Panasonic 20mm f/1.7 Auto Focus

Started Feb 14, 2013 | Discussions thread
Anders W Forum Pro • Posts: 21,468
Re: Lots of talk about AF times but what about some measurements (part II)

The discussion in this thread prompted me to renew some of the tests of the AF performance of the Panasonic 20/1.7 in comparison with other lenses that I have already performed in the past (see here) and to execute and document them in an even more rigorous manner than I have hitherto found necessary. What follows below is first a description of my methods, then the results in numeric terms, and finally the conclusions I draw from them.

Camera: Olympus OM-D E-M5 with firmware 1.5 (the latest at the time of writing). The use of this camera is appropriate not only on the ground that it is currently one of the most popular as well as advanced MFT cameras but also because some posters have argued, in this thread as well as in others, that the AF performance of the 20/1.7 is worse on this camera and/or on Olympus cameras more generally than on recent Panasonic cameras.

Lenses: Apart from the 20/1.7 (with firmware 1.1), I chose to include three other lenses for comparison, namely, the Olympus 12/2 (firmware 1.0), the Olympus 45/1.8 (firmware 1.0), and the Panasonic 14-45/3.5-5.6 (firmware 1.2). The firmware of the lenses is in all cases the latest available at the time of writing.

The Olympus 12/2 and 45/1.8 are both fast primes, like the 20/1.7. Unlike the 20/1.7, however, they both have internal focus using Olympus MSC (Movie & Still Compatible) technology along with a reputation for being about as fast-focusing as it currently gets (both are often held out as opposites to the 20/1.7 in that regard) and without any known or rumored compatibility problems with the Olympus E-M5. In terms of focal length, they differ in that one is a fairly strong wide-angle and the other a short tele, with the 20/1.7 falling nicely inbetween.

The final lens, the Panasonic 14-45/3.5-5.6 is in several regards the opposite to the 12/2 and the 45/1.8. It is a slow (in terms of aperture), standard, "kit", Panazonic zoom rather than a fast Olympus prime. Further it is old (by MFT standards) rather than fairly recent. In fact, it is the oldest native MFT lens along with the 45-200/4-5.6, introduced with the very first MFT camera, the Panasonic G1, in the fall of 2008. Like the 12/2 and the 45/1.8, but unlike the 20/1.7, however it has internal focus. In fact, the 20/1.7 and the 17/2.8 are the only native MFT AF lenses that do not have internal focus and thus move the entire array of lens elements when they focus.

According to some posters, old and slow kit zooms like the 14-45/3.5-5.6 have significantly slower AF (in spite of their internal focus) than recent fast primes like the 12/2 and the 45/1.8, which is one of the reasons why I found it appropriate to include the 14-45 in the test.

Focus target, focus distance, and focus mode: As the focus target I used the brand logo on my freezer door. The focus distance for the 20/1.7 and the 14-45/3.5-5.6 which was tested at the same focal length as the 20, was 75 cm. For the other two lenses, I made sure that the magnification and framing was the same as for the 20 and 14-45. This in turn meant that I moved somewhat closer with the 12/2 and somewhat further away with the 45/1.8. All tests were performed with standard, single-area AF-S, with the AF-area at the center of the frame.

Sources of illumination and light level: The sources of illumination were a mix of halogen and fluorescent bulbs. Since it has been argued by some posters that low light makes the alleged difference in AF speed between the 20/1.7 and other lenses more prominent, I consciously chose about the lowest light level where fast AF might conceivably be of any importance. When metering off my white freezer door, I obtained an exposure of 1/60 and f/1.7 at ISO 1600. When replacing the door surface by a gray card, the metered shutter speed fell to 1/15. This in turn means that with the actual target (the door rather than the gray card), the light level was just above (about one stop) the point where the AF-assist-light goes on.

Tests: For each lens, I performed two tests. In the first, I simply refocused on the same target that the camera had already locked focus on in a previous trial. In the second, I started the AF process with focus set to infinity. The latter was accomplished by turning the camera off and then on again between trials, in which case the lens AF mechanism goes to infinity focus when the camera is turned on. Since I discovered that my 45/1.8 for some reason does not go to infinity focus when reset in this manner but to a fairly short distance, I manually refocused the lens between trials in that particular case.

The two tests combined span the range of focus distances in which fast AF might normally be of importance. The first test covers the common case of minimal change of focus distance between shots and the second one the case of significant change of focus distance from infinity to 75 cm or, in the case of the 12/2 and 45/1.8, the distance required for the same magnification and framing as for the two lenses shot at 20 mm.

Note that lenses with a short FL, like the 12/2, have to move the lens elements a shorter distance in/out to cover the same range of focus distances as a longer lens, like the 45/1.8. Since, however, I chose to shoot the lenses at the same magnification rather than at the same focus distance, this is compensated for by means of the test setup.

Note also that unlike the video comparison between the 20/1.7 and the 25/1.4 on which I comment in this post earlier in the thread or the test reported in this post, both of which, for reasons explained in the comments of mine that I link to, are only relevant under special circumstances were fast AF is rarely critical, my tests are chosen so as to be representative of normal usage were fast AF might be of importance.

Measurements: The camera was mounted on a tripod and each test repeated five times since AF times may vary slightly from one instance to the next even if the circumstances remain unchanged. While testing, I shot a video of the AF process as shown on the OLED by means of my iPhone 4, held on an L-bracket behind the camera. I then played back the video, frame by frame, and counted the number of frames from the point at which the AF process begins (as indicated by a change in brightness on the OLED) to the point at which the green indicator in the upper right corner of the OLED came on (indicating that the AF process has finished and focus been obtained). Finally, I averaged across the five trials and divided by the video frame rate the iPhone 4 had decided to use (in this case 23 fps or 27 fps depending on the brightness of the OLED with the particular lens tested) in order to obtain the average time in seconds. The video whereby I measured the performance of the 20/1.7 can be seen here as an additional documentation/illustration of my test procedures.

Results: The results obtained are as follows:

Panasonic 20/1.7

Test 1: 0.33 s

Test 2: 0.48 s

Panasonic 14-45/3.5-5.6

Test 1: 0.45 s

Test 2: 0.43 s

Olympus 12/2

Test 1: 0.41 s

Test 2: 0.49 s

Olympus 45/1.8

Test 1: 0.39 s

Test 2: 0.46 s

Conclusion: It is easily seen that the differences between tests as well as lenses are generally so small as to not be of much significance for most practical purposes. The shortest average recorded is 0.33 s (the 20/1.7 on test 1) and the longest 0.49 s (the 12/2 on test 2) but not too much can or should be made of this or other differences. As I pointed out, AF times may vary somewhat from one try to the next even if circumstances remain unchanged and if I had made more than five trials per lens and test, the exact times reported may well have been slightly different. The reason that I didn't extend the test to 50 trials instead of five is simply that the level of precision is already sufficient to prove the major point (not much of a difference) and that it is a quite time-consuming and boring task to play back videos one frame at a time and count the number of frames between certain critical events.

In essence, the results are the same as those obtained by Pekka Potka in this earlier comparison of the AF performance of the Olympus E-P3 versus the Olympus E-P2. While Potka's times are significantly longer than those I report, that's primarily because they include the time required for aiming the camera and actually firing it rather than the time required by the AF process alone. The important common denominator between his results and mine is that there is little difference between the AF performance of various lenses but considerable difference between the AF performance of older versus more recent bodies.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow