Canon EOS-D30 Review
Noiseless "grain free" images
The first thing that strikes you about the D30's images is their virtual lack of any noise (obviously some becomes apparent at higher ISO's but still impressive), zoom in as far as you like, all you'll find are smooth gradients and sharp details. The next thing you'll notice is a lack of any kind of sharpening artifacts; jagged lines or black / white halo's, that's because the D30 has very little internal sharpening (or if any it's very mild), that produces these beautiful clean images, so noiseless in fact that in this respect I can't really find another camera to compare to the D30.
The two samples below each have three areas cropped then magnified 200%, just look at that smooth clean detail with virtually no noise (ISO 100).
|Lens used: Canon EF 28 - 70 mm F2.8 L|
Overall the D30's colour rendition was very accurate, white balance was often hard to fool (although it's always good practice to use a preset where possible, to avoid "white balance drift"). The D30 features three colour settings (examined earlier in this review) at the default "Normal" setting colours are neutral and good for printing, the "High" setting producing the most pleasing colours on an average monitor (for the web) deep colour without over saturation or loss of detail.
|Lens used: Canon EF 24-85 mm F3.5 - F4.5|
Although good, colours were still not quite as good as the Fujifilm S1 Pro (our current top "colour demon"), the D30 was occasionally shy of reds with the odd hue shift in over saturated rich reds (normally corrected with a -0.3 EV compensation before exposure). That said they were more than impressive, combined with the smooth detail the results are very pleasing.
Resolution & Sharpness
It's deceiving, you'd think by looking at the D30's images that Canon are doing something special to "clean" the images (I'm not talking about noise reduction which only kicks in for >1 second exposures), when you get in close to a D30 image it looks so clean that you're sure some detail must have been lost to produce such smooth gradients. The second thing that would enhance this feeling is the relative lack of internal sharpening applied, this cuts down any sharpening artifacts but can leave the images looking soft to the untrained eye.
However, I'm glad to report that the D30 (3.1 megapixel) exhibits more resolution than Nikon's D1 (to be expected at just 2.74 megapixels) and more than our current to pick of the prosumer digital cameras, the Nikon Coolpix 990 (3.1 megapixel), as you'll see when we compare resolution charts later in this review.
This means that the images are not only incredibly clean but still full of detail and resolution. And I personally prefer Canon's approach of not over sharpening images in-camera but giving you the option (and the clean image) to sharpen later if preferred (though I also acknowledge that some users would prefer slightly sharper images straight out of camera, perhaps we could see an improved selection of sharpening settings in the next firmware?).
|Lens used: Canon EF 24-85 mm F3.5 - F4.5|
Dynamic range simply defines the range of light the camera is able to capture before it either loses detail in darkness (shadows for example) or blows out a highlight (edges of chromed metals are good examples of this). Most consumer digital cameras only have a 8-bit analog to digital converters, plus their CCD's are not built to have a particularly large dynamic range. The D30's CMOS sensor has a 12-bit analog to digital converter.
Metering was very good, you have the choice of three different metering modes, Canon's "Evaluative metering" produced very similar results to Matrix metering seen on other cameras, I found the camera did have a tendency to over expose, because of the amount of shadow detail captured by the D30 it wasn't a problem to simply dial in a -0.3EV compensation to alleviate this (something I'd recommend doing if you're shooting in bright sunlight).
Now the bad news, it's a real shame that Canon haven't spent longer on their TWAIN acquire module, the RAW format out of the D30 shows a lot of promise, acquire an image at 16-bits Linear (no processing) and the amount of extra dynamic range above what you get out of a normal JPEG is obvious to see, however the linear image takes A LOT of work to get it looking anything like the scene. I've no idea WHY Canon haven't got a way to perform exposure compensation on the RAW image BEFORE acquiring it. The second problem with the acquire module is its speed, taking between 25 and 40 seconds to acquire a single image is VERY slow, it's the main reason I didn't shoot more in RAW, I simply couldn't face sitting and waiting so long for each image. Lets hope that either (a) Canon come up with an upgrade acquire module or (b) a third party developer decodes Canon's new RAW format and develops a better processing tool.
|Lens used: Canon EF 28 - 70 mm F2.8 L||Lens used: Canon EF 24-85 mm F3.5 - F4.5|
Night photography is a real possibility with the D30, very little noise in exposures up to 10 seconds, enable noise reduction and you can go further still. Two samples below both shot without and with noise reduction.
|6 sec, F4.5, No Noise Reduction|
|6 sec, F4.5, With Noise Reduction|
|15 sec, F11, No Noise Reduction|
|15 sec, F11, With Noise Reduction|
As expected the D30's auto whitebalance was fairly good, though it did struggle under incandescent light. As always the best and most accurate option being taking a manual reading from an existing image. Samples below shot at 1440 x 960.
"Weird Dots" on resolution chart
This is something very specific which only seems to occur when shooting our standard ISO 12233 resolution chart. At 700 and 800 lines/picture height frequencies some random dots seem to make their way into the test pattern. These look like what we'd expect to see from "hot pixels", however they only appear at the above mentioned frequency and the exposure is only 1/3s (not long enough for hot pixels). Thanks to Mike Chaney for noticing these which has enabled me to update the review.
Strange, right? They don't appear at lower or higher frequencies, just 700 and 800 lines/picture height and always the angled lines... What's even stranger is that close examination of "every day" shots shows they don't exhibit this phenomenon which kind of indicates it could be related to the consistent frequency of the test chart. Whatever the problem it appears only to be isolated to these frequencies for this specific shot, we've scanned and re-scanned through all the samples we've shot and can't see a single example of this in a real-life shot.