Olympus E1 SLRThe first "Four Thirds" system (almost) sees the light of day!
<<Viewfinder :(Previous) | (Next): Exposure & Flash>>
Page 5:OpticsReview First Posted: 06/24/2003, Updated: 03/16/2004
Learn how to use lens aperture to control depth of field - Visit our free Photo Lessons area!
Free Photo Lessons
Learn how to use lens aperture to control depth of field - Visit our free Photo Lessons area!
At the introduction of the E-1 camera, the Zuiko Digital lens system offered a variety of focal lengths, including 50mm and 300mm lengths, and two zoom lenses (14-54mm and 50-200mm). A 1.4x teleconverter is also available, and Olympus has plans to expand the range in the near future. Note that the 22.3mm diagonal dimension of the sensor translates into a 1.94x focal-length multiplier relative to 35mm cameras. The lenses mentioned above thus translate into 97mm, 582mm, 27-105mm, and 97-388mm equivalent focal lengths on a 35mm camera. In early 2004, Olympus introduced a 150mm f/2.0 lens and an 11-22mm f/2.8-3.5zoom.
Olympus has for some time insisted that the three-dimensional structure of CCD sensors demand a radically different lens design for optimum performance. Their E10 and E20 fixed-lens SLRs embodied such a design, in which additional optical elements collimated the light , insuring that light from the subject would strike the CCD surface perpendicularly across its entire surface. By contrast, with conventional lenses, light from the subject strikes the film or sensor plane at an increasingly oblique angle, as you move towards the edges of the image circle. (See the illustration above right, courtesy of Olympus.) Depending on the sensor design, this varying angle of incidence can cause problems in one of two ways. If the sensor employs microlenses to concentrate light on each pixel's active area, changes in the angle of incidence can lead to unwanted optical effects due to diffraction by the microlenses themselves.On the other hand, if no microlenses are used, collection efficiency is lower, and the decidedly three-dimensional structure of the CCD's surface can result in some of the light being shadowed from the active silicon surface by surrounding surface structures on the chip. Either case results in imperfect coupling of the light to the sensor elements.
In Olympus' "Digital Specific" lenses, an extra group of optical elements collimates the light (makes all the rays parallel), so it impinges on the CCD at right angles to its surface all across the frame.
Actually, Olympus has done a number of things in the design of their lenses for the Four Thirds system to improve image quality. To call attention to the extent of these design improvements, they've branded them as "Zuiko Digital (tm)" lenses. (I'm told that Zuiko means "Light of the Gods," presumably in Japanese.) In addition to the special "digital specific" design described above, Zuiko Digital lenses also incorporate improvements in lens molding and polishing accuracy, multi-coating, centering of the lens elements within the mounting system, increased use of ED glass and aspheric elements, and dual-sided aspheric elements, the latter of which Olympus claims as an industry exclusive. The actual impact of these enhancements remains to be seen (whenever I can test production models of both camera and lens), but the promise is that Zuiko Digital lenses will have better resolution, color rendering, and flare characteristics than even the best conventional designs. (The 14-54mm lens that shipped with my prototype sample of the E-1 indeed seemed to show very little distortion or chromatic aberration, particularly for a prototype-model lens.)
As I noted in the overview section of this review, Zuiko Digital lenses also support a greater degree of communication between lens and camera than has heretofore been the case, with some interesting consequences. Olympus claims that part of this increased communication will benefit autofocus speed and exposure determination, although they haven't yet said how this will work. It does appear though, that Zuiko Digital lenses pass information about their optical characteristics to the camera body, including information on geometric distortion. I first thought that the use of this information would be limited to storing it in the file's EXIF headers, facilitating automated correction of barrel and pincushion distortion post-exposure, on a host computer. It appears that the real story is quite a bit more dramatic though: The camera itself can perform these corrections, and apparently does so automatically. (This probably explains why I saw essentially zero barrel or pincushion distortion from the 14-54mm zoom that shipped with my evaluation unit.)
