Julian Loke83476
Active member
Hi Bryan,
Thanks for the response. Can you increase the sensitivity for metering? E.g. change to ASA 200 or 400?
The popup EV calculator can cope with other ASA values too.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
AF Sensitivity Test Target
- Thread starter Julian Loke83476
- Start date
Thanks for the response. Can you increase the sensitivity for metering? E.g. change to ASA 200 or 400?
The popup EV calculator can cope with other ASA values too.
Bryan Biggers
Senior Member
Oh! Good idea. I can't get the pop up calculator to work so I have to use my HP200
Let me try at ISO 800...
OK, I get at ISO 800
White 1/500 at F4.5 = EV 10.3
Black 1/6 at F3.5 = EV 3.2
Patch 15 0.5 at F3.5 = EV 1.6
(better check my math, I had to guess how to calc EV at higher ISO (just multiplied the shutter speed by 8 in this case)
The funny thing is the way that it will focus on patch 15 fine, but I can't ever get it to focus on 14, even though there is no difference by eye.
Bryan
Bill Branham
Forum Enthusiast
Great work, Julian!!! I have an Elan IIe, A2e, EOS 3, and D30. When I get my D60 this should make a nice test. I agree with other's comments that monitor variables might be too great to compare my data with that someone else might get, but it should do a great job on my battery of AF cameras. I have added your sites to my links and will let you know my results.
Thanks again for the test materials.
Mike Price
Forum Enthusiast
How ever a brief test to compare 1D and D30:
1D could find focus down to 11
D30 could find focus down to 24
Both would confirm focus a couple of levels below if already focused.
Some other factors that prevent this from being a true global test:
Does the AF sensor use horizontal, vertical or diagonal contast measurement?
Does the screen refresh interfere with the AF sampling?
Screen resolution/focus
Screen type:
LCD v CRT
Trintron v traditional (what's it called?)
Julian Loke83476
Active member
Hi Bryan,Oh! Good idea. I can't get the pop up calculator to work so I have
to use my HP200
The funny thing is the way that it will focus on patch 15 fine, but
I can't ever get it to focus on 14, even though there is no
difference by eye.
The pop-up calculator requires Javascript with Math functions.Can you e-mail me privately about why the pop-up calculator is not working so that I can fix it?
Make sure that your observation is not the result of a bit-depth artifact, ensure that your display is set to at least millions of colors (24 bit or 32 bit). Also, changing the brightness on the monitor changes the black and white points, so you should get a different AF limit patch (e.g. patch 35 instead of 15).
Julian Loke83476
Active member
Hi Mike,
The AF sensors in EOS cameras are sensitive to horizontal and to vertical detail. The central AF point(s) have a pair of sensors dedicated to both horizontal and vertical, while most peripheral sensors are only sensitive to details in a single direction. The lens used can also affect which sensors are used for AF. Check Canon's own technical information for further information.
I accept your concern about the validity of comparisons between different observers, using different monitors, different lenses, and different cameras. I think you will agree that a standardized AF sensitivity target is better than anectodal reports of being (un)able to focus on a green couch lit by fluorescent lights in someone's basement.
Ideally, if someone used the same monitor with a fixed refresh rate, the same part of the monitor screen, same lens, same focus distance, but different cameras, and different patch brighness, then the RANK ORDER (1D, D60, D30, etc) should remain the same, and is statistically robust.
Metering the patches is however dependent upon the accuracy of the in camera meter. Again, however, if the same camera for metering within a test session using the same monitor, then that reading should also be valid and reproducible.
To my way of thinking, you have just verified that the method works since you were able to induce AF failure. If you could even focus on the screen on patch 0, the test becomes invalid, because the camera was able to detect the shadow mask, or a moire pattern, or the lcd pixel array. (I have some other test patterns to show that this is not a confounding factor).
We just need to find an interested photo retailer or photo reviewer (hint hint) to test the AF sensitivity of several cameras on a single monitor, perhaps even with a calibrated handheld spot meter
Hugo So
Leading Member
unfortunately, you forgot to see with your eyes, on the subject headings and the amendment.
P
Phil Askey
Guest
Ouch... This is pretty dangerous, there's no way for you to know what gamma a certain person is using on their monitor, nor what brightness / contrast settings or the luma output of the monitor itself...
Julian Loke83476
Active member
Hugo So posted at http://eosdoc.com/manuals.asp?q=AFSensTest
I could follow your posts here at dpreview.com, but I was referring to the completely separate feedback form at eosdoc.
P.S. The EV calculator should now work for NS 6.2 users too.
Hi Hugo,
I could follow your posts here at dpreview.com, but I was referring to the completely separate feedback form at eosdoc.
P.S. The EV calculator should now work for NS 6.2 users too.
Julian Loke83476
Active member
http://eosdoc.com/manuals.asp?q=AFSensTest
Thanks for your site, and for the feedback.
You are absolutely correct. Gamma is not important, so long as patch 1 is brighter than patch 0, and patch 2 is brighter than patch 1, and patch 3 is brighter than patch 2. The patch numbers are not important. Rather, metering the patches (e.g. black=EV3, white=EV 10, focus=EV 6) helps even out the differences between monitors.
As I have written in several posts above, the methods works best if there is one monitor used, and several cameras tested against that same monitor. Then, the RANK ORDER of AF sensitivity of the cameras should be the same, and should be reproducible even if using a different monitor in a different location.
