Panasonic Lumix DMC-FZ20

Image Stabilization Shootout: The Panasonic DMC-FZ15 vs the Konica Minolta DiMAGE Z3
Initially comparing the viewfinder display of the FZ15 with that of the Konica Minolta DiMAGE Z3 (a directly competing 4-megapixel long-zoom digicam, "KMZ3" for short from here on), I felt that the KMZ3 offered noticeably more effective anti-shake operation. I don't have a standardized test method for evaluating image stabilization effectiveness though, so it was hard to say definitively which system performed the best.

At the prodding of a number of staunch Panasonic fans though, I came up with a direct comparison between the FZ15 and KMZ3, that produced some very interesting results. First and foremost among these was the conclusion that, without absolutely rigorous controls to insure exactly consistent conditions, any sort of with/without sample pictures purporting to show the effects of image stabilization are misleading, to say the very least. (See the photos presented below for evidence of this.) Given the amount of interest that seems to have been raised about image-stabilization effectiveness, I decided to devote a fair bit of attention to the issue here, and will also copy this text into my review of the FZ3 (and eventually the FZ20, whenever Panasonic will condescend to send me a sample of that model), since all three of Panasonic's current FZ-series cameras use the same anti-shake technology.

To test the two cameras against each other, I physically strapped them together (as shown at right - ugly, but effective), and then shot a large number of images of text on a CRT, at the 12x maximum telephoto zoom setting, with a range of very slow shutter speeds (1/10-1/25 second), and in various combinations of the camera's stabilization modes.

With this setup, the amount of shaking the pair of cameras saw varied from shot to shot, but at any given moment, they would see essentially the same motion. Holding the cameras in both hands, it wasn't hard to press the two shutter buttons at nearly the same instant (I estimate that there was a good bit less than 1/10 second separation between the release of the two shutters), so in any pair of shots, the shaking seen by both cameras should have been more or less identical. (It should be noted though, that movement acting to rotate the pair of cameras about the shooting axis could result in one seeing more motion than the other, but inspection of the images that resulted seemed to show little or no rotational movement.)

One of the first things I observed when looking at the resulting images was just how much variation there was in the sharpness of the images from one shot to the next. I was deliberately shooting at very slow shutter speeds, to insure that I was stressing the cameras' stabilization systems enough to see noticeable differences between them, but the variation from shot to shot made it clear just how much the amount of shaking can vary over time with a handheld camera. Given this, it's clear that any A/B comparison showing image sharpness with/without a camera's stabilization system engaged will be completely meaningless unless you're capturing simultaneous frames with two cameras having the same optical characteristics and a common support structure, in the way that I was doing here. That is, you could get a sense of how much difference the FZ15's OIS made if you strapped together two FZ15's and captured simultaneous shots with them, as I did with the FZ15 and the KMZ3. Unless you used a calibrated "shaker table" though, images captured successively with the same camera would convey no meaningful information. - And then, even a calibrated shaker table wouldn't take into account phase and range-of-motion effects between the shaking and the camera's internal mechanics.

Here's an example of what I mean. The crops below are from shots captured at the same shutter speed (1/40 second) within a few seconds of each other, with image stabilization turned off on the FZ15. I tried to hold the camera as steadily as I could for all shots. As these images demonstrate, even though I was trying to be as consistent in my handling of the camera from shot to shot, there was nonetheless an enormous range of variation from image to image. This range of variation makes it clear that any sort of with/without examples of anti-shake effectiveness will be wildly misleading unless done with two identical cameras, being shot at the same moment.

Given the extent to which the camera shake varied between shots, it became clear that I needed to look at a large number of shots for each combination of test anti-shake mode and shutter speed, and evaluate the amount of blurring produced by both cameras statistically. I looked at pairs of images shot with both cameras side by side, scored them based on the amount of blurring I was seeing, and then compared the average scores across a minimum of 10 or more shots for each test condition.

