Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username   Password

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Canon Forum | Join Upload & Sell

  

Archive 2012 · FoCal's test results on 6D AF consistency
  
 
thw2
Offline
• • • •
Upload & Sell: Off
p.1 #1 · p.1 #1 · FoCal's test results on 6D AF consistency


The latest blog entry from Rich.

Interesting results. 6D AF consistency is as good as D800. But for 5D2 and 7D, both phase and contrast detect are pretty bad. This may be surprising to many, but not for me (about 7D AF).



Dec 18, 2012 at 01:15 PM
PhilDrinkwater
Offline
• • •
Upload & Sell: Off
p.1 #2 · p.1 #2 · FoCal's test results on 6D AF consistency


Very interesting. It's great to get some actual data on this kind of stuff.

Will be interesting to see the 5diii and 1dx when they're added



Dec 18, 2012 at 01:54 PM
ben egbert
Offline
• • • • •
Upload & Sell: On
p.1 #3 · p.1 #3 · FoCal's test results on 6D AF consistency


good stuff. I wonder how the 1DS-mk3 compares here? I notice focus variability between shots and when testing, I always adjust to infinity between shots. The focus distance EXIF data also varies.

A theory I have about needing to go to infinity between shots is that the system may not have a closed loop system and when the lens is already close, it gets an impossibly small correction value.

The greater accuracy of the Nikon and 6D could indicate a closed loop.

Way out of my expertise here but accurate AF is a holy grail for me so I have a lot of interest.




Dec 18, 2012 at 05:02 PM
RCicala
Offline
• • • •
Upload & Sell: On
p.1 #4 · p.1 #4 · FoCal's test results on 6D AF consistency


One thing to note, though: different lenses have different AF accuracy within a system. There are many other factors involved, but for simplicity:

1) Camera tells lens to go here.
2) Lens goes here, camera rechecks
3) If needed repeat

Let's say camera A can be accurate to +/- 0.5 and camera B to +/- 0.1.
If the lens is accurate to + / - 0.5, changing cameras may make no difference even though B is better.

This is fairly true. For example the Canon 50mm f/1.4 can recognize about 750 steps of command, while the 85 f/1.2 can recognize over 4,000 steps. Plus steps are not linear throughout the focus range. A macro like he tested with Canon has fewer steps at infinity and more at the macro focusing range. I'd be very hesitant to compare between brands unless very similar lenses were used, and even then only with great care.

One other thing to note: contrast detection AF is much more light and ISO dependent than phase detection. If you shoot at ISO 3200 phase may be just as accurate as at 400, but contrast is not.

I agree the results are very interesting. But having done some of these type tests myself, I found there's an awful lot of variables and extrapolating too much is a temptation.

Edited on Dec 18, 2012 at 06:22 PM · View previous versions



Dec 18, 2012 at 06:15 PM
snapsy
Offline
• • • •
Upload & Sell: On
p.1 #5 · p.1 #5 · FoCal's test results on 6D AF consistency


RCicala wrote:
I agree the results are very interesting. But having done some of these type tests myself, I found there's an awful lot of variables and extrapolating too much is a temptation.


Excellent points Roger, and knowing how careful you are with methodology makes me far more apt to trust your data over others'.



Dec 18, 2012 at 06:20 PM
RCicala
Offline
• • • •
Upload & Sell: On
p.1 #6 · p.1 #6 · FoCal's test results on 6D AF consistency


One other thing of note, a couple of people who have reverse engineered Canon's AF algorithms point out on different lenses you require a different number of steps to go from near-to-far than from far-to-near. That could make accuracy different depending on which direction the AF is attempting.

I corrected the 80mm f/1.2 numbers above, it has 4,000. I was going off the top of my head and I've looked them up now.

For some comparisons:
50 f/1.4 660
16-35 f/2.8 1,500
35mm f/1.4 2,500
70-200 f/2.8 IS mk 1 2,500



Dec 18, 2012 at 06:22 PM
Beni
Offline
• • • • •
Upload & Sell: On
p.1 #7 · p.1 #7 · FoCal's test results on 6D AF consistency


Fascinating that a ultra wide angle f2.8 zoom has so many more steps that a 50mm 1.4!


Dec 18, 2012 at 06:43 PM
RCicala
Offline
• • • •
Upload & Sell: On
p.1 #8 · p.1 #8 · FoCal's test results on 6D AF consistency


Beni wrote:
Fascinating that a ultra wide angle f2.8 zoom has so many more steps that a 50mm 1.4!


I think that's more about the age of the design than anything else. Plus it has the older type motor.



Dec 18, 2012 at 06:59 PM
ben egbert
Offline
• • • • •
Upload & Sell: On
p.1 #9 · p.1 #9 · FoCal's test results on 6D AF consistency


RCicala wrote:
One other thing of note, a couple of people who have reverse engineered Canon's AF algorithms point out on different lenses you require a different number of steps to go from near-to-far than from far-to-near. That could make accuracy different depending on which direction the AF is attempting.

I corrected the 80mm f/1.2 numbers above, it has 4,000. I was going off the top of my head and I've looked them up now.

For some comparisons:
50 f/1.4 660
16-35 f/2.8 1,500
35mm f/1.4 2,500
70-200 f/2.8 IS mk 1 2,500


Does this help explain why some lenses do so poorly at infinity?



Dec 18, 2012 at 07:07 PM
ben egbert
Offline
• • • • •
Upload & Sell: On
p.1 #10 · p.1 #10 · FoCal's test results on 6D AF consistency


RCicala wrote:
One thing to note, though: different lenses have different AF accuracy within a system. There are many other factors involved, but for simplicity:

1) Camera tells lens to go here.
2) Lens goes here, camera rechecks
3) If needed repeat



So this sounds like a feedback loop rather than open ended. Answers one of my questions.

