Reading mode: Light Dark

AF Sensitivity Test Target

Go to last post
Interesting and could be a usefull test. Of course would depend on monitor calibration and room light, lens etc. For example, on my D60, 28.70L, Adobe Gamma calibrated monitor, normal indoor lighting, I can AF on number 49 with some hunting at first. It would be interesting if others could input their equally somewhat unscientific results.

Rich
Julian Loke said:
FWIW, there is an Autofocus Sensitivity Test Target at the the EOS
Documentation Project.

http://eosdoc.com/manuals.asp?q=AFSensTest

Comments and feedback (and test results) are welcome.
Click to expand...
 
Julian,

Great site. Thanks. I just tried my D30. Is there a particular lens you used to perform the test, 50 f/1.8?? Although, I think there are quite a bit of variables to be considered when doing the test, so I'm not sure it would prove to be a valid site for comparisons of two different cameras in different environments.
Matt
Julian Loke said:
FWIW, there is an Autofocus Sensitivity Test Target at the the EOS
Documentation Project.

http://eosdoc.com/manuals.asp?q=AFSensTest

Comments and feedback (and test results) are welcome.
Click to expand...
 
Agreed. Could be a very subjective test. On my uncalibrated Sony 420GS monitor the left square is no longer visible to my eye at around 50. If I place the monitor in "Presentation Mode" then the boxes are visible to my eye down to 25. So without a precisely calibrated monitor the tests are useless in my opionion. I would also expect a wide variation in CRT vs. LCD readings.

-- John
RichW said:
Rich
Julian Loke said:
FWIW, there is an Autofocus Sensitivity Test Target at the the EOS
Documentation Project.

http://eosdoc.com/manuals.asp?q=AFSensTest

Comments and feedback (and test results) are welcome.
Click to expand...
Click to expand...
 
Member said:
http://eosdoc.com/manuals.asp?q=AFSensTest
Click to expand...
It is only unscientific if you don't follow the instructions. The patch number between different monitors is meaningless unless you report the lens used, and the meter reading at ISO 100 for:
. the black point,
. the white point, and
. the darkest patch you could AF

Optionally, convert meter reading to Exposure Values:
http://www.chem.helsinki.fi/~toomas/photo/ev.html

and leave a comment in the guestbook at the bottom of the page.

On the other hand, someone with a D30, a D60, and a 1D could make a valid comparison between the cameras on the same monitor with the same lens.

I feel that the comparison is more valid than pointing at real world objects in variable lighting conditions with different lenses.
 
Julian Loke said:
FWIW, there is an Autofocus Sensitivity Test Target at the the EOS
Documentation Project.

http://eosdoc.com/manuals.asp?q=AFSensTest

Comments and feedback (and test results) are welcome.
Click to expand...
Hi Julian:

I did not "take the test". You offer a very interesting test; it is a great idea and approach - a good starting point! However, IMO the test is very dependent on display brightness and gamma - there is room for some improvement.

To really compare (or test) camera/lens AF performance between users, each user (people like me) should be instructed on the following before performing the test:

1. Pretty important - display brightness and to a limited extent, contrast should be identical for all users. For example, to set up brightness, you might consider the approach outlined by Timo A. at his AIM site. Although I personally don't follow Timo's approach to setting up brightness, his approach is excellent, and anybody can do it. IMO, the brightness setting is critical for comparing AF performance. Contrast is not nearly as important - you need a photometer to set up contrast in a repeatable manner - in a pinch, a "good" camera's light meter will do. BTW - room lights should be turned OFF.

2. Now this is very important - the display gamma MUST be identical for all users. Pick any approach (e.g. Adobe gamma, Praxisoft WiziWYG gamma (free), etc.); but pick one. Again, room lights should be turned OFF.

If my suggestion(s) is already covered elsewhere on your web site, I'm sorry to bother you. Again, my compliments to you on a very good idea for testing AF performance.

Joe Kurkjian
 
Matt Lewis said:
http://eosdoc.com/manuals.asp?q=AFSensTest
Click to expand...
Matt Lewis said:
Great site. Thanks. I just tried my D30. Is there a particular
lens you used to perform the test, 50 f/1.8?? Although, I think
there are quite a bit of variables to be considered when doing the
test, so I'm not sure it would prove to be a valid site for
comparisons of two different cameras in different environments.
Click to expand...
Hi Matt,

So long as you specify the lens used, the test should be consistent and repeatable. The camera can also be used to meter brightness, or at least exposure value using shutter speed and aperture at ASA 100.

Of course, the best way would be to use the same monitor, the same lens, the same focus distance, changing only the camera body. E.g. compare D30, D60, 1D, 1V, etc.

I suspect that only someone at Canon or in photo retail would have access to so many toys.
 
J A K said:
http://eosdoc.com/manuals.asp?q=AFSensTest
Click to expand...
J A K said:
To really compare (or test) camera/lens AF performance between
users, each user (people like me) should be instructed on the
following before performing the test:
...
If my suggestion(s) is already covered elsewhere on your web site,
I'm sorry to bother you. Again, my compliments to you on a very
good idea for testing AF performance.
Click to expand...
Hi Joe,

You are right in wanting to test in a dark room, where the computer monitor is the only source of light.

But rather than relying on a spectrophotometer to generate an ICC profile, my method suggests that you use the excellent light meter in your camera to meter the patches. Convert to Exposure Values at ISO 100 for an objective measure of brightness:
http://www.chem.helsinki.fi/~toomas/photo/ev.html

If you read the text, you can see how I used Exposure Values (EV) to quantify the light levels at the black point, the white point, and the patch. So, my Elan 7E can focus down to EV 6, when presented with 3 EV step to the black point.

