![]() |
EOS Documentation Project | ||||||||||
| Julian's Autofocus Sensitivity TestContentsIntroductionThe issue of low-light autofocus performance has been the subject of much recent debate amongst EOS camera users. Initial reports from early adopters of the EOS D30 digital camera and the EOS Elan 7E / 7 / 30 indicated that the low light AF performance was annoyingly worse than the official Canon specification. Some reports compare cameras at different market points, and attempt to draw a conclusion based on the speed of autofocus acquisition, or on the tracking of a moving subject. Sometimes, autofocus is assisted with a built-in lamp on the camera, or from a patterned beam emitted from an accessory flash device. Autofocus ability will not only depend on the brightness of the scene, but also on the pattern presented to the camera. It is even possible to fool the autofocus with patterns that cannot be focussed at all! MethodThis document presents an AF sensitivity target that can be used for objective measurement of the limit of autofocus sensitivity in EOS cameras in low light. The target consists of a web page with a series of rows of alternating light and dark patches. The camera should be able to autofocus onto the transition, but it will fail if the difference in brightness is too small for the AF sensor to detect. By metering the black/white/limit patches, it should be possible to compare the autofocus performance from different cameras. Perhaps those duelling EOS D30 and EOS D60 users, or the Elan 7E detractors, can finally settle their longstanding arguments. Or perhaps start new ones! | ||||||||||
| AF Sensitivity Test TargetThe test target is the sharp transition between the light patch on the left and the black patch on the right. Scroll up and down the page and test if the camera is able to lock onto the transition. Of course, the AF assist beam has to be disabled to make comparisons meaningful between different cameras. Record the spot or partial meter reading for BLACK, and the reading for WHITE, and the patch number, as well as the meter reading over the patch.
For instance, My Elan 7E is able to AF down to patch
number 65, with some hunting. You are welcome to enter the results from testing your own camera in the comments at the end of this page.
| ||||||||||
| Revision History
v0.4b2002Sep04 jul: minor reformat
Commentsblack : ev10White : ev2.0 lowest patch: ev2.5 hugoso 20/04/2002 19:44:55 Uh, Which camera? Which lens? Julian Loke 21/04/2002 01:46:21 How does the EV calculator work? Oly E-10, can focus on 5 but false locks on 4. F2.8/6" drg 21/04/2002 15:06:53 Black = 4"@f/4 White = 4"@f/5.6 Patch 36 = 4"@f/4.5 reliable on this patch Sigma 24mm f2.8 for Canon EOS Elan II ISO set at 100 focus at 1 foot I note it will not foucus on the horizontal bar below the patch. Julian, your EV calcualator would not work, stays at 0. Your test is a good idea, time will tell if variation in monitor contrast and brightness nullify the test, though the light meter readings should help evaluate this. Ken Newcombe 21/04/2002 15:16:40 Thank you for the feedback. The EV calculator has been updated to cope with an incompatiblity between Netscape 6.2 and dynamic scripting. Try again with version 0.3 Julian Loke 21/04/2002 18:16:24 Camera Canon D30 Lens Canon 50mm F1.8 ISO 100 No AF light Auto focus at patch #26 on the junction between white/black. Don Schultz 21/04/2002 19:49:50 Julian, It might help if you asked people to note which numbered pair they can just see a density difference on by eye. For example I could see the horizontal line on pair 11 but not 10. This would help compare monitor settings and make the comparisions more meaningfull. Ken Newcombe 22/04/2002 02:15:55 E10 black: 1.3" @ f2.4 white: 1/100 @ f2.4 patch: 2 DocDoc 22/04/2002 05:17:54 Canon EOS 30 (Elan 7E) EF50mm f/1.8 Mk II Focusing distance 1.5ft ISO manually set to 100. Using partial metering mode. Black: 2" f/1.8 = EV 0.5 White: 