Analysis of EOS 7D AF performance, part 2

Printable View

10-07-2009, 01:35 PM
arash_hazeghi

Analysis of EOS 7D AF performance, part 2

Hi guys, I was not planning to post a follow up to the original thread, however, since yesterday I have received many emails requesting a follow up post regarding the performance of 7D AF system when tracking birds in flight, so I decided to post a second thread.

Since the raptor shoot last weekend, I have been able to speak with CPS and learn more about the features of the new AF system, according to them the 7D AF system is supposed to be a step up from the TTL-SIR module currently used in 1DMKIII and 1DSMKIII cameras. The main feature of this new AF module is that it is coupled to the RGB meter and uses color information from the 63-zone RGB CMOS to help track the subjects in AI-servo mode. No other EOS camera has this feature.
This morning, I met with another shooter who also had a 7D, the two bodies were purchased new from different vendors at different times and the serial numbers were completely different in their entire digits.

We went to Palo Alto Baylands nature preserve to do a shoot, a world-famous site for shore birds, I made about 500 photos. The conditions were very different from those in Half Moon Bay, in this case most species are bright white against low contrast shallow water or sky in early morning light. Unlike Raptors, shorebirds like fly in more predictable patterns, usually straight lines or large circles so the behavior is less erratic.
I did not change any of the camera settings since the raptor shoot.

Here is what I learned:

• Both cameras behaved similarly, there is no indication that my camera is a “bad copy” or it needs calibration.
• When using a static subject (a chain with lots of contrast attached to a post) both cameras exhibited unstable behavior in AI-servo mode and 4-point expansion, both would randomly lose focus and lock on BG in this mode, so by now I am confident the performance of AF expansion mode is erratic at best, I did not use this mode for any of the subsequent shots.
• In single center AF point, tracking Willets against plain BG, results are mixed there are sharp and soft shots within a sequence.
• When tracking Willets against plain BG at close range in Zone AF mode, camera had much better success rate, delivered higher numbers of in focus shots, and it was able to capture “peak of landing” moment (see one sample below)
• Static focus stability with zone AF still remains mixed, please see example.
• I also tested static AF stability in spot AF mode, in some cases camera does focus on tiny features (see the spider example) consistently and delivers a sequence of sharp photos and it some cases it does not deliver a sharp photo at all.
• Overall The keeper ratio was much higher compared to raptor results and camera was able to capture the “peak of action” moment, even if AF was just slightly soft. In other words there were few “full blown” AF shots as in the raptor case.

It is now clear to me that Canon’s new AF system is optimized for speed and it performs best when tracking high contrast, large subjects against low contrast plain BG with a different color. It does not perform well in the cases of erratically moving camouflaged raptors against busy BG, in these cases camera AF is apparently fooled by similarity of color between subject and BG, in these conditions a Canon 1DSMKIII TTL-SIR or the Nikon MULTICAM3500FX module is clearly superior. Overall, my impression of the new Canon AF systems remains mixed, it has potential but it also requires fixes , with current firmware I can see it lead to situations like those experienced in summer 2006. I am sure that there will be people who disagree and will attest to its 100% perfectness, but this is my honest opinion. I hope that Canon provides a mode that can disable color-tracking feature and improves overall AF stability, especially in expansion mode and under busy BG conditions.

Regards,
Arash

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

Willet landing, EOS 7D and 400 f/5.6 L
10-07-2009, 01:40 PM
arash_hazeghi

1 Attachment(s)

Here is a sample of spot AF stability
10-07-2009, 02:10 PM
arash_hazeghi

Example of failed AF in tracking the same Willet moments later, this is a shot from a sequence of 17, none is critically sharp
http://www.stanford.edu/~ahazeghi/Ph...llet%20oof.jpg

I also wanted to add a point, since the camera is 18 mpixels, sometimes if you get the bird to fill the entire frame you can get away with down sampling the photo to web size and sharpening up,
but you wouldn't be able to make prints. If I were a Canon rep I would just show you the smaller frame :D
This is why I do not trust random photos that are posted on the internet, without knowing who took them, what the conditions were and what size the original crop was :)
10-07-2009, 02:34 PM
arash_hazeghi

Here is a demonstration of AF instability, two shots in a row taken from a burst, single AF point kept on the bird.

http://www.stanford.edu/~ahazeghi/Ph...nstability.jpg
10-07-2009, 02:58 PM
arash_hazeghi

Regarding IQ, when in focus, 7D photos are sharp and detailed, noise is a very tight and easy to remove. Dynamic range is also improved to compared to 50D, its files do not have the "harshness" and the noisy look of 50D. It is excellent overall, but not as good as 5DMKII, 1DSMKIII or other full frame cameras. Overall, I don't find noise to be an issue with this camera, files are clean through ISO 800, and 1600 is usable in good light conditions.

Here is an example, at ISO 400.

http://www.stanford.edu/~ahazeghi/Ph...all%20bird.jpg
10-07-2009, 05:13 PM
Aravind Krishnaswamy

Thanks for sharing Arash. I am still hoping that the 7D will be a decent BIF camera (especially hand held BIF with the 400 DO), but your observations are giving me reason to pause.
10-07-2009, 06:29 PM
James Prudente

Check these out.

http://www.fredmiranda.com/forum/topic/821100
10-07-2009, 10:49 PM
James Shadle

A Nikon users 2 cents worth.
I had the opportunity to use(briefly) the 7D this last weekend.

The AF sensor configuration reminded me of the Nikon D200 / D2X cameras.

Using the center sensor only, it was very fast and accurate. Using the 9 and 21 sensor array, I found that photographing BIF was very easy and accurate in a clear sky. However the 9 and 21 sensor array results were nowhere near as accurate when photographing against a background with contrast.

Unlike the Nikon, you can not(from my brief experience anyway) select the sensor to start with when using the 9 and 21 sensor array. The 7D decides where to start. Thus the problem with contrasty backgrounds.

IMO Nikon's ability to allow the user to select the sensor that initially acquires focus (before tracking with other sensors) is what makes them more accurate.

AF Speed Single Sensor Canon 7D
AF Speed Auto AF (clear sky) Canon 7D
AF Tracking Accuracy Nikon D300/700/3/3x
AF Accuracy Auto AF Nikon D300/700/3/3x
10-07-2009, 11:03 PM
Flavio Rose

Thank you for this information, even though it is not good news for me as an amateur photographer who has developed a pretty considerable attachment to his 400 f/5.6.

Just wanted to make sure I've understood one point correctly -- static AF problems with the 7D are limited to situations where you're using expansion points, a feature that you recommend not using at all?

Thanks.
10-07-2009, 11:18 PM
arash_hazeghi

Hi James,

Canon also let's you pick the initial AF sensor for tracking, but only in All-point AF mode which means after that the camera will chose any subset of 19 AF points it wishes, just like Nikon 51-point with 3D tracking, in practice just like 51-point 3D-tracking in an instant, AF points will be all over the place picking whatever they wish.

The major difference between Nikon Dynamic Area AF and Canon expansion mode AF are

1) with Nikon system you can chose how many extra surrounding points you want the camera to "look at" around the point that you select, 9, 11, 21 or all 51, in other word how large the AF net is. With the new Canon system expansion points are always 4 in a diamond pattern, you cannot change the number or pattern.
2) Nikon 3500FX module tries to lock on the point you pick based on the information it gathers from the AF net, however Canon new AF picks whichever point has the highest contrast within the net.
3) Nikon uses overlap servo operation while Canon has discrete servo operation.
4) Nikon AF does not use color information in modes other than 51-point 3D (which is useless anyway as you know), 7D uses color information in all modes, there is no way to disable this. I guess birds in Japan always have different color than BG :D :eek:

The new zone-AF system is something that has no equivalent in Nikon system, you can select an entire zone of AF points (9 points at center or 4 points in corners, total of 5 zones), the camera then selects a number of points within the zone that it wishes based on contrast and possibly subject recognition algorithms.

In practice I agree that Canon locks faster, however many times the result is OOF (at least for me).
10-07-2009, 11:21 PM
arash_hazeghi

Quote:

Originally Posted by Flavio Rose

Thank you for this information, even though it is not good news for me as an amateur photographer who has developed a pretty considerable attachment to his 400 f/5.6.

Just wanted to make sure I've understood one point correctly -- static AF problems with the 7D are limited to situations where you're using expansion points, a feature that you recommend not using at all?

Thanks.

IM experience yes, although in other modes such as spot AF there was still some static instability as the example above shows. I observed no instability or misses when using single AF as opposed to AI servo, so it is an artifact of AI servo operation. For now I recommend switching to single AF for stationary subjects whenever possible.
10-07-2009, 11:49 PM
Morkel Erasmus

thanks for all the info and discussion guys. It's quite interesting to note what various users on various sites have experienced with the 7D...guess I'll have to see if Canon updates firmware before placing my order since I'm not switching brands :)
10-08-2009, 11:06 AM
Joerg Rockenberger

Arash,

Thanks for continuing to share your impressions. Just to make sure: are the pics in this thread still with your original 7D or a second copy?

Thx, JR
10-08-2009, 02:01 PM
Manas Khan

Quote:

Originally Posted by arash_hazeghi

4) Nikon AF does not use color information in modes other than 51-point 3D (which is useless anyway as you know), 7D uses color information in all modes, there is no way to disable this. I guess birds in Japan always have different color than BG :D :eek:

In practice I agree that Canon locks faster, however many times the result is OOF (at least for me).

