A working photojournalist's digital camera needs to be many things. Three of the most important attributes are:
- Good image quality at all ISO settings
- Reliable flash exposures
- Fast, accurate autofocus
In writing Canon EOS D30 shows promise back in July, these were the three things that couldn't be assessed, since Canon had only pre-production cameras to show at that time. After 10 days of steady D30 shooting, however, a clearer picture of the D30 is emerging: D30 colour is nothing short of stunning in most situations, and both single and multiple 550EX flash photography is markedly more reliable than with the DCS 520/D2000. Serious sports shooters, however, will find the camera's autofocus system, and perhaps shutter lag and buffer operation too, hinders the making of good peak action sports photos.
Canon EOS D30 with the Battery Grip BG-ED3 attached
Note: this article includes 14 D30 photos. Most may be viewed larger in your web browser by clicking zoom; the original image may be downloaded via the full-res link. Each original image contains shooting information in the IPTC Caption field.
When it comes to digital SLR colour, the D30 is my new best friend. As long as the appropriate white balance (WB) setting is selected, D30 photos easily eclipse the colour coming from the D1 and even Kodak DCS/Canon D2000 cameras in both good and not-so-good light. The two photos below are good examples of the camera's range.
Left: ISO 100, Multiple wireless 550EX flash (zoom)
Right: ISO 1600, dim available stadium light (zoom|full-res)
The frame of my son Fergus at left (my favourite crash test dummy for evaluating image quality currently) was shot on the Custom White Balance setting, while Auto White Balance was set for the frame at right. Both show vibrant overall colour for the shooting situation, pleasing skin tones and few colour oddities, except perhaps for a slight tendency to turn hot highlight areas slightly pink.
Window light, cool white fluorescent, household incandescent, the D30 seems to handle it all with flair. It's Custom White Balance setting doesn't drag every colour in the scene into line as effectively as the D1's Preset does. That capability remains the D1's chief image quality strength. Instead, the D30's Custom White Balance does display some variation in colour depending on the colour temperature and colour characteristics of the light source. But the variations are minor, and the results almost always pleasing and not overly difficult to tune for printing. In short, if good colour matters in a variety of shooting environments, the D30 is going to be hard to beat.
ISO 400, window light, Cloudy WB, photo brightened in Photoshop
to compensate for D30 metering error (zoom|full-res)
Left: ISO 400, cool white fluorescent, Custom WB (zoom|full-res)
Right: ISO 1600, 60 watt household lamp, Custom WB (zoom|full-res)
As you can see, there is some variation in skin tone in the three photos above, where the primary difference is in the reds, or pinks to be exact. Photos shot under warmer- than-daylight conditions show a consistency tendency to shift a narrow portion of the red spectrum to oversaturated pink. It's visible in the too-pink lips of the fluorescent and incandescent frames (view the enlarged version of each to see what I mean). Call it Magenta Cast Lite if you like, because it's so far the single biggest consistent colour problem that I can see. And calling it a problem is a stretch.
Properly exposed images at ISO 100 are effectively free of noise. In fact, it's not until about ISO 400 that the visibility of digital camera noise begins to intrude into the shadows of properly exposed frames at all. At ISO 800, properly exposed images show a remarkable lack of noise, and at ISO 1600, the D30's noise levels are noticeably less than the D1, DCS 520/D2000 and DCS 620. Short of using a DCS 620x, you'll be hard pressed to produce cleaner digital camera images on the printed page.
The noise is almost exclusively in the image's chrominance. That means that Quantum Mechanic Lite and Quantum Mechanic Pro can, and do, filter it out. As an added bonus, the noise is often pleasing in appearance. Huh? How can noise be pleasing? When it has the same random, scattered appearance as film grain, and when it forms a part of the rich out-of-focus backgrounds that the D30 often records.
Left: ISO 800, daylight + stadium lights, Auto WB (zoom|full-res)
Right: ISO 1600, dim available stadium light, Auto WB (zoom|full-res)
Left: ISO 800, mystery light source, Custom WB (zoom|full-res)
Right: ISO 200, daylight, Auto WB (zoom|full-res)
Detail in D30 photos is impressive. Small text, sports jersey netting, eyelashes and other fine edges are distinct at 100% magnification in Photoshop. The only problem is that even High in-camera sharpening is slightly too weak, if one understands the purpose of this sharpening. Namely, it's to restore fine detail lost to the camera's optical low-pass filter. This optical filter is necessary to ensure the best colour and to limit digital artifacting, but it does take a toll on the crispness of hard edges. Fortunately, the D30 records fine detail with such smoothness and clarity that the detail sharpens back with little drama. Once sharpening has been applied to restore fine detail, the photo is in an optimum state to be resampled and sharpened for printing.