The in-camera correction capability goes beyond simple geometric distortion though. Many wide-angle lenses suffer from light falloff in the corners of the frame. This is a fairly natural consequence of lens design, and one that's very difficult to avoid in some lenses. (In the film world, certain ultra-wide lenses come with special graduated-density filters to compensate for this, and produce an even exposure across the frame.) Here again, Olympus has taken advantage of the E-1's inherent image-processing capability, through an option called "Shading Compensation." When enabled, this option tells the camera to use information communicated by the lens to correct the captured images for any light falloff resulting from the optics. This option can be enabled or disabled via a menu option. When enabled, it does add considerably to the processing time for each image. (When shooting a normal SHQ quality JPEG image, the camera finished writing to the memory card after about 2.5 seconds. With Shading Compensation enabled, the camera takes about 19.7 seconds to finish its processing.) Shading Compensation does not affect buffer capacity however: The number of shots you can take in rapid succession is the same whether Shading Compensation is on or off. (The only difference is how long it takes the camera to process the images and write them to memory.)
The Four Thirds lens mount itself is a bayonet design, similar in function to those employed on most removable-lens SLRs on the market today. A release button on the side of the lens unlocks it from the body, after which a partial twist removes it.
Focus can be automatically or manually controlled on the E-1, and a small switch on the front of the camera selects between Single, Continuous, and Manual focus modes. Single AF mode sets focus only when the Shutter button is halfway pressed, while Continuous AF mode continuously adjusts the focus (good for moving subjects). Manual focus mode lets you adjust the focus by turning the focus ring on the outside of the lens. (A menu setting on the camera lets you determine which rotational direction adjusts focus in or out, making it easier for users familiar with other manufacturers equipment to adapt to using the E-1.) The E-1's autofocus system judges focus based on a TTL phase difference detection system, and has a three-point AF area. The three points are spread across the center of the frame, with one point dead center, and the other two flanking it on either side. An adjustment button on the camera's back panel lets you select one of the three focus areas, or tell the camera to employ all three, choosing the one corresponding to the portion of the subject closest to the camera. The AE Lock button can be programmed to lock both focus and exposure when pressed, or to lock only focus. With only three focusing zones, the E-1's AF system is less sophisticated than those on some competing SLR models, but seemed to work very well in all my testing. (I don't have any good test to measure AF tracking speed with moving subjects, so that's a parameter we'll probably have to wait to hear from working pros about.) Also, while basic AF speed with vary greatly with the lens used and the subject being photographed, I generally found full-autofocus shutter lag times of between 0.2 and 0.4 seconds, quite fast indeed. (And prefocused shutter lag was a blazing 76.5 milliseconds.)
The E-1's specs state that it employs a phase-detect autofocus approach, versus the more common contrast-detect method. We'll have to see how this plays out on the production models: Phase-detect AF should theoretically result in faster AF response, because the phase-error signal can indicate not only that the lens is out of focus, but in which direction, and by approximately how much. Used properly, this "predictive" focus-error signal can be used to home in on the correct focus position more quickly than the magnitude-only signal from contrast-detect systems. In practice though, this isn't always the case. Phase-detect AF systems can easily be slower than contrast-detect ones, and may have either higher or lower light sensitivity. - Stay tuned in this important area for my tests of a production-level E-1.
A nice feature on the E-1 is the ability to combine autofocus with manual "fine tuning." A menu option labeled "S.AF + MF" stands for "single autofocus plus manual focus. In this mode, the camera will autofocus when you half-press the shutter button, but then transfer control to the manual focus ring, allowing you to make fine adjustments manually. This strikes me as potentially very handy, for situations where you'd like the camera to get you into the right focus range quickly, but then let you focus manually, to make small adjustments, or perhaps to track a moving subject. The way this works in practice is that, in single-AF mode, the E-1 adjusts the focus whenever you half-press the shutter button. If you hold down the shutter button after half-pressing it, the manual focus ring becomes active as soon as the camera has set the focus, allowing you to tweak the focus back and forth slightly. The manual focus ring is a "fly by wire" design though in that it doesn't connect to the lens elements mechanically, but rather just tells the camera in which direction and by how much to adjust the focus for you. I generally dislike fly-by-wire focus controls, as they don't provide the quick, sure operation that you get from a purely mechanical system. With the Zuiko lenses though, I found the focus control very responsive, at least for small "tweak" adjustments. - It's still slower than a manual control for large focus changes, but has a very good "feel" when making small movements.