I predict that the rank order (Best to worst) should be EOS 1D, EOS D60, EOS D30, etc., but it would be very interesting to see the actual results of comparing the D60 and D30.
Hi Phil,
Thanks for your site, and for the feedback.
You are absolutely correct. Gamma is not important, so long as patch 1 is brighter than patch 0, and patch 2 is brighter than patch 1, and patch 3 is brighter than patch 2. The patch numbers are not important. Rather, metering the patches (e.g. black=EV3, white=EV 10, focus=EV 6) helps even out the differences between monitors.
As I have written in several posts above, the methods works best if there is one monitor used, and several cameras tested against that same monitor. Then, the RANK ORDER of AF sensitivity of the cameras should be the same, and should be reproducible even if using a different monitor in a different location.
I predict that the rank order (Best to worst) should be EOS 1D, EOS D60, EOS D30, etc., but it would be very interesting to see the actual results of comparing the D60 and D30.
P
Phil Askey
Guest
Also don't forget that we don't know what affect the monitor refresh flicker will have on the AF systems.
Julian Loke83476
Active member
Hi Phil,
Thank you for your question, which seems to invoke some aspects of Englishman Francis Bacon's ideas about scientific induction
http://www.blupete.com/Literature/Biographies/Philosophy/Bacon.htm
Because my AF test method relies on inducing AF failure, it is its own internal control. By that I mean that if AF was affected by the monitor shadow mask, or the refresh rate, or dust in front of the screen, or gamma, or any other monitor factor, then those confounding factors are constant, and should interfere with AF on patch 255 (white) through to patch 1 (black). Or, allow AF to occur even when there is no change in brightness (e.g. patch 0).
I can also repeat the experiment any number of times to show that it was not some time dependant factor (such as refresh rate, room lighting, etc). If I get varying results (e.g. patch 24, patch 64, patch 13), then the test is invalidated.
So, if you point the lens at the exactly the same place on the same monitor, and can induce AF failure only by varying the brightness transition (i.e. scrolling the document), I can only conclude that since the sole change was the brightness step, it must have been the causal factor for inducing AF failure.
As I wrote before, analysis of the results is a lot easier if the test is used to generate a RANK ORDER of AF sensitivities between cameras, rather than an ABSOLUTE measure of the brightness step required to allow AF to occur.
Tom Crowning
Senior Member
I don't understand several things:
White: 1/125 sec f/3.5
Lowest patch 5
Lowest patch exposure 1.5sec f/3.5
tc
- How can the exposure value for black be smaller than patch 15?
- I get different values for my Dimage7:
White: 1/125 sec f/3.5
Lowest patch 5
Lowest patch exposure 1.5sec f/3.5
tc
Bryan Biggers
Senior Member
My result is odd because my exposure for the patch (EV 1.6) is longer than the exposure for black. (EV3) Ummmm I don't know why this is, maybe I should repeat it. It is hard to meter the patch with spot, should you meter on the bright side or the dark side or in the middle. I used the middle.
Bryan
Bryan
Tom Crowning
Senior Member
I've metered the bright side because metering the dark side would be useless(same as black) and metering both would give the avaerage of both sides as result.
tc
tc
DavidP #28649
Forum Pro
Gotta love LCD panels.
Julian Loke83476
Active member
Hi Bryan,
For a valid result, you need to have
Black
Note that you do not have to be in focus to perform metering.
For most reliable results, I suggest that you jam your lens as close to the lens as you are physically able without damaging your lens or the monitor. It also helps to use a simple averaging mode, like centre-weighted, instead of spot or evaluative/matrix metering.
Julian Loke83476
Active member
And LCD screens on notebook computers too!
But are you sure that the fluorescent tube backlight does not have a finite refresh rate?
Julian Loke83476
Active member
Here is a collation of Mark's results:
D60 + EF 50mm f/1.8 II
White / Patch 23 / Black
EV 8.5 / 3.5 / 1.5
--
REBEL 2000 + EF 50mm f/1.8 II
White / Patch 18 / Black
EV 10 / 5.5 3.5
--
EOS IX + EF 50mm f/1.8 II
White / Patch 15 / Black
EV 10 / 5.5 / 3.5
If the same monitor and the same lens were used, then the rank order seems to be (in order of increasing sensitivity):
D60 .. REBEL 2000 .. EOS IX
And it seems that the D60 meter is biased by about 2EV, if it is metering the same patch (unless the meter was also reading the surrounding dark border).
D60 + EF 50mm f/1.8 II
White / Patch 23 / Black
EV 8.5 / 3.5 / 1.5
--
REBEL 2000 + EF 50mm f/1.8 II
White / Patch 18 / Black
EV 10 / 5.5 3.5
--
EOS IX + EF 50mm f/1.8 II
White / Patch 15 / Black
EV 10 / 5.5 / 3.5
If the same monitor and the same lens were used, then the rank order seems to be (in order of increasing sensitivity):
D60 .. REBEL 2000 .. EOS IX
And it seems that the D60 meter is biased by about 2EV, if it is metering the same patch (unless the meter was also reading the surrounding dark border).
Mike Price
Forum Enthusiast
Simple: the further you get from a source of light, the dimmer it is. To match image sizes with the same lens on the D60 you have to go further away.
EOS IX and Rebel should have been different. (The IX is APS format?)
Keyboard shortcuts
- f
- Forum