I'll stop short of a full explanation of this sharpening workflow, because the point I really want to make is this: for D30 photos to print with maximum sharpness, fine detail has to first be restored. And the in-camera sharpening, even on High, falls short of the mark. Still, High is close enough that in most workflows it shouldn't make an appreciable difference and seems safe to use at even higher ISOs. That's because the kind of noise in D30 files is not overly accentuated by High in-camera sharpening. Since sharpening of JPEGs can't be turned off in the camera, all this is a good thing. I'll have more on image sharpening strategies, as well as a look at some minor D30 image oddities, in a complete D30 review later this month.
When I conduct training sessions on the DCS 520/D2000, I devote a large chunk of time to the 550EX flash and its Evaluative TTL (E-TTL) view of the world. That's because the 550EX is the best available option for that camera. And also because it produces inconsistent exposures in some situations in which it really shouldn't.
The 550EX problems of the DCS 520/D2000 seem all but eliminated with the D30. I'm a flash guy, and have been waiting for a digital SLR to deliver the same level of goof-proof TTL flash photography that I associate with Nikon film SLRs especially. So far, the 550EX mated to the D30 is a whole lot closer to flash nirvana than I expected, which has been a pleasant surprise. Whether the flash is on the camera or at the end of the Canon Off Camera Shoe Cord 2, pointed straight ahead or bounced, flash output is reasonably consistent and reasonably close to the mark every time. The only major quirk has been when shooting in flat light: sometimes, the flash needs a +0.7 or more boost. It's not a coincidence I'm sure that the ambient metering system seems to require the same compensation bump to prevent underexposure in similar flat or front lit situations. Prior to the D30, I can't think of a shooting situation in which I've applied a +0.7 or +1.0 bump to 550EX flash output.
Multiple 550EX flash photography is downright fun with the D30. Two and three 550EX strobes, triggered by an ST-E2 transmitter at the camera, sing in harmony like never before with a Canon-body digital SLR. As is obvious, I shoot a lot of photos of my son. But I've never shot so many of him cruising about the livingroom with two and three wirelessly-controlled TTL flashes lighting his way. Even with one 550EX in a softbox (as is the case in the Superman photo below), flash exposures so far have been consistent. The D30 also supports the modeling lamp and ratio control features of Canon's wireless flash system.
Left: ISO 100, two 550EX strobes, Auto WB, RAW format (zoom|full-res JPEG)
Right: ISO 100, two 550EX strobes, Custom WB, RAW format (zoom|full-res JPEG)
To get decent flash colour, however, has meant using either Auto WB, which balances flash-lit pictures well much of the time (above, left), or Custom WB when it doesn't (above, right). The Flash WB setting isn't even close for some reason: photos have a strong yellowy-green cast. I'm anxious to try another D30 body, should the problem be a calibration hiccup in my early production model from Canon.
The 550EX in conjunction with the D30 even seems to handle long lens high school football:
ISO 1600, 550EX flash, Auto WB (zoom|full-res)
The camera has many other strengths, including a compact and seemingly rugged body, well-designed menus on the rear LCD and a number of clever Canon touches that will make a professional photographer feel right at home with the D30.
Autofocus, shutter lag and the "Buffer Problem"
Image quality and flash performance are there. This has made it all the more frustrating for me that the D30's peak action shooting capabilities are sub-par. Autofocus, shutter lag and a mysterious pause between bursts of several frames have so far limited both myself and other early D30 shooters from covering sports events properly.
The camera contains the same autofocus module from the Elan II, which may explain why the camera's autofocus performance comes up short. Last weekend I shot night high school football, university basketball and afternoon Canadian Football League action. With the exception of the first minutes of afternoon football, which were played under bright sunshine, the weekend was an autofocus wipeout. Frame after frame of decent moments were just not sharp.
So poor was my take from the weekend that I spent Monday morning locked to the phone, commiserating with D30-toting colleagues around North America about my autofocus experience, trying to gauge whether they were having better luck. It quickly became clear that they were not. San Francisco Chronicle photographer Michael Macor's experience shooting the 49ers Sunday afternoon mirrored my own. In fact, when he checked in with the editor going over his D30 photos from the first half, Macor was told that while he had a number of photos that were close, there was virtually nothing that was sharp enough for the paper. Macor abandoned the D30 and shot the remainder of the game on a Canon D2000.
ISO 200, AI Servo, photo front-focused (100% magnification zoom)
(photo by San Franciso Chronicle/Michael Macor)
There are four things that affect the speed and accuracy of Canon's autofocus system (and probably that of other manufacturers too): subject contrast, light level, the focus point selected and whether the user selects the autofocus point or the camera does. Macor set the D30 so that it would dynamically select one of the camera's three focus points, which may have contributed to the camera's focusing problem. But manually selecting a focus point is by no means a cure. As an early adopter of the Canon EOS system, I learned that keeping a Canon autofocus SLR camera manually set to the centre focus point translates into the best possible autofocus performance. And that's how I shot the D30 last weekend. The result, as I mentioned, was many more blurry photos that I'm accustomed to from Canon gear, even when compared to a camera like the midrange Canon A2.