1" f/22 = EV 9 AF limit: Patch 24 1/3s @ f/1.8 = EV 3 Hunts, but reliably locks on every time. Julian, perhaps the lighting condition where the monitor is facing affects the contrast, thus the variation in the AF results. I'm sitting in a dark room as there's a thunder storm this afternoon. Thanks. BlueSky 23/04/2002 08:16:40 Elan 7, 24-85/3.5-4.5 @ 85: locks on to patch #30. D60, 24-85/3.5-4.5 @ ~50 (~85 35mm equiv.) locks on to patch #27. Central AF points for both. Strange. Alex Barnes 6/30/2002 2:09:17 PM Minolta 800si AF assist off Minolta 50mm f/1.8 Lock every time from "card" 25, can lock in the card 22. Exposure times: aprox, 1/20 at f/1.8. Obs: My monitor is not working fine, the blacks are not really blacks, in good conditions the camera will for sure do much better. Alexandre Escada 7/17/2002 11:58:22 PM Please note the different standart gamma for pc (2.2) and Mac (1.8) causing different contrast between patches. lufik 8/19/2002 2:33:11 PM Yes, monitor gamma will result in different patches for different monitors. But the intention is to use the same monitor to compare different cameras at the same sitting. The rank order of the AF sensitivity for different cameras will not vary with the monitor gamma. Julian Loke 9/2/2002 1:35:04 PM My EOS 300 would focus on patch 38 before hunting. I got readings of black 1/20f1.8,white 1/15 f 1.8 (%0mm f1.8 Mark II) Larry 9/22/2002 1:45:54 PM EOS1D -16mm-35mmF2,8L patch 32 no probs, 31 hunted a little but locked on, didnt go any further paul burgman 25/09/2002 09:34:48 EOS 30 with 28-135 f3.5/5.6 IS USM lenses. Down to patch 39, with manual setting ISO100, aperture 3.5 , 1" and 0.5" respectively for black and white, EV=4 and 3.5 Les Scholz 10/26/2002 11:15:52 AM Also, down to patch 29 with the monitor contrast turned up to max. Aperture 3.5 etc. Les Scholz 10/26/2002 11:21:15 AM EOS D30, Canon EF 50mm f/1.4 @ 2' AF lock (with some hunting): Patch 23 ISO 400, f/2.0 . Patch 23 = 1/2 sec, EV 1 . Black = 1 sec, EV 0 . White = 1/90 sec, EV 6.5 . (Patch 65 = 1/6 sec, EV 2.5 - for reference only) Monitor: iiyama VisionMaster Pro 17" CRT Video card: Matrox G550 OS: Win2K Pro Calibration/Profiling: Colorvision CRT/LCD Spyder, OptiCAL 3.7 - Gamma 2.2, 6451 K - Black Level: 0.30 cm^2 - White Intensity: 89.8 cm^2 WH Yi 11/14/2002 2:29:23 PM EOS 50E (Elan IIE) with 28-105 USMII easly goes down to 6-th frame (does hunting but shoots) white ev 10 black ev 2 Romanchuck J. Dimitry 1/27/2003 10:53:07 PM black: EV 1 white: EV 9 patch 48: EV 1.5 Elan 7 50/1.4 Seth 4/26/2003 8:12:07 AM eos 600 (630) ISO 100 ca 1.5 ft EF 50mm/f1.8 II Black 1.5" @ f2 = EV 1 White 1/45 @ f2 = EV 7.5 Focus limit at patch 50, 0,5" @ f2 = EV 3 EMK 08/05/2003 18:04:33 Canon 10D, Sigma 28-70 f/2.8. Hunts but eventually finds focus after repeated tries with patch 37. Monitor calibrated by OptiCal Spyder to gamma 2.2. EV -0.5. Adam 7/4/2003 10:12:37 AM I don't know if this is at all relevant for EOS cameras, but on a whim I tried this test out on my P&S digicam (I need something new badly) and got unusual results. It turns out, after a little experimentation, that it can reliably and correctly focus on the detail present in the subtly glowing pixels of the screen even when displaying a full-screen pure black document in Photoshop in a dark room. Bryce 11/15/2003 5:56:40 PM EOS 300D (Digital Rebel) EF 50mm f/1.8 (Original, Mark I) Evaluative Metering: 0.6 sec @ f/1.8 (ISO 100) Center focus point is enabled and used exclusively. The camera is is able to focus reliably between patches #5 and #6. | ||||||||||
© 2002 Sep 04 Julian Loke for EOS Documentation Project. All Rights Reserved. Redistribution of any EDP document, text, image, or code, or any work derived therefrom, in any form, electronic or printed, is not allowed without prior written permission from EDP.
Copyright notices, including this notice, must remain intact.
Use of this document constitutes agreement to our
terms and conditions. | |||||||||||