The E-1 also lets you tell it whether to adhere to focus- or release-priority. In focus-priority mode, the shutter won't fire unless the subject is properly focused. Conversely, release-priority means that the shutter will fire whenever you tell it to, whether the subject is focused or not. In a nice touch, the E-1 lets you determine select focus or release priority independently for single-shot and continuous shooting modes. (I can imagine myself wanting to insist on focus priority for single shots, but preferring release priority for continuous shooting, to let the camera just take its best shot at tracking a moving subject, perhaps settling for slightly misfocused images, rather than missing the shot entirely.
Though aperture settings will differ depending on the lens in use, the E-1 does feature a Depth of Field Preview button that stops down the lens to the set aperture when pressed. Common on most high-end SLRs, this lets you focus with the lens wide open for a bright viewfinder image, and then preview depth of field by momentarily stopping down to the shooting aperture. The E-1 also features a bright AF assist lamp on the front of the camera, which lights in dark shooting conditions when the camera might otherwise not have enough light available to focus accurately.
Third-Party 4/3-System Lenses
One of the drawbacks to the E-1 system has been the high cost of the Olympux Zuiko Digital-Specific lenses. While of very high quality and not dramatically higher-priced than pro-grade lenses from manufacturers like Nikon and Canon, their cost could put the whole E-1 system out of reach for even well-heeled amateur photographers. (Or pros with limited budgets, for that matter.) For quite a while after the E-1's announcement and retail availability, there was no option in the marketplace for E-1 lenses other than Olympus' own offerings. Just as I was "going to press" with the production-level update of this review though, I received word that Sigma would be announcing 4/3-compatible lenses at CeBit. (This article will go up on our website slightly after the non-disclosure embargo on the Sigma lenses, so I could include mention of them here.) Sigma's announcement covers three lenses, an 18-50mm f/3.5-5.6 zoom, a 55-200mm f/4-5.6 zoom, and an 18-125mm(!) f/3.5-5.6 zoom. Pricing hasn't been set as of this writing, but Sigma is well-known for producing optically sharp lenses at very attractive prices. With an independent lens manufacturer now making lenses, the 4/3 system has taken a big step forward toward being a true multi-vendor standard. (All that's missing now is for another manufacturer to produce a 4/3 camera body.)
"Supersonic Wave Filter (tm)" Automatic Sensor Cleaning
Here's a feature that made me sit up and take notice: Built-in ultrasonic sensor cleaning! Here again, it remains to be seen just how well this will work, but the concept sounds promising. Dust has proven to be a bane for digital SLR users from the beginning. In film cameras, the imaging surface (the film) is constantly refreshed as each new frame is advanced. Any dust that might accumulate on one frame will thus not affect subsequent ones. In digital SLRs though, the sensor surface is fixed, so any dust falling on it tends to stay there, the surface becoming increasingly dirty over time. Various accessories are available to clean CCD surfaces, but their use presents an ongoing risk of accident. (That is, while the cleaning gadgets themselves may be perfectly safe, every time you open your SLR and start sticking things inside the camera body, there's a finite risk that you'll do something stupid and damage the sensor chip.)
Olympus has apparently developed new technology by which the E-1 can clean its sensor itself. Details are sketchy at this point, but it appears that every time the unit is turned on (or commanded to do so via a separate menu setting), an ultrasonic system activates, dislodging any dust particles that may have settled on the sensor's surface. (Dislodged dust is collected and trapped in an internal receptacle, so it won't float around the mirror compartment to cause more problems down the line.) It's not clear whether the ultrasonic system works by vibrating the sensor itself, or if it instead directs an airflow over the sensor, but Olympus claims that it's quite effective, and that a full cleaning cycle takes only 200 milliseconds. (0.2 seconds) I don't have any way to objectively measure the effectiveness of this system, but can say that I saw virtually no evidence of dust on the sensor throughout my testing and use of the E-1.
To set appropriate expectations for Olympus' Supersonic Wave Filter system, it's important to note that it almost certainly won't be effective against grease smudges caused by fingerprints. - So continue to be careful about putting your fingers inside the mirror compartment, when the sensor is exposed.
The sensor chip used in the E-1 calls for special comment as well, although the test results I obtained from a production-level camera lead me to wonder slightly whether the special attention is in fact deserved. Its claims to fame should be lower noise and increased dynamic range, but there are a lot of system-level factors that can affect noise levels and dynamic range, regardless of sensor characteristics.
The Four Thirds initiative is a joint effort by three companies: Olympus, Kodak, and Fuji. We haven't heard anything about Fuji's possible plans yet, but Kodak will clearly be a major partner of Olympus in the E-1, as it's one of their sensors that will be used in the camera. Kodak was a dominant player in the early digital SLR market, thanks largely to their advanced sensor technology. Now, with the advent of Four Thirds and their participation in the E-1 with Olympus, they appear poised to regain significant market share for their chips. While Kodak has recently struggled in the SLR marketplace, their CCD sensor technology has historically been second to none: Kodak's specs for quantum efficiency, electron capacity, and thermal noise levels are thoroughly state of the art. They also have a very well-developed design base and semiconductor manufacturing process for creating "frame-transfer" CCDs, which have considerable inherent advantages over the more common interline-transfer designs used in most digital cameras currently on the market.
While considerably more difficult to manufacture than interline sensors, the frame-transfer design potentially provides better light sensitivity and a significantly improved signal to noise ratio. This is because almost 100% of the silicon's surface area is available for light collection, since the structures used to transfer charge out of the photodiodes and off-chip are buried beneath the photodiodes themselves. By contrast, in an interline-transfer CCD design, the charge-transfer registers are located alongside the photodiodes, consuming considerable silicon real estate. (This also means that frame-tranfer CCDs don't need the "microlenses" commonly used with interline-transfer chips to improve light collection efficiency. Dispensing with these optical structures on the CCD's surface can further improve efficiency (since the microlenses themselves aren't perfectly transmissive), and reduce other undesirable effects such as diffraction.
The electronic structure of frame-transfer CCDs also results in a much higher "saturation voltage" than that of equivalent interline-transfer designs. Combined with the low thermal noise that characterizes Kodak's chips, the overall result is that the CCD used in the E-1 should have nearly twice the dynamic range of competing interline-transfer units. (Dynamic range is the range of light to dark values that can accurately be recorded.)
As I noted above, image noise and dynamic range are also strong functions of the design of the rest of the camera's electronics, but the Kodak frame-transfer sensor used in the E-1 should at least give it a head start in this regard. - Stay tuned for the final word on all this though, once I can test a production-level E-1. Note that as of this writing in early September, 2003, Olympus has told me that none of the evaluation units thus far released to reviewers have any noise-suppression processing implemented in them. Any image-noise data published as of this time thus will not be representative of the results from final production models, so I've elected to refrain from any evaluation or comment at this point on image noise. (It is indeed higher than average for a high-end D-SLR, but that would certainly be expected if there's no noise-suppression processing being applied.)
Lest this sound like a completely uncritical paean to frame-transfer sensors though, I should hasten to point out that they do have one significant limitation relative to interline-transfer designs - Their charge collection can't be electronically "gated" to produce the ultra-fast exposure times of some consumer-level cameras. In the case of the E-1, this shouldn't be much of a limitation though, since the camera's shutter can produce minimum exposure times of 1/4000 second on its own.
The above-mentioned theoretical considerations aside though, my testing of a production-level E-1 revealed that its image noise was actually somewhat higher than most competing d-SLR models (and greatly higher at high ISO settings), rather than lower as one would expect, given its use of frame-transfer CCD technology. Also, while it seemed to show good dynamic range, I don't know that I'd say it was significantly better than much of its competition. All I can conclude is that either the particular sensor chip it uses has noise characteristics that fall somewhat short of the promise of frame-transfer technology, or that the analog and A/D electronics inside the E-1 have higher noise levels than they might. Whatever the case, the E-1 hasn't thus far lived up to the promise of the sensor technology it uses.
Reader Comments! --> Visit our discussion forum for the Olympus E1 SLR!