in Canon's lineup with the best AF system (IMHO the 1D3) has the sensor with the lowest pixel density and thus can benefit least from fast, precision focusing. The camera that could benefit most from terrific AF - the 50D - has a consumer grade AF system. It might be pretty darn good, but its flexibility is somewhat limited and there is a mighty chasm between focus points when it comes to tracking a difficult/small subject.

If Canon had put pro spec AF into the 50D, and kept the pixel count at 10MP, with gapless microlenses and DIGIC 4 for great noise performance, I would not now be the owner of a 1D3, or a 16-35 or a 24-70 - absolutely no need for that kind of camera or that kind of expenditure, for my photography. So because the 50D isn't everything I wanted it to be in terms of AF and noise I've spent an extra ~£4,000 on Canon gear? Who is the fool - Canon or me?

Quote:

---

Originally Posted by arash_hazeghi

BTW, with the 50 why are you still keeping the 40D?

---

Because for travel purposes I will be taking my 40D and 50D with appropriate lenses - typically 10-22, 17-55 and 100-400. There is no room in my bag for the 1D3 and I don't want the weight of that thing plus my 16-35 and 24-70 to drag through airports and carry with me all day long. If I took the 50D with EF-S glass and the 1D3 and my 50D went down then the EF-S glass would be no use. If I took all EF glass and the 1D3 went down then I'd lose a lot of my wide angle capability and the benefit of IS in my standard zoom. For travelling light and compact the XXD cameras make more sense to me. Also, if the 1D3 and 50D were my only cameras, what would be my spare?

A parting example of a BIF shot with my 50D - This won't win any awards so please don't bother to critique, but judge whether the AF was up to the job and whether the pixel density of the 50D brought any value to the detail and sharpness of the image.

Full image - 50D + 85/1.8 at 85mm, 1/2500, f/4, 200 ISO. Sharpened in Lightroom....

http://lh5.ggpht.com/_4_R8TkwT74w/Si...47_5657_LR.jpg

100% crop....

http://lh5.ggpht.com/_4_R8TkwT74w/Si...47_5657_LR.jpg

It's not stellar in IQ, but I suspect it gives me more to work with than a 1D3 or 1D2n might.

At the end of the day everything is a trade off - price, performance, features, durability, size, weight, ergonomics etc.. Due to the laws of physics and economics we can't have everything our own way. All we can do is pick the best tool for the job for any given budget. For some that may be the 50D, for others the 40D, for others the 1D3 or 1Ds3 etc.. Heck, for some it may not even be a Canon at all. Remember, Canon did not target the 50D exclusively at BIF photographers. If the 50D is not delivering the result you seek, is it the right tool for you? The thing is, is there actually a tool that would do better? My own conclusion is that the 50D is a fine camera, but it cannot be blamed if I insist on trying to photograph BIF with a 400mm zoom lens from 100m away and then am unhappy with the results viewed at 100%. What camera would yield pleasing results under those circumstances?

As things stand, in my opinion the 50D is the best camera I have when things are "too far away". If the light is poor or I have no trouble filling the frame with my subject, or there are high speed movements to be tracked then it'll be the 1D3 for me. As a general walkabout camera, when shooting casually, I'm happy to take either the 50D or 1D3. My choice will be based on the weight and bulk of the system, relative to what I'm doing and where I'll be, and the expectations I have for possible subjects, their distances from me, and the lighting.

p.s. Don't forget that the farther away you are from your subject the more atmosphere you have in the way. On a hot summer day there can easily be a heat haze/shimmer and the farther you are from your subect the more severe the effects. What the naked eye doesn't see, or disregards, the camera surely picks up at large magnification levels. e.g. this example of a 50% crop from a photo at 380mm, 1/1600, f/8, 400 ISO....

http://lh6.ggpht.com/_4_R8TkwT74w/Si...49_6957_LR.jpg

Get closer and things begin to clear up nicely. Here are a couple of 50% crops from a sequence, shot at 235mm, 1/2000, f/7.1, 400 ISO....

http://lh4.ggpht.com/_4_R8TkwT74w/Si...48_7231_LR.jpg

http://lh5.ggpht.com/_4_R8TkwT74w/Si...51_7243_LR.jpg

I have the same problem recently, a lot of shots make with the 40D was out of focus and or the lazy AF catch a more easy place to grab (in front or in the back). Even if I better like the image resolution and quality from the 40D, I go back to the 1D2n for a more solid AF. it's Better a lower resolution than no image, especialy in nature photography where you wait so long until the opportunity happen.

I have found the autofocus on my 50D to be dead-on accurate when subjects are relatively close and I've also found that it consistently focuses closer than the subject for farther away subjects. With BIF, you often have nothing but sky around the bird and you would only notice that the bird isn't tack-sharp ( you wouldn't be able to tell if it focused closer). It's noticeable for the landscape shots or birds on the ground when the focusing is too close. I haven't yet figured out how to deal with this for action shots, but manually focusing using LiveView 10x produces wonderful results.

