Update:
We have now been able to replicate some autofocus issues. They may not be relevant for your type of shooting, however. Please click here for an ongoing detailed report.
Having read the reports of a few people indicating problems with the autofocus system the first thing I did was conduct some of my own tests. My tests did not indicate any significant issues in this area. I did think I noted a few instances of a delay in locking focus on a still subject. But, upon reviewing the actual image files I saw nothing consistent worth reporting. It all seemed well within the norms of the standard digital autofocus systems we all use. So, I did not spend a lot of time on the issue when writing up the review.
In fact, I was just about to hit the publish button when I saw Rob Galbraith's dire report on the performance of the Auto Focus system. With all due respect to Rob and Mike, my tests did not reveal the issues they seem to be experiencing. So, I have decided to include more detail about our experiences with the autofocus system. If indeed there are issues, perhaps this will add to the body of knowledge regarding the autofocus system.
Again, just for the record, I did not experience any focus issues that would make me hesitate to purchase this camera.
Perhaps Ron Holt will also provide his experience in the forum, given he shot 748 frames of a little league game on the 1D Mark III sandwiched in between two sessions on his trusted 1D MKII body.
Weather: First, some information about the weather, as this is central to Rob's report on the Autofocus system.
The following photographs were shot on June 15 in Atlanta, GA from mid afternoon to early evening. The temperature at the time was in the low 80s (Fahrenheit). Subjectively, it felt very hot and humid. It was clear with very partial cloud cover. By now, anyone who has ever been to Atlanta in the summer is just nodding their head wanting me to get on with it. They don't call it HOTlanta for nothing.
Camera Settings:
No custom functions relating to autofocus were changed for any of the following photos. In other words this is what we got straight out of the box in terms of autofocus.
Sequence of 15 consecutive images of a baseball player swinging and running down the first base line. Atlanta, GA. June 15th, Little League Baseball.
Images shot with Canon 300mm f/2.8 + 1.4x. Aperture 5.0, ISO 250, Shutter Speeds of 1/500 and 1/640. After clicking to view full-res image you may right click to download the full-res file complete with all exif information including settings, times, etc. All images straight from the camera with no processing whatsoever.
Photos courtesy of Ron Holt.
Here is the EXIF info for the last frame in the sequence.
File name
_B9P2401.JPG
Camera Model Name
Canon EOS-1D Mark III
Shooting Date/Time
6/15/2007 18:29:09
Tv(Shutter Speed)
1/640Sec.
Av(Aperture Value)
F5.0
Metering Modes
Centerweighted average metering
Exposure Compensation
0
ISO Speed
250
Lens
EF300mm f/2.8L IS USM +1.4x
Focal Length
420.0 mm
Image size
3888 x 2592
Image Quality
Fine
Flash
Off
White Balance
Auto
AF mode
AI Servo AF
Picture Style
Faithful
Parameters
Tone Curve : Standard
Sharpness level : -
Pattern Sharpness : -
Contrast : 0
Sharpness : 0
Color saturation : 0
Color tone : 0
Highlight tone priority : No
Color matrix
-
Color Space
Adobe RGB
File Size
3195 KB
Dust Delete Data
No
Drive Mode
High-speed continuous shooting
GPS Data
Latitude : -
Longitude : -
Altitude : -
Geographic coordinate system : -
Owner's Name
-
Camera Body No.
504439
Only 1 truly out of focus shot out of 15 consecutive movement filled frames. Perhaps this 1 OOF shot was due to movement on the part of the shooter. Just nothing worth being concerned with.
There are many other such sequences in the batch of 748 that were shot. In addition, action photos I shot the day before, on June 14th, show a similar high percentage of in focus shots. It was just not an issue for us.
Update: (June 21, 2007)
There seems to be a lot of interest in more information about this camera and the autofocus system. I may be able to review the autofocus system next week in more detail to try and replicate the issues Rob and others seem to be having.
Additional Gallery Images In the meantime, I have uploaded 56 more images from the baseball coverage into a gallery folder here for viewing. In addition, I have uploaded one short series of the catcher showing the most consecutive out of focus images I could find in the whole batch.
For those not familiar with our gallery system you may have it play a slideshow after clicking into one of the images and/or use the drop-downs to change the number of images viewable on one page. You may click on any thumbnail for a medium sized image, and then click again for the full size unprocessed image.
Highlights
Here are a few highlights from these additional images.
In the above two images, and the series they come from, it almost seems like the AF system should have been fooled but wasn't. It maintained the lock on the correct subject even with a more prominent image target in the frame.
This image is taken from the worst series of consecutive out of focus images I could find.
Weather
For those who doubt the weather conditions at the time of this coverage here is the link to the hourly weather report for June 15th in Alpharetta, GA (suburb just north of Atlanta). It shows 79 degrees at 7:00pm. I believe Rob reported having significant problems around 77 degrees. It should be obvious from the images that the sun was indeed out at the time.
I was outside in the 80 degree temp all afternoon with the camera body at all of the different parks (from about 2:30pm onwards). The shots on the backfocusing page were shot at the same park earlier in the afternoon. So, while the body was not in direct scorching sunlight the entire time, it was outside and operational in the heat for a long time. If I get the body back next week, we'll try it in the direct sunlight for a few hours in our now 90 degree weather. As far as heat. We're really just getting started. Wait until July and August. I could show you some nice images of past brides who were literally melting away because they didn't take the heat seriously and booked an outside wedding in the summer here.
Additional Note (June 23, 2007) : I may have access to the 1D MKIII again next week. If so, I will attempt to more directly replicate the problematic conditions being reported by some other websites. Hopefully, I will also be able to report on some additional features of the camera. If you have any specific requests please let me know in the forum thread.
Update:
We have now been able to replicate some autofocus issues. They may not be relevant for your type of shooting, however. Please click here for an ongoing detailed report.
This was a very useful review, Drew. In my opinion, this is how technical reviews ought to be carried out. It is worth far more to me to be able to read about real world usage rather than seeing the endless, and often sterile, identical test images that have been de rigeuer in the technical press for decades.
I think it is very worthwhile having access to the RAW images so that the potential of the camera can be evaluated while using our own workflows. It speaks far more to me than having to look at (and compensate for) another person's own image processing techniques. I look forward to more of this type of review. Thank you.
Thanks for all the effort you put into this review, Drew. I had just read Rob's review this afternoon, and I think I learned a lot from both of them.
I wasn't planning to purchase the camera at this time anyway but I'm keeping my up with this in case I decide to purchase one later on, after some of the focus questions that some have experienced are getting answered.
I appreciate that you put images that are completely unprocessed up for everyone to dudge the image quality at different ISO settings for themselves.
I've read both your AF review and the AF review by RG. He is not the only one reporting AF problems with the 1DIII. I've been jumping around on a lot of forums over the last month and the AF issue is far from consistent...I'd call it random at best. However, it has been reported often enough to cause me to postpone my current 1DIII order until the issue is resolved satisfactorily. It is entirely possible that the future resale value of these first releases will be impaired and I do not like the Canon dogma that..you bought it... you own it...you can't return it. And by inference...we don't know what's wrong with it, if anything, yet.
IMO your test does not exactly duplicate RG's, if I correctly interpret his conditions...bright clear mid-day skies on an 80+ day. You test appears to have been done at 6:30 in the evening with some clouds.
Is it possible that direct sunlight on the 1DIII black body under the clear mid-day sun is causing a heat build up that is the source of this phenomenon that is not repeatable with the waning sun rays at 6:30?
Scientific method would mandate EXACT conditions be repeated in every respect to refute the other test including exact framing. Your runner was slightly off the lens axis...his were directly perpendicular to the sensor. HIs test showed the principal point of focus...yours does not. To repeat his test you would have to be standing on the baseline looking directly at home plate with the baseline fixed in the viewfinder throughout the series. I don't recall that is the case. His test had a closeup of a stationary person with focus point indicated...I saw no such test of yours. His tests show the OOF focus shots graphically...yours do not. There may be other differences.
As I recall RG was given an unrestricted pre-production 1DIII with the proviso that he not publish the images and there is an AF issue with the pre-production models that will be corrected before mass production begins. If you have not read his pre-review you might find it of interest, as the same AF anomaly existed in both pre-production and mass production versions of the 1DIII.
RG is one of the more respected reviewers on the planet. If I recall correctly, he was invited to Tokyo by Canon. For him, or anyone with his reputation, to jeopardize the loss of that stature by posting a negative review of that magnitude takes a great deal of fortitude. His tests were not just checked and rechecked, and checked again, they were also shared with Canon.
One could surmise that Canon's usual "so what" attitude would be the only thing that prompted him to go public...to protect the buyers.
I applaud him.
I will give you the same accolades, once you EXACTLY and REPEATEDLY replicate his conditions with more than one camera regardless of your outcome.
Until then I hold your AF test to be suspect and unscientific.
Sorry, but there is no camera movement in the one out of focus shot. That is clear from the background fence, which is tack-sharp and without blur at all.
The shot is back-focused.
I suppose one out of so many is no big deal. In my day (the 60s) most shots were unsharp and one had to pick and choose and then, for commercial shots, retouch the whole photo to get anything really sharp.
I, personally, would have used one single sensor and always placed it on the boy's face. Not even on his helmet, where the sharpness usually is in this series. As good as these photos are, had the focus been exactly on his face, the photos would have been even better.
In these photos, the helmet has more plasticity than the face itself (see papers on absolute focuc at www.anstendig.org for explanations).
Mark B Anstendig
Last edited by mbanstendig; 06-20-2007 at 08:21 AM..
What seems to have happened is that the boy was next to the junction of the pipes holding up the fence, which had blue and green things behind it to the right. The blue and green things were similar to the boy's colors and the pipes of the fence were large enough to be seen by the sensors as important. So the sensors averaged the distances between those three things, settling in between, which was pretty much on the fence.
The auto-focus is working just fine. That is why I never use it with my EOS-1Ds MK II.
I only use single sensors, which is the only way to be sure you get the focus exactly where you want it with the image types that it can handle. Some image types cannot be exactly focused with this system or any other auto-focus system, except perhaps large format systems, with magnifiers, etc.
But, of course, I don't do sports.
However, sequences of photos I did live during the sixties, using the old lenses of the day and the Messraster focal-point-exact focusing screen in SLR cameras, are better focused than these, even without the depth of field (they were mostly wide open) because the focus is almost always exactly on the nearest eye of the main subject, and they were done in poor lighting. They are at www.anstendig.com at the URLs:
Great job, Drew. You'e done my kind of real-world review: Actually use it, and show the results.
Heh, one OOF shot, with rest nice and sharp. It's pretty clear what happened there... The bit of fence showing through between the boy's appendages tricked the camera into focusing on that. Nothing wrong there with the camera at all. Just keep the focus point on the boy's body and everything would be fine.
Regarding the noise at high ISO, I've been experimenting... It's pretty much standard Canon fare, though there's less of it than in prior models. I have a banding noise reduction tool I'm about to release, and I already have luminance and color noise reduction tools that work a treat. Between these two ISO 6400 Mark III images can be made into publication quality.
Linear Mode