This proved to be a pretty laborious process: In the process of sorting out the performance of the two cameras, I shot, inspected, and scored about 250 images with each model. At the end of the day, the performance of the two image stabilization systems turned out to be surprisingly similar to each other, but a few facts became evident. Here's a digest of what I found:

1) Any with/without stabilization comparison images published by anyone (myself included) are bound to be hugely misleading. Results can vary wildly from one shot to the next, depending on just exactly how the camera is moving at the moment of exposure, and where the mechanical elements of the anti-shake system happen to be in their range of travel. The only way I could come to firm conclusions about how the various IS modes worked relative to each other was by looking at the statistics across a large number of shots. (Since I was comparing the two IS systems to each other, and given the wide variation I was finding between shots, I didn't bother to shoot any non-IS images. Based on what I saw though, I venture to say that I could find "without" shots that looked better than "with" ones, just according to the luck of the draw.) - So anyone who's showing with/without example photos without very rigorous controls on how they're being shot (exactly simultaneous shots from a common platform, or calibrated, mechanical shaker table, for instance) is misleading you at the very best.

1) In Mode 1 operation (anti-shake enabled all the time), the KMZ3 did indeed beat the FZ15 most of the time, especially at slower shutter speeds.

2) In Mode 2 operation however (anti-shake only enabled when the shutter is tripped), the FZ15 beat the KMZ3 quite handily.

3) Running the KMZ3 in its equivalent of Mode 1 and the FZ15 in Mode 2 (that is, running each camera in the mode in which it performed best), the FZ15 generally edged the KMZ3 in image sharpness, albeit not by a huge margin.

Overall, it looks to me like the Panasonic approach to OIS can respond better to higher frequencies of vibration/shake, but it has a smaller range of motion available to its elements. Thus, if you leave it running all the time, the chances are much higher that it'll end up hitting the limits of its travel, and produce a blurry photo as a result. It's quick enough though, that it can respond while you're pressing the shutter down, in which case it's less likely to run up against its mechanical limits, and so is more likely to produce a sharp picture. By contrast, the KMZ3's anti-shake system seems to take longer to get in sync with the camera's motion, but appears to have a greater range of travel within its elements. This is part of why it can handle lower-frequency vibration better, and also why it does better in its equivalent of Mode 1.

Bottom line, I'd give the Panasonic IS system a slight but noticeable edge in performance, but I personally like being able to watch the stabilized LCD display, so I can pick the best moment to fire the shutter. Running in their respective optimum modes though, the differences between the two systems are largely academic: Either does so much better than an unstabilized lens that the differences between the two probably aren't all that important.

So how much of a difference does image stabilization make? As I noted earlier, the only truly valid way to know exactly what sort of vibration a given camera is compensating for would be to strap together two identical models, turn the image stabilization on in one and off in the other, and then capture simultaneous images from both cameras while they're being subjected to the same shaking.

Not having duplicate units of any of the image-stabilized digicams in question available to me, I couldn't perform this sort of a test with the full rigor it deserved. I did feel though, that I could give readers at least a general idea of how well Panasonic's OIS worked, by comparing unstabilized images shot with one camera with stabilized ones shot with another.

This was the genesis of the image crops you see here: I again used my lashup of the FZ15 and KMZ3, this time with the FZ15 set to Mode II stabilization, and the KMZ3 unstabilized. Both cameras were set to 1/40 second, full 12x zoom, and ISO 200. All images were subjected to an "auto levels" adjustment in Photoshop, so they'd appear with equal brightness. (Hence, don't look at these images to compare image noise, they simply aren't comparable in that respect.) Each pair of shots were captured at as close to the same instant as possible, by my estimate within less than 1/10 second of each other. Hence, both cameras saw more or less identical movement at the moment of exposure. Even though different cameras were used for the with/without comparisons here, the fact that the images were shot at the same focal length, the same shutter speed, and within a fraction of a second of each other makes the comparison entirely valid.

While there's some variation between shots here, the results are pretty dramatic: There's no question that image stabilization is a huge boon for long-zoom digital photography. Even the shots in which the OIS didn't completely eliminate the blur would probably be considered usable, particularly if they were the only shot that you had of some special moment. These shots were captured at 1/40 second, a shutter speed at which the FZ15's OIS system produced usable images about 90% of the time for me. This "usability" threshold will vary from user to user, and even with an individual shooter's physical condition. For instance, I have a much harder time holding the cameras steady when I'm physically tired, or (obviously) if I'm at all winded or my pulse is racing. - All this is to say that you should experiment a bit with you camera, taking a large number of stabilized shots at a range of shutter speeds, to get an idea of just where the exposure-time cutoff is for your particular capabilities and physical condition.