Variations in AF performance: any reason why?

[photo nuts]

So the 7D is colour blind to red as a source for focus? This thread was going in the direction of poor focusing ability in low light but these photos are in fine light. So I suppose in low light with red as your only option to focus on, you'd just be out of luck altogether?

If the size of the red object is about the same as that of the normal AF point in low light, I think the 7D will just make wild guesses. Not all shots will be OOF but only a small fraction will be in focus. I have noticed that in low light, the 7D AF also struggles with dark reddish brown background.

Again the 450D does not exhibit such issues.

 

[ryanc]

Same exact problem here.. I'm on my 2nd 7D and it's going back to amazon soon.

Does the 5D has this same issue? I'm considering spending the extra cash...

 

[Kirk27]

Thank you for this test.

I have performed a similar one (single shot, center point, tripod, IS on and off outdoors on a sunny well lit day using buildings, contrasty bill boards and other static objects for test objects) and got same results with my 17-55 and 7D combo - majority of shots out of focus (and this was after sending camera and lens to canon NJ). No such problem with my 30D or even my old Nikon D70.

However, also found that the number of OOF was significantly less with my other canon lenses (100-400L, 70-200f4L, 100L macro and even the 15-85) when I repeated the test with them. In fact I shot at an international orchid show using the 100L IS macro handheld and only had a couple OOF out of couple hundred shots. The detail that can be obtained with the 7D and macro is simply amazing. But at same show I took 20 shots with the 17-55 (in good light at F8, ISO 100, single center point and shutter speed 320 of still objects/landscapes) and only 3 were in focus.

Also my Tokina 11-16 behaves like the canon 17-55 - majority of shots come out OOF.

And it is not a MF issue either, as I've tested that thoroughly. Not sure what would cause certain lens to behave worse than others on the camera though.

 

[rwbaron]

Thank you for this test.

I have performed a similar one (single shot, center point, tripod, IS on and off outdoors on a sunny well lit day using buildings, contrasty bill boards and other static objects for test objects) and got same results with my 17-55 and 7D combo - majority of shots out of focus (and this was after sending camera and lens to canon NJ). No such problem with my 30D or even my old Nikon D70.

However, also found that the number of OOF was significantly less with my other canon lenses (100-400L, 70-200f4L, 100L macro and even the 15-85) when I repeated the test with them. In fact I shot at an international orchid show using the 100L IS macro handheld and only had a couple OOF out of couple hundred shots. The detail that can be obtained with the 7D and macro is simply amazing. But at same show I took 20 shots with the 17-55 (in good light at F8, ISO 100, single center point and shutter speed 320 of still objects/landscapes) and only 3 were in focus.

Also my Tokina 11-16 behaves like the canon 17-55 - majority of shots come out OOF.

And it is not a MF issue either, as I've tested that thoroughly. Not sure what would cause certain lens to behave worse than others on the camera though.

I did extensive AF tests with 4 7D bodies shooting B&W resolution charts under artificial light. My findings were at 50X FL and the widest aperture of the lens I had nearly 100% in focus with my 50f1.4, 100f2.8 macro, 24~105f4LIS, 70~300DO and 300f4L. I experienced the worst performance on 2 bodies (5 out of 10 not in focus) with my 70~200f4LIS which has no problem on my 40 or 50D under the same testing conditions. I agree that there are some lenses that do not AF well with the 7D although I have no clue as to why.

There was a thread about the 1Dmk4's low light AF performance a while back. I don't know if this is fact but there were several posters who mentioned Canon was using a CMOS AF sensor in the 7D and 1Dmk4 instead of a CCD. They indicated this was a new direction for Canon and that the CMOS was not as sensitive for low-light AF as a CCD. This could all be BS too but the 7D from my tests is not as sensitive for low light AF as my 40D or 50D and there has to be a logical explanation.

Bob

--
http://www.pbase.com/rwbaron

 

[rwbaron]

Your test is interesting and I'm not disputing the results but did you try the same using spot AF? I'm curious to know if it would make any difference.

Bob

The following test shows how sensitive the 7D AF sensor is to color differences.

Camera mounted on tripod. Canon 17-55 f/2.8 IS lens used. IS turned off. Single shot. Center normal AF point, f=35 mm, ISO 200, f/2.8, 1/800 sec

10 shots are fired at the target. The lens is intentionally defocused between shots by turning the lens towards the minimum focusing distance or infinity.

When the AF point rests on the demarcation between the red tiles and yellow walls,



7 out of 10 shots are NOT in focus.

When the shot is out of focus, both the red tiles and trees are out of focus.



so no one can argue the camera focused on the trees. Anyway, at f=35mm and this target distance, the DOF should cover both the trees and red tiles.

The same test when administered on a lowly XSi/450D throws up 10 out of 10 shots which are in PERFECT focus.

Such AF results from an advance camera like the 7D are pathetic.

Some may argue the lens/camera combo is not properly calibrated. To disprove this, the same test is repeated with the AF point resting on the demarcation between the green and yellow portions of the same building.



Result: 10 out of 10 shots in focus (see left portion of above pic).

The test (10 shots fired each time) is repeated several times and the results are always the same.

I only carried out extensive testing 'cos I experienced a number of OOF shots from the 7D during an event shooting.

For all its advance features and superb handling, a camera that cannot AF reliably for simple single shot photography is completely useless.

--
http://www.pbase.com/rwbaron

 

[photo nuts]

Your test is interesting and I'm not disputing the results but did you try the same using spot AF? I'm curious to know if it would make any difference.
Bob

I've previously carried out tests to identify behavioral differences between spot and normal AF. I find their relative advantages vary according to the shooting situation. For the tests I show in the above post, there is little to be gained by employing spot AF.

 

[ryanc]

Does the 5D has this same issue? I'm considering spending the extra cash...

Never mind, I just visited some 5D forums -- Looks like the same problems exists for many 5D owners What a bummer..

 

[schmegg]

So the 7D is colour blind to red as a source for focus? This thread was going in the direction of poor focusing ability in low light but these photos are in fine light. So I suppose in low light with red as your only option to focus on, you'd just be out of luck altogether?

Just been playing around with your question here at home.

I have a situation here with a red subject in low light that gives 100% focus lock with phase detect and 0% focus lock with contrast detect. That's using a 24-105/f4.

Using the 17-55/f2.8 on the same subject I get 100% focus lock with phase detect (again) and about 50% focus lock with contrast detect. Worse still is that the contrast detect method reports focus lock often when the focus is obviously WAY off (and I mean way, way off!)

Pointing at a non-red subject in the same light gives 100% focus lock with both methods.

This is pretty rough and ready testing, and I'm intending to do more over the weekend in a more controlled manner, but so far it seems to me that red is indeed a problem (for my 7D, at least).

 

[jm67]

So the 7D is colour blind to red as a source for focus? This thread was going in the direction of poor focusing ability in low light but these photos are in fine light. So I suppose in low light with red as your only option to focus on, you'd just be out of luck altogether?

Just been playing around with your question here at home.

I have a situation here with a red subject in low light that gives 100% focus lock with phase detect and 0% focus lock with contrast detect. That's using a 24-105/f4.

Using the 17-55/f2.8 on the same subject I get 100% focus lock with phase detect (again) and about 50% focus lock with contrast detect. Worse still is that the contrast detect method reports focus lock often when the focus is obviously WAY off (and I mean way, way off!)

Pointing at a non-red subject in the same light gives 100% focus lock with both methods.

This is pretty rough and ready testing, and I'm intending to do more over the weekend in a more controlled manner, but so far it seems to me that red is indeed a problem (for my 7D, at least).

I must admit I wrote what's above with a grain of salt but I too had a little time on my hands and did the same as you.

Just a few quick shots at low, low light and two targets, one a red box with white text, another blue with white text. Using only the 17-55 at 2.8 my in-focus results on the blue box had better results than the red. All times my focus point flashed "lock".

Very rough initial shots and again, low enough light that I needed 3200 or 6400 which may have influenced the results somewhat. I'll also have to try it with my 40d which can take this lens and also other lenses and more reasonable lighting as I'm not sure low light is necessarily a factor as photonuts' shots where well lit.

Odd though, that in "real life" shooting I've used the camera for since last fall, I've not really noticed a lot of unexplainable oof shots. Maybe I've been lucky so far.

 

[40d_dane]

You should be careful - this forum is filled with people that will cry "user error" if someone claim that their 7D doesn't focus properly ;-)

I've been through 4 7D's previously which were all returned. Three were returned because they miss focused in the field (several x the DOF). In controlled testing they showed no problems with AF. I even printed out an image of bird that the 7D could not focus on (taken with my 40D) and used that as a test target at home in my "lab" (my living room). No problems - perfect focus. It may have been an low light problem. I didn't test that. I also submitted one of the field shots showing bad AF (front focus by several times the DOF) to Canon support. They were unable to say anything remotely intelligently about the miss focus, but I also doubt that the sample ever made its way to someone with real solid knowledge of the 7D focus system.

I am now on my fifth 7D which seem to work just fine both in my "lab" and in the field. However, I haven't had the opportunity to shoot under the same conditions that caused the AF problems with the 7D's I tested previously.

As I suspected that the issue I had with the previous 7D's could have been related to the color of the target and as you claim colorblindness I decided to take my new 7D to the "lab" with my 400mm f5.6. I see small variations in AF on multiple shots but I don't really see anything that would indicate blindness to red. This is a representative sample of focus on a black/red target using phase detect AF:





This is a representative sample of focus on the same target using contrast detect (live view) AF:





This is a representative sample of focus on a black/white target (which according to a Canon service technician is far to busy for AF testing - however it works just fine for that purpose):





Please note that these tests were carried out in a hurry without paying appropriate attention to perfect alignment. To be honest, they are kind of sloppy. However, I think that they show that my fifth 7D works fine AF'ing on red.

I've done no real low light testing but as you can see on the shutter speeds used the lighting on these test shots were pretty dim !

 

[mmullen]

This agrees with the results of a simple test I did against my 40 and 50D. The 7D AF system is excellent and highly configurable but it appears that Canon took a step backward in low light sensitivity.

True, and the reason for this is quite simple:

The regular precision AF sensors in the 7D are more accurate than the regular precision sensors from the 40D and other previous cameras. To achieve the higher accuracy, the pixels of the AF sensor are smaller (and thus they collect less light).

If Canon did not make the AF sensor pixels smaller, the AF accuracy would not be high enough to consistently take advantage of the extra resolution the 7D offers over previous models.

--
Mike Mullen

 

[jm67]

Interesting. With those speeds light must have been low as was my informal few shots a while ago. Sorry, I didn't keep them as they were not shot 'properly' as to rule out other causes of slight shift in dof.

A few things with your tests though. Maybe I did my quick and dirty ones wrongly but I used only the 17-55 2.8 @ 2.8. Far from a scientific conclusion but I get the impression there are some lenses that just seem to not work 100% and this may be one of them with schmegg mentioning the 24-105 as well. And I suppose I don't have much clue as to how af works but I always though simply contrast was all that was needed. Wouldn't dark red/white be all that's needed for af to 'find' the mark?
Hopefully this weekend I'll try some stuff out if I have time.

 

[schmegg]

Odd though, that in "real life" shooting I've used the camera for since last fall, I've not really noticed a lot of unexplainable oof shots. Maybe I've been lucky so far.

I should have added that too - in real world I've been more than happy with the AF performance of my 7D.

I'll do some better testing this weekend and get back with results.

I should mention that I'm still at V1.1.0 so I'll test at that and then, if there seems to be a problem, I'll go up to the latest and test again. Hope it doesn't get worse though because you can't go back apparently (thanks Canon :-( )

 

[40d_dane]

A few things with your tests though. Maybe I did my quick and dirty ones wrongly but I used only the 17-55 2.8 @ 2.8. Far from a scientific conclusion but I get the impression there are some lenses that just seem to not work 100% and this may be one of them with schmegg mentioning the 24-105 as well. And I suppose I don't have much clue as to how af works but I always though simply contrast was all that was needed. Wouldn't dark red/white be all that's needed for af to 'find' the mark?

The lens is an integral part of the phase detect auto focus system. So it is natural that the choice of lens influence the results. This is also why micro adjust can be done per lens. I've used my 400mm on all the 5 7D's I've tested so the lens has been a "constant" in my testing.

It was implied that the AF system doesn't pick up red. If that was the case, testing on a black/red target as I did should have the AF system in serious trouble - basically the target should be seen by the sensor as no-contrast black if the red I used was a "perfect" red. When I look at the color on my test shots, the red color actually contain quite a bit of green (about 25%) and a little blue so I will try to redo over the weekend... it seems that I did the test too fast... the green may come from the light used. The test patch looks quite a bit more red in real life.

 

[MM1]

[snip]

There was a thread about the 1Dmk4's low light AF performance a while back. I don't know if this is fact but there were several posters who mentioned Canon was using a CMOS AF sensor in the 7D and 1Dmk4 instead of a CCD. They indicated this was a new direction for Canon and that the CMOS was not as sensitive for low-light AF as a CCD. This could all be BS too but the 7D from my tests is not as sensitive for low light AF as my 40D or 50D and there has to be a logical explanation.

Yes, this was mentioned here, but also in the past, as a difference between film EOS cameras including very first digital cameras, versus the later models. I read it here when I joined in 2006 already.

--
Cheers,
Martin

 

[40d_dane]

I've found that on the 7D, the color temp of the light source causes a focus shift. It seems that my 7D focuses quite precisely in low light if the light source is 4 "white" LED's. Using incandescent light, my 7D front focuses quite a bit. My 40D focuses quite independently of light source it seems. I uploaded some test images to my gallery but realized that I can improve on the performance by MA'ing the lens so I will try to do that over the weekend and repost. This focus shift most likely is not something that vary significantly from 7D to 7D. I fear it's the nature of the beast :-(

 

[jm67]

I've found that on the 7D, the color temp of the light source causes a focus shift. It seems that my 7D focuses quite precisely in low light if the light source is 4 "white" LED's. Using incandescent light, my 7D front focuses quite a bit. My 40D focuses quite independently of light source it seems. I uploaded some test images to my gallery but realized that I can improve on the performance by MA'ing the lens so I will try to do that over the weekend and repost. This focus shift most likely is not something that vary significantly from 7D to 7D. I fear it's the nature of the beast :-(

The shots with the 40d/incandescent seem a bit front focused too. I don't think ma will help unless you plan to shoot only under incandescent light (if this theory were to hold true).

I did some quick ones myself today just with 7d/17-55 @ 2.8, 55mm, 35mm. The target was the old 40d box at 6' which we all know is red with white "Canon" on it. I shot tripod, no IS, sunlight and out of 7 shots, they were consistent (and in focus), matching LV shot for reference. The only time I didn't get consistent focus was focusing on the word 'Canon' when wide. Focus improved if I focused on the top of the box (focus square half on box, half wall). Not scientific but it's all I had time for. Maybe there's more to this 'shooting under incandescent' theory but it still doesn't explain photonuts' shots.

 

[Bob Blount]

Do you have any large images to post?
--
Bob

'I can look at a fine art photograph and sometimes I can hear music.' - Ansel Adams

 

[J A K]

I did extensive AF tests with 4 7D bodies shooting B&W resolution charts under artificial light. My findings were at 50X FL and the widest aperture of the lens I had nearly 100% in focus with my 50f1.4, 100f2.8 macro, 24~105f4LIS, 70~300DO and 300f4L. I experienced the worst performance on 2 bodies (5 out of 10 not in focus) with my 70~200f4LIS which has no problem on my 40 or 50D under the same testing conditions. I agree that there are some lenses that do not AF well with the 7D although I have no clue as to why.

Darn it, now you have me worried! I just bought a new lens and it is without camera. So anyway, I was thinking about a 7D because a couple of shooters I know are doing very well with that camera shooting BIF; however, to my knowledge they have never used the 7D at low levels of illumination. Do you recall what the shutter speed, aperture (I know you said widest), and ISO settings were for your tests? If you could provide that information I can get a rough feel for the illumination's EV by assuming proper exposure on your part. Just out of curiosity, when you shot the resolution chart was there a patch of 18% gray somewhere in the scene?

There was a thread about the 1Dmk4's low light AF performance a while back. I don't know if this is fact but there were several posters who mentioned Canon was using a CMOS AF sensor in the 7D and 1Dmk4 instead of a CCD. They indicated this was a new direction for Canon and that the CMOS was not as sensitive for low-light AF as a CCD. This could all be BS too but the 7D from my tests is not as sensitive for low light AF as my 40D or 50D and there has to be a logical explanation.