The only respite from this was in the first quarter or so of afternoon football, where the light level and subject contrast was enough to give the autofocus system a fighting chance. By comparison, shooting indoor basketball under the flat and comparatively dim light of the University of Calgary's main gymnasium led to some nice moments that were simply not usable. I switched to the D1 for the last few minutes of basketball on the D1. There were a number of tack sharp frames to choose from.
USA Today photographer Bert Hanashiro introduced me to a new term this week: "EOS Moment." It refers to the ability of Canon's best autofocusing cameras to get seemingly impossible peak action sports frames sharp, and especially the first frame, as the ball tickles the receiver's fingers and that kind of thing. Except perhaps in bright, contrasty light, don't expect the D30 to deliver as many EOS Moments as the best Canon pro cameras.
Shutter lag, the time between when the button is depressed to when the shutter trips, seems longer than it should be as well. Ultimately I've been able to tune myself to the D30's shutter button, and as I scan over my photos from the weekend there are plenty with the ball in the frame. But I was conscious of having to squeeze the button well before the peak moment, or the basketball or football would be nowhere to be seen (photo below). Compared to the D1, which I've been shooting sports with for the past year, the D30 seems slower to trip the shutter when the button is pressed.
ISO 400, overcast, Auto WB (zoom|full-res)
The D30 may very well boast a shutter lag time that's comparable to the D1 and pro Canon and Nikon film cameras. But the shutter button's design may be masking that fact. Cameras like the EOS-1v, for example, have a shutter button that places increasing back pressure on the user's finger as the button travels downward from the midpoint. This makes it possible to find the sweet spot in the shutter button's range, just shy of tripping the shutter. The D30's shutter button does not allow for the shutter button to be squeezed past the midpoint in the same manner, effectively increasing the distance the shutter button has to travel when depressed to take a picture. This explanation is convoluted, I know: you probably have to shoot with the D30 to follow my point.
Perhaps the most perplexing D30 trait of all is what will hereafter be known as the "Buffer Problem." Sometimes, when shooting a burst of a few frames and then stopping for a second or so as the play unfolds, the camera will not fire right away when the shutter button is pressed again. The pause the camera inserts can be infuriatingly long. This behaviour would be expected when the buffer completely fills, but that's not the case here. In fact, for me it seems to happen if I break after shooting exactly three frames, then try to squeeze off another a moment later. It's as if the camera interprets the break in shooting as an opportunity to get busy processing and writing images, and then protests as its being yanked back to firing the camera. If I simply hold down the shutter button without a break, the camera will fire continuously without the pause, at least until the camera can't hold any more images in its processing queue.
The Microdrive may contribute to the problem, in that during the unwelcome pause I can hear its tiny mechanism spinning up. USA Today's Bob Deutsch reports that a colleague at the World Series in New York was fit to be tied when the D30 repeatedly did the same thing, and she had a Microdrive loaded in the camera. I can't replicate the problem with a Lexar Media Flash RAM card, though freelancer photographer Greg Fulmes indicates that he has experienced the pause even with a Kingston 96MB Flash RAM card in his D30. So, the Buffer Problem may be present with Flash RAM cards too but is exacerbated by the spin-up time of the Microdrive. Or it may be caused by something else entirely. I really don't know at this point.
Canon USA's Chuck Westfall indicates that they are well aware from early D30 owners of the nature of the problems described here, and are communicating with Canon in Tokyo about the matter. There is so much to like about the D30 that I hope solutions for these problems are forthcoming, though I'll confess to not being optimistic. Sluggish autofocus, the most significant of the three problems, is not normally the kind of thing that can be improved through a firmware update.
On a related note, Canon's specifications for the number of frames that may be shot in a sequence now makes sense. As many shooters have discovered, sometimes the camera will shoot many more than the 8 Large Fine JPEGs the manual states is possible. And sometimes it won't. The total number of frames is directly related to the size of the JPEG once its compressed. Since noisy images compress less than clean images, the camera will shoot fewer frames in a burst at higher ISOs than lower ISOs. I've been able to fire off 31 consecutive frames of a blank wall at ISO 100. While shooting night high school football at ISO 1600, however, the maximum was consistently 9-10 frames. It seems that Canon's 8 frame burst spec is the minimum number of frames one can expect in a burst. When shooting at higher ISO settings in particular, then, its important to not get too trigger happy or the image processing queue may fill before the play is done.
The D30 is ready to go for all sorts of news jobs. In fact, a typical newspaper photographer's daily diet is made up of large quantities of static or slow moving assignments, for which the D30 is a perfectly good match. Its images approach the vibrancy of the best slide films, flash photography is effortless and handling the camera is a pleasure. If you spend your shooting day making environmental portraits, the D30 will not disappoint. But if you need one camera to do that and cover peak action sports, and you're wedded to the Canon platform, the DCS 520/D2000 or, presumably, Canon's upcoming pro digital SLR, will be a better choice.