Hi Arash,

I am new here. Have a plan to buy 7D hence following the related threads and articles....

From where did you get this information ? Or is it from your observation ? I am asking because according to CPN Newsletter (http://cpn.canon-europe.com/content/...d.do#container), the new AF system use the color information from the dual layer metering sensor only to correct the chromatic aberration of the AF optics, not to "identify" the BIF from the BG. And they states that the metering is done concentrating mainly on the subject (in focus) and giving very less priority (if at all, in some situations) to the background (out of focus part of the frame).

Secondly, I have read in the downloaded instruction manual that there is a C. Fn. that allows one to give priority to either fast capture (priority to shutter release) or accuracy (priority to focus lock). Can't one sacrifice the speed to have more "focused" shots ??

Reading the users reactions I am totally confused. Some says it work like dream and they are going to sell their DII/N for the 7D and others say it's far behind the 1D series AF. For some people, setting the "re-focus speed" to slow, it worked so well that the AF kept lock on the subject and didn't go to the background even if the focus point missed the subject http://forums.dpreview.com/forums/re...ssage=33278436 . Seen some clear example of good performance in BIF with busy and tempting background (http://www.birdforum.net/showpost.ph...94&postcount=1, http://forums.dpreview.com/forums/re...ssage=33186206 )... and on the other hand there are some complains too.

Probably will wait for some more time to see some more reviews / user experiences before deciding....

regards,
Manas.
10-08-2009, 03:59 PM
kevinmat

There is new firmware coming out for the 7D, most shipped with 1.0.7 and the update is 1.0.9 - should be available on Canon's website shortly, no details available, but it might be worth trying to see if it makes any difference unless you already have it from CPS.
10-08-2009, 04:21 PM
arash_hazeghi

Quote:

Originally Posted by Manas Khan

Hi Arash,

I am new here. Have a plan to buy 7D hence following the related threads and articles....

From where did you get this information ? Or is it from your observation ? I am asking because according to CPN Newsletter (http://cpn.canon-europe.com/content/...d.do#container), the new AF system use the color information from the dual layer metering sensor only to correct the chromatic aberration of the AF optics, not to "identify" the BIF from the BG. And they states that the metering is done concentrating mainly on the subject (in focus) and giving very less priority (if at all, in some situations) to the background (out of focus part of the frame).

Secondly, I have read in the downloaded instruction manual that there is a C. Fn. that allows one to give priority to either fast capture (priority to shutter release) or accuracy (priority to focus lock). Can't one sacrifice the speed to have more "focused" shots ??

Reading the users reactions I am totally confused. Some says it work like dream and they are going to sell their DII/N for the 7D and others say it's far behind the 1D series AF. For some people, setting the "re-focus speed" to slow, it worked so well that the AF kept lock on the subject and didn't go to the background even if the focus point missed the subject http://forums.dpreview.com/forums/re...ssage=33278436 . Seen some clear example of good performance in BIF with busy and tempting background (http://www.birdforum.net/showpost.ph...94&postcount=1, http://forums.dpreview.com/forums/re...ssage=33186206 )... and on the other hand there are some complains too.

Probably will wait for some more time to see some more reviews / user experiences before deciding....

regards,
Manas.

Manas,
I was told about the use of "color information for tracking" By canon CPS representative, I assume what I was told is correct, unless the rep was not familiar with the camera.

Regarding your second point I have the CF-III 2 set to 0 which is AF priority/track priority and CFIII-3 set to 1 (AF track priority) as well as CF-III-1 set to slow or slow+1, I disclosed all of these details in the first thread.

Also nothing can be said about the tiny and poorly exposed samples posted in armature forums like dpreview.com, there is no way you can tell how sharp or in focus the original photo was, how many shots were taken etc. etc. most of the people there don't know what they are talking about any way. for example the captive raptor shots look very soft and lack feather detail.

I am looking forward to more comments from accomplished BPN photographers here, all of whom I trust. So far, it appears that only Doug has extensively used 7D in the field and he has been satisfied. so there is one data point.

At the end of the day, as usual opinions will be mixed, exactly like the 1D MKIII, to date there are many photographers the believe 1DMKIII had no AF issues either, and there are many who switched brands because of that camera. The best way is to buy one and try for yourself.
10-08-2009, 04:29 PM
arash_hazeghi

Quote:

Originally Posted by John Rain

Arash,

Thanks for continuing to share your impressions. Just to make sure: are the pics in this thread still with your original 7D or a second copy?