The 1DmkIV white paper talks about a "new low-noise CMOS AF unit" but unfortunately there is no white paper (that I'm aware of) for the 7D. So, I think your "BS" conclusion is probably a good one.

Additionally, I believe "you" regarding the low light AF capability of the 7D versus the 50D/40D (AFAIC you are trusted source). Let me ask a question, do you still have your 40D and/or 50D available for further AF testing, TIA for your reply?

Regards,

Joe Kurkjian

Galleries: http://www.pbase.com/jkurkjia

--



SEARCHING FOR A BETTER SELF PORTRAIT

 

[rwbaron]

I did extensive AF tests with 4 7D bodies shooting B&W resolution charts under artificial light. My findings were at 50X FL and the widest aperture of the lens I had nearly 100% in focus with my 50f1.4, 100f2.8 macro, 24~105f4LIS, 70~300DO and 300f4L. I experienced the worst performance on 2 bodies (5 out of 10 not in focus) with my 70~200f4LIS which has no problem on my 40 or 50D under the same testing conditions. I agree that there are some lenses that do not AF well with the 7D although I have no clue as to why.

Darn it, now you have me worried! I just bought a new lens and it is without camera. So anyway, I was thinking about a 7D because a couple of shooters I know are doing very well with that camera shooting BIF; however, to my knowledge they have never used the 7D at low levels of illumination. Do you recall what the shutter speed, aperture (I know you said widest), and ISO settings were for your tests? If you could provide that information I can get a rough feel for the illumination's EV by assuming proper exposure on your part. Just out of curiosity, when you shot the resolution chart was there a patch of 18% gray somewhere in the scene?

Hi Joe,

All the above tests were done under artificial (halogen) lights at EV 11 shooting at the max aperture for each lens at ISO 100. The 7D's that did not AF well were very early units (preorder) and were inconsistent with the 70~200f4LIS in daylight too even after extensive MA work. I need to say that the 7D I currently have from late December 09 AF's extremely well with all of my lenses. IMO Canon had an AF calibration issue with early batches of 7D's that I believe has been resolved. This was a standard USAF 1951 resolution chart and there was no patch of 18% gray in the scene.

There was a thread about the 1Dmk4's low light AF performance a while back. I don't know if this is fact but there were several posters who mentioned Canon was using a CMOS AF sensor in the 7D and 1Dmk4 instead of a CCD. They indicated this was a new direction for Canon and that the CMOS was not as sensitive for low-light AF as a CCD. This could all be BS too but the 7D from my tests is not as sensitive for low light AF as my 40D or 50D and there has to be a logical explanation.

The 1DmkIV white paper talks about a "new low-noise CMOS AF unit" but unfortunately there is no white paper (that I'm aware of) for the 7D. So, I think your "BS" conclusion is probably a good one.

Additionally, I believe "you" regarding the low light AF capability of the 7D versus the 50D/40D (AFAIC you are trusted source). Let me ask a question, do you still have your 40D and/or 50D available for further AF testing, TIA for your reply?

I do still have both but am looking to sell the 40D shortly as I never use it or the 50D for that matter but I still need a backup body. I can tell you that Imaging Resource found the 7D AF to be less sensitive than the 50D in low light as follows;

"The Canon 7D's phase-detect autofocus system was able to focus on the subject to just below 1/8 foot-candle light level unassisted, and in complete darkness with the AF assist enabled. That's not quite as good as most semi-pro SLRs. (The Canon 50D's AF system could focus unassisted to less than 1/16 foot-candle.) We were hoping the production-level Canon 7D would improve over the prototype in that regard, but unfortunately, it didn't. In Live View mode, the 7D's contrast-detect autofocus was able to focus down just below the 1/4 foot-candle level without assistance. That's fairly typical for contrast-detection AF."

This follows with a very rudimentary test I did where the 7D would hunt and never lock but the 40 and 50D both locked immediately. This was on a low contrast target indoor with natural light at EV 9.

What tests are you looking for?

Bob

SEARCHING FOR A BETTER SELF PORTRAIT

--
http://www.pbase.com/rwbaron