Eric, I guess you could try a different AF microadjustment setting if you know you will be shooting at some distance. e.g. maybe for subjects <20m away you might need +5 and perhaps for subjects >20m away you might need +8 or +10, for example.

Canon does recommend that you set your adjustment "in the field" under the conditions of actual use (lighting, temperature, distance, focal length), rather than just to suit some "laboratory conditions" set up for the initial AF microadjustment. This is mentioned on page 181 of the 50D instruction manual, albeit as an unobtrusive bullet point. I must admit that I need to look into that myself too, as I do try to calibrate somewhere around the 50X focal length mark, which would be 20m for my 100-400.

The odd thing is that in some of my most disappointing photos at longer range - perhaps 50m-100m I can't find any sharp point in the scene. Everything just looks dreadful. It does not seem to be an issue of missed focus, or camera shake or subject blur. At other times there are no problems at all. I have no problem with shots of the moon, for example, so it isn't a problem with longer distances all the time. So, either my lens has something loose inside or there is some sort of atmospheric condition that is causing loss of IQ.

Here's an example from the sixth frame of a sequence of eight shots spread over three seconds. This is a raw file with sharpening at Lightroom defaults. The full frame looks OK and you would not notice a problem when looking through the viewfinder, but, viewed at 100% the resulting image is clearly junk. They all were. It was shot at 400 ISO (not too demanding), f/7.1 (should conceal some of the 100-400's softness at 400mm) and 1/1250 (which hopefully should put paid to camera shake). It looks like I am using a horrible, cheap filter, but I do not use filters, and the lens was clean and the hood in place.

http://lh3.ggpht.com/_4_R8TkwT74w/Si...50_7208_LR.jpg

100% crop :
http://lh3.ggpht.com/_4_R8TkwT74w/Si..._7208_LR-2.jpg

There is no reason for this picture to be anything but near perfect, unless I am missing something. I was locked onto the bird and tracking it without problems. If anyone has ideas I'd be delighted to hear. This does not look like shake, or poor tracking - there is no directionality to the "softness" - but it also doesn't look like missed focus either - and for all eight images to be similarly low in IQ.....? Like I said, there is nothing sharp within the scene. It looks like a really crappy lens, or your classic cheap filter problem.

On another day, with the same camera and lens, and no sharpening beyond Lightroom defaults, I've had this (400mm, 1/640, f/8, 400 ISO)...

http://lh5.ggpht.com/_4_R8TkwT74w/Si...47_6128_LR.jpg

100% crop :
http://lh6.ggpht.com/_4_R8TkwT74w/Si..._6128_LR-2.jpg

My moon shots turn out OK with this body/lens, even with a 1.4X, so it is not just a problem of distance alone. This is why I suspect that the problem is atmospheric rather than with the kit or my own (lack of) skills.

One issue that I have with 50D and it also existed with 40D and 20D is that in AI-servo mode camera is in release priority mode. you can test this by aiming your lens at a very close subject where it can't focus and pressing the shutter all way, it will gladly fire off! When tracking birds I have found the camera fires regardless of focus lock sometimes resulting in 17 OOF shots in a burst :( A better approach would have been to slow down the frame rate automatically so that AF could keep up, it's just a normal servo that sometimes overshoots or undershoots since it has a short rise time, camera however should wait for convergence before firing!

Thanks for posting these images. Definitely they've made me reconsider the idea of upgrading from my 400D to a 50D or 40D. I tend to suspect at this point that while people who have topnotch tracking skills can do BIFs with a 50D, people with less tracking ability are better off with a 1D series camera or a Nikon CAM3500 camera like your D700. This is just a guess...

Since posting in this thread I have had discussions on another forum and also been out practicing and testing some more. Here are some further findings....

- My AF is calibrated perfectly and the camera can focus accurately whether at 5m, 20m, 150m;
- For a static subject One Shot AF does seem to have the edge over AI Servo, including when using the back focus button to simulate One Shot focus when in AI Servo mode. Unless you compared the shots side by side, at 100%, you probably would not notice.
- Atmospheric disturbance is a genuine issue when shooting at longer distances and longer glass is not the solution. You need to be closer to your subject or pick a time when the air is cooler. In hot conditions, or with the air filled with dust or pollen, use long glass to make small things bigger, not distant things appear closer;
- In summer, (very) early starts are the best way to avoid atmospheric disturbance through heat and also provide low angle lighting that brightens the underside of the bird. Get out and shoot at dawn, not midday;
- Your focus tracking accuracy must be faultless. Even though each AF point is larger than the indicators suggest, you really do need to get the indicated point on the bird.
- Panning technique must be beyond reproach. You'll need that in order to keep the focus point where it belongs, but also to contribute towards the elimination of camera shake;
- Shutter release technique should also be beyond reproach. I find it easy enough to release smoothly when shooting static subjects, but for BIF everything seems to be moving so fast that I think I get careless. Perfect shutter release also helps reduce shake and is worth paying attention to. Firing a short burst of maybe 3-5 shots may help get a shake free image amongst them.