Thx, JR

John, we exchanged bodies and tried some static shots in AI-servo, but I shot with my own camera after that.
10-08-2009, 04:33 PM
arash_hazeghi

Quote:

Originally Posted by kevinmat

There is new firmware coming out for the 7D, most shipped with 1.0.7 and the update is 1.0.9 - should be available on Canon's website shortly, no details available, but it might be worth trying to see if it makes any difference unless you already have it from CPS.

where did you get this information from?
10-09-2009, 07:34 AM
Pao Dolina

arash is the AF performance of the 7D superior to those of the 50D? I am asking as I am currently using a 40D and assume the 50D has better AF performance.
10-09-2009, 08:05 AM
Ed Cordes

Quote:

Originally Posted by Pao Dolina

arash is the AF performance of the 7D superior to those of the 50D? I am asking as I am currently using a 40D and assume the 50D has better AF performance.

Well I'll weigh in now for what my small opinion will count for. I hate to get involved in potential disputes, but to answer the direct question - My wife has a 40D and I have been using a 1DMK2. I find the AF of our new 7D to be excellent. To our mind it is better than the 40D by a lot. The 7D is at least as good as my 1DMK2. Granted, this is based on limited experience of about 200 exposures of both static and moving targets using a 300 F4. I hope to test it more this weekend. RE the 50D I think the larget ability to crop more with the 7D is an advantage over the 50D with somewhat better noise quality.

Just my advanced amature $0.02. BTW - This is based on PRINTS at 8 1/2 X 11 and 11 X 14 - not jsut computer monitors.
10-09-2009, 11:40 AM
Pao Dolina

Quote:

Originally Posted by Ed Cordes

Well I'll weigh in now for what my small opinion will count for. I hate to get involved in potential disputes, but to answer the direct question - My wife has a 40D and I have been using a 1DMK2. I find the AF of our new 7D to be excellent. To our mind it is better than the 40D by a lot. The 7D is at least as good as my 1DMK2. Granted, this is based on limited experience of about 200 exposures of both static and moving targets using a 300 F4. I hope to test it more this weekend. RE the 50D I think the larget ability to crop more with the 7D is an advantage over the 50D with somewhat better noise quality.

Just my advanced amature $0.02. BTW - This is based on PRINTS at 8 1/2 X 11 and 11 X 14 - not jsut computer monitors.

Thanks Ed. I am asking as I am considering selling the 40d (good camera for family activities like school futbol matches but a bit wanting for BIF) but the most I could get for it would have me shell an additional $1100 to get the 7d.

I know the ISO is years better than the 40D and the 78% pixel density increase would make taking photos of sea birds much simpler by a tens of meters.

I am deadset in getting the 1DMK4 when it comes out Jan/Feb. Or I could put the $1,100 to use by buying that 100L Macro IS and keep the 40D.
10-09-2009, 11:41 AM
Jamie Strickland

Quote:

Originally Posted by arash_hazeghi

where did you get this information from?

I saw a few posts on FM about this as well, I think it was a guy from Germany that got one with the 1.0.9 on it, not sure what the diff in the firmwares are
10-09-2009, 11:46 AM
Jamie Strickland

Quote:

Originally Posted by Pao Dolina

Thanks Ed. I am asking as I am considering selling the 40d (good camera for family activities like school futbol matches but a bit wanting for BIF) but the most I could get for it would have me shell an additional $1100 to get the 7d.

I know the ISO is years better than the 40D and the 78% pixel density increase would make taking photos of sea birds much simpler by a tens of meters.

I am deadset in getting the 1DMK4 when it comes out Jan/Feb. Or I could put the $1,100 to use by buying that 100L Macro IS and keep the 40D.

might as well wait in that case, the MK4 should be a beast based on the 7D specs

I have not had a good chance to play with my 7D really, but it definitely seems to focus faster than the 40D, based on the very very limited usage it seems to be about the same as the 1D MKII, I will be on vacation next week so I will finally get to use it :D
10-09-2009, 12:41 PM
Pao Dolina

Have fun Jamie. :) I may just forgo the 7d as I have the 5D2.
10-09-2009, 12:47 PM
Jamie Strickland

Quote:

Originally Posted by Pao Dolina

Have fun Jamie. :) I may just forgo the 7d as I have the 5D2.

the 5D2 is an awesome camera, for pic quality I would take it over the 7D for sure... I sold my 5D2 and went back to the original 5D as I don't use FF often

the 1D MK4 & 5D2 will be an awesome combo :)
10-13-2009, 04:00 PM
arash_hazeghi

Based on the Doug's comments I have decided to try a 2nd copy of 7D, I have returned the current one and will update this thread when I get a 2nd copy from B&H.
10-14-2009, 11:23 AM
Paul Granone

I do not see any of those focusing problems with my 7D and 400mm F5.6. The focus for both single shot and servo have been fine.