I do see a difference in focus during MA testing depending on whether if I start at infinity or near focus. I always assumed the af tolerance was based on some percentage of DOF, and DOF math is sensitive close and insensitive as you approach infinity.



Dec 18, 2012 at 07:11 PM
 

Search in Used Dept. 



RCicala
Offline
• • • •
Upload & Sell: On
p.1 #11 · p.1 #11 · FoCal's test results on 6D AF consistency


ben egbert wrote:
Does this help explain why some lenses do so poorly at infinity?


Ben, I don't think so. I think that may be more lens design or lens calibration. But I'm not certain we're talking about the same thing.

As to the feedback loop, as best I know in recent years it has been closed; there is at least a check after the first lens direction command.

I've heard that early lenses and systems may not always have been (hence the 're-autofocus' or 'hit focus button twice' that some of us were taught helped AF be more accurate).



Dec 18, 2012 at 07:21 PM
D. Diggler
Offline
• • • • •
Upload & Sell: Off
p.1 #12 · p.1 #12 · FoCal's test results on 6D AF consistency


RCicala wrote:
I've looked them up now.

50 f/1.4 660
16-35 f/2.8 1,500
35mm f/1.4 2,500
70-200 f/2.8 IS mk 1 2,500


Curious where these numbers come from. The figures for other lenses listed online anywhere?



Dec 19, 2012 at 12:10 AM
M Lucca
Offline
• • •
Upload & Sell: Off
p.1 #13 · p.1 #13 · FoCal's test results on 6D AF consistency


thw2 wrote:
The latest blog entry from Rich.

Interesting results. 6D AF consistency is as good as D800. .


HUh? That means the 6D won't AF on the left side too? Ack!



Dec 19, 2012 at 01:11 AM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #14 · p.1 #14 · FoCal's test results on 6D AF consistency


RCicala wrote:
One other thing of note, a couple of people who have reverse engineered Canon's AF algorithms point out on different lenses you require a different number of steps to go from near-to-far than from far-to-near. That could make accuracy different depending on which direction the AF is attempting.

I corrected the 80mm f/1.2 numbers above, it has 4,000. I was going off the top of my head and I've looked them up now.

For some comparisons:
50 f/1.4 660
16-35 f/2.8 1,500
35mm f/1.4 2,500
70-200 f/2.8 IS mk 1 2,500


I figure that these steps correspond to the AFMA steps used in recent DSLR that offer AF microadjustments.

This could explain why AF on the 80/1.2L is relatively slow; 4000 steps is a lot more than the other lenses. OTOH, with such a slim DOF, it makes good sense.

Roger - do you know the number of AF steps for the EF 50/1.2L ?



Dec 19, 2012 at 01:17 AM
skibum5
Offline
• • • • • •
Upload & Sell: Off
p.1 #15 · p.1 #15 · FoCal's test results on 6D AF consistency


jcolwell wrote:
I figure that these steps correspond to the AFMA steps used in recent DSLR that offer AF microadjustments.

This could explain why AF on the 80/1.2L is relatively slow; 4000 steps is a lot more than the other lenses. OTOH, with such a slim DOF, it makes good sense.

Roger - do you know the number of AF steps for the EF 50/1.2L ?


I think the 85 1.2 is slow, at least in part, because it has a really huge focusing element with a ton of mass.



Dec 19, 2012 at 02:16 AM
RCicala
Offline
• • • •
Upload & Sell: On
p.1 #16 · p.1 #16 · FoCal's test results on 6D AF consistency


Here's the list compiled by Berger when they reverse engineered the mounts for RED

http://redmodz.com/red-birger-lens-mount-unofficial-mini-faq-and-thread-compilation/

RED had some other numbers they've released but I don't have the link anymore.



Dec 19, 2012 at 03:19 AM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #17 · p.1 #17 · FoCal's test results on 6D AF consistency


Thanks Roger.


Dec 19, 2012 at 03:35 AM
thw2
Offline
• • • •
Upload & Sell: Off
p.1 #18 · p.1 #18 · FoCal's test results on 6D AF consistency


RCicala wrote:
One thing to note, though: different lenses have different AF accuracy within a system....

Let's say camera A can be accurate to +/- 0.5 and camera B to +/- 0.1.
If the lens is accurate to + / - 0.5, changing cameras may make no difference even though B is better.

This is fairly true. For example the Canon 50mm f/1.4 can recognize about 750 steps of command, while the 85 f/1.2 can recognize over 4,000 steps. Plus steps are not linear throughout the focus range. A macro like he tested with Canon has fewer steps at infinity and more at
...Show more

Thanks for your input Roger.

I gave some thought to what you said and realize that FoCal's test results are not exactly invalid. While platform to platform comparison (i.e., Canon to Nikon) may be difficult due to inherent differences in the lenses and cameras, all the Canon results tested with the same 100 f/2.8L lens should hold. Even if the 100 mm macro lens has fewer steps near infinity, the results show the 6D phase contrast is still more reliable than what can be achieved with the 5D2 and 7D under identical test conditions. And if the test conditions were too dim (although I find it hard to believe), then it shows that 6D contrast AF works better under dim light conditions than 5D2 and 7D. Yes?


Edited on Dec 19, 2012 at 05:13 AM · View previous versions



Dec 19, 2012 at 03:51 AM
D. Diggler
Offline
• • • • •
Upload & Sell: Off
p.1 #19 · p.1 #19 · FoCal's test results on 6D AF consistency


jcolwell wrote:
do you know the number of AF steps for the EF 50/1.2L ?


Haha. That's the lens I was wondering about.



Dec 19, 2012 at 04:40 AM





FM Forums | Canon Forum | Join Upload & Sell

    
 

You are not logged in. Login or Register

Username   Password    Reset password