The patch numbers are only used to help the unintiated, or where you have multiple cameras to test in front of the same monitor. So, try the test, and please report the Exposure Values to make this a valid comparison.
 
Most of us who got the 1D sold our D30.

Overall, though, it's a very good test for those that have two different cameras. I, for one, would like to see somebody who has a D30, Elan-II, and Rebel to conduct the test. That would prove most interesting.

This test focuses more on contrast than just overall lighting, too. The contrast is very important, and is difficult to control in subjective, anecdotal tests.
Julian Loke said:
On the other hand, someone with a D30, a D60, and a 1D could make a
valid comparison between the cameras on the same monitor with the
same lens.

I feel that the comparison is more valid than pointing at real
world objects in variable lighting conditions with different lenses.
Click to expand...
 
DavidP said:
Most of us who got the 1D sold our D30.

Overall, though, it's a very good test for those that have two
different cameras. I, for one, would like to see somebody who has
a D30, Elan-II, and Rebel to conduct the test. That would prove
most interesting.

This test focuses more on contrast than just overall lighting, too.
The contrast is very important, and is difficult to control in
subjective, anecdotal tests.
Click to expand...
You make a good point about other variables. That's why my method limits the AF target to a single left-to-right hard edged transition between patches that differ in brightness.

The pattern, as well as the contrast, can be vitally important. I have made some optical illusions that can fool phase-contrast Autofocus with a high-contrast, but repeating pixel column patterns.









If you like to push the limits with your EOS gear, have a look at the other pieces at EOSDOC http://eosdoc.com

Voluntary contributions are always welcome.
 
Hi all!

I just performed this with my D60 and thought I would share the results.

First the setup:

First batch D60, 50mm 1.8 MkII lens. (Others were similar) Camera on P with partial metering (the one like this "()" without the dot)

White patch reading: 90 2.5 EV 8.5
Black patch reading: 1" 1.8 EV 1.5
at lowest focusable point: 0"3 1.8 EV 3.5

I was able to focus all the way down to 23. As a note, my lens didn't hunt until it could no longer lock. At 22, it zooms in then CLOSE to focusing then out and then finally gives up. At 23 and higher, the lock was virtually instant.

Good luck all!

Mark
 
Ok, I drug out my Rebel 2000 and these are the results:

First the setup:

Like new Rebel 2000, 50mm 1.8 MkII lens. Camera
on P with center focusing point selected.

White patch reading: 125 2.8 EV 10
Black patch reading: 6 1.8 EV 3.5
at lowest focusable point: 10 1.8 EV 5.5

I was able to focus all the way down to 18. At 17, it zooms in then
CLOSE to focusing then out and then finally gives up. At 18 and

higher, the lock was very fast. Subjectively, the lock seemed to be just as fast as my D60.

By the way, I just realized that my damned REBEL 2000 has 7 AF points... GRR...

Good luck all!

Mark
 
I tested with the Canon 1D.

Lens used: 70-200/2.8 IS at 200mm

AF method: selected center AF point only

Note: results were extremely dependent on the ability to keep the AF sensor steady over a point of contrast. I didn't use a tripod, but just kept trying. I was about 2 meters away. You'd be surprised how much motion occurs at 200mm focal length and a distance of only 2m. I finally found the limit where I could consistently get an AF lock (steady green light) vs. no AF lock (blinking green light).

Patch #10 f/2.8 1.3 sec EV 2.6
Black f/2.8 2.5 sec EV 1.6
White f/2.8 1/160 sec EV 10.3

So, it appears that EV 2.6 is the limit for that f/2.8 lens on the 1D. If I'd used a 50/1.4 lens, presumably I'd have been able to achieve focus lock 2 stops darker, or EV 0.6
 
Dr_ML wrote:

Ok, I drug out my IX also, since people say the AF in the D60 is based on it and these are the results:

First the setup:

Like new Rebel IX, 50mm 1.8 MkII lens. Camera

on P. Unable to select the center focusing point. NO assist lamp on any of the tests.

White patch reading: 125 2.8 EV 10
Black patch reading: 6 1.8 EV 3.5
at lowest focusable point: 10 1.8 EV 5.5

I was able to focus all the way down to 15. At 14, it zooms in then

CLOSE to focusing then out and then finally gives up. On a subjective level, this camera hunts WAAAAAY more than the D60 or the Rebel 2000. It had to hunt beginning in the 50 range. It always locked, it was just slow.

Also on a subjective note, the D60 feels LIGHTYEARS ahead of these cameras in terms of build quality and "feel". Just my $0.02 :)

Good luck all!

Mark
 
Julian Loke83476 said:
http://eosdoc.com/manuals.asp?q=AFSensTest
Click to expand...
I added a nifty Javascript EV calculator to help those like me who can't figure logarithms in their heads.

Thanks to Hugo So for adding some data to eosdoc, but unfortunately, he forgot to write which camera and which lens he was using.

The lens is important because EV 6 in front of an f/2.8 lens is EV 0 at the film/sensor plane. There is also the effect of maximum aperture on depth of field for the focussing sensor(s).
 
Results for Dimage 7

Black 1.5 sec F3.5 = EV 3
White 1/125 sec F3.5 = EV10
Lowest patch 15
Lowest patch exposure 4 sec F 3.5 = EV1.6

The problem that I have is metering the lowest patch, the camera will not meter below 4 second exposures, so the value is approximate.

Bryan
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top