I think the 50D probably is up to the task, as far as the AI Servo performance is concerned, but it takes a lot more skill from the photographer than I first realised. I also think that to get stellar results also requires rather more thought, planning, preparation, knowledge, commitment etc. than I have applied to the task. I realise now that turning up to a wild bird reserve, where the birds are too far away, and it is impossible to get closer due to fencing or other barriers, and opening times restrict access to the least favourable times of day I am on a bit of a hiding to nothing. I think my own approach has been far too casual and if I want to improve my results I need to buck up my ideas as well as improve my technique.

I suspect also that the 100-400 zoom lens is not the best choice to make the 50D shine. The 1D3 probably makes things easier and the relatively low pixel density probably conceals the user errors far more readily than the 50D does. For now I'm sticking with the 50D as my birding/wildlife camera but I need to practice my skills and I will not be wasting my time shooting individual birds at 50m or 100m on a hot day, if ever.

I shall endeavour to get up and go out far earlier (before breakfast), to catch the best light and cool air, when things are still quiet, and I will practice, practice, practice. While I have no interest in shooting pigeons, crows or gulls, for example, I shall use them as practice targets so that hopefully I shall be prepared when something more exciting comes along. As I said before, use the 50D's high pixel density to reveal details in small subjects that are close by(<20m), not to make things that are far away (>50m) appear that much closer. When things are too far away, at or near ground level, the atmosphere will probably conspire to prevent you realising the benefits of all that the 50D has to offer.

Good for you Tim! I think there is a tendency among bird photographers to blame their gear when they can't get flight shots as good as what the pros get. Photographing BIF is hard and it takes lots of practice. I can assure you that the more you practice, the better your gear is going to perform. :)

Tim, You mentioned atmospheric disturbance at longer distances as a possible contributor to autofocus problems. I hadn't given that a thought previously (obviously heat waves can make the entire image appear soft, but that's not the problem I've been dealing with). Is it possible for autofocus to lock onto disturbance in the air at a point between myself and my subject? I've not heard of this before and it seems unlikely to me that something like this would give consistent front-focusing.

Sorry if I suggested that longer distances muck up the AF. What I'm experiencing is a scene where nothing is sharp, whether I focus manually, using traditional AF or Live View manual or Live AF to focus. I can get focus lock confirmation. I can manually move through the sharpest point of focus and back again but viewed at 100% the image is truly an abomination. I've had this not only with my 50D but also my 1D3 too. I do not think it is an AF issue at all. I'm quite sure the atmosphere is acting like the world's worst ever UV/protection filter and is basically a wobbly "lens" stuck in the way between my own lens and the subject.

I shot a short video of my Live View display when "focused" on a TV aerial about 150m away. The air temperature in the shade was not high - around 20/68 - but the aerial was above a roof in bright sunshine. The display can be seen to wobble and shimmy the whole time. I also try AFing a few times and that works OK. It's just that the actual image is rubbish. I didn't video an example at shorter distances but picking an aerial about 30m away and doing the same thing there is no visible problem.

Here is the video. I'm afraid I hand held the video camera (compact camera) but even so I think you can see the shape-shifting and morphing that appears on the Live View display....

http://www.youtube.com/watch?v=MsOSBy8P7eY

p.s. Cheers, Doug. I realise there is quite a gulf between my skills currently and those needed to turn in a cracking shot. At least I am now coming to understand the shortcomings in my own experience and techniques. Probably the biggest one, amongst many, is my failure to get anywhere near close enough to my subject. The other big one is panning skills, or lack of. I was thinking a 500/4 might be the fix I needed, but now I somehow doubt that very much. I need subject knowledge, fieldcraft, time, patience and dedication, not more kit.

Apologies for reviving this thread, but it does cross my mind that the pixel pitch of the D700 used for the hawk photo in the first post is almost twice that of the 50D -- 8.45 um versus 4.7um. Thus, to compare the 50D shots fairly to the D700 hawk shot to judge the cameras' focusing ability, the 50D shots need to be reduced by quite a bit -- as an approximation, shrunk 1:2 linearly. I did that with one of the out of focus avocet shots (don't want to post the resulting picture since it isn't mine) and spent some time staring at the hawk and the avocet side by side. I can't say for sure that the avocet looks more out of focus. What I can say is that the avocet seems farther away than the hawk, is a smaller bird, and being largely white doesn't present as good a focus target.

It's tough being a consumer thinking to buy this kind of expensive gear and trying to get a feeling for what bang you can expect to get for the buck, particularly when that bang depends on the development of your own skills as you use the gear. I don't know if there is really an answer applicable to all shooters to the basic question of which is better to buy for BIF, 50D or 1D Mark IIn.