Five E-System DSLRs Compared

Five E-System DSLRs
Compared by RAW Results Using Silkypix Developer Studio 3

Olympus E-1 E-300 E-330 E-400 E-510

After the Olympus E-510 came to market in 2007, there were particularly two negative issues pointed out by reviewers and critics: highlights in pictures turning pink when brightness is reduced in the RAW converter, and a limited dynamic range which makes the camera comparatively prone to clipped highlights. In October 2007, when I slowly became interested in the E-510 because of its newly introduced in-body image stabilizer, I had the chance to borrow an E-510 and so I set out to explore how severely those issues would affect my pictures, which I usually shoot in RAW format and develop with Silkypix Developer Studio. After the first tests, I was sufficiently convinced to buy one for myself, and since I still had an E-300, an E-330 and an E-400 and I could lay hands on an E-1 as well, I decided to take the started comparison a little bit further, seizing the opportunity to compare all image sensors Olympus had employed in an E-System DSLRs until then:

Fortunately, the pink highlights problem was already resolved at that time through updates for most RAW converters including Silkypix, so I could have restricted myself to the issue of clipping highlights – but then I thought better of it and took the time to find out how the five cameras compare in terms of noise and colour reproduction as well, as far as my humble 'test environment' allowed. Here are the results, of which I will mainly discuss the the dynamic range aspects in more detail:

1. ISO performance
   ISO performance comparison from ISO 100 to 1600, part 1 unsharpened and without noise reduction, part 2 with individually optimized sharpening and noise reduction parameters for each ISO step.

   What do we see there? We see what everyone knows, that more megapixels mean more detail resolution, and more noise means less detail preservation. We see that the detail resolution disadvantages of the 5 megapixel E-1 were already significant compared to its early 8 megapixel follow-up, the E-300. We also see that really bad noise can outweigh high megapixel counts, and that we don't have to look at point-and-shoot type cameras for proof: Above ISO 400, the 7.5 megapixel E-330 actually preserves more detail than the 10 megapixel E-400. And we see that, noisewise, the new NMOS type sensors by Panasonic perform better than any of the Kodak CCDs used by Olympus so far. And we find that the winner of a strict low noise competition disregarding detail resolution would be the E-330, with the E-510 coming off second-best, while, probably the most relevant result, the actual winner in final detail preservation through both pixel count and noise clearly is the E-510, with E-330 and E-300 competing for second-best. Though showing more noise, the E-300 seems to deliver slightly sharper detail than the E-330. While the pixel count difference ought to be insignificant, it may well be the E-300 used to have the less aggressive antialiasing filter.

2. Highlight range by example, part 1
   Test shots of a contrasty scene, correct exposure leaning to 'just under', already showing a slight tendency to overexposed highlights.

3. Highlight range by example, part 2
   Test shots of the same scene, slightly longer exposed, still barely good exposure for the main scene, more significant overexposure of highlights.

4. Highlight range by example, part 3
   Test shots of the same scene, clearly overexposed (+1.3 EV), extreme overexposure of highlights.

   At least the E-510 with its increased tendency to blow highlights shows a significant chance to recover detail through the use of RAW, here. But even if we do have more highlight range in RAW – all in all, we see here that, with any of these cameras, there's only little to repair even with the help of RAW and the so-called RAW headroom, once a picture is thoroughly overexposed and in need of a substantial amount of highlights to be 'brought back'. Especially when the colour of lost details is anything but grey, because as soon as one of the three colour channels clips, the best the converter can do is to render the recovered details greyscaled.

   The slightly darker results from the E-1, by the way, do not come from a greater highlight range, but from a slightly lower effective sensitivity, which could not be sensibly corrected for this comparison, because the brightness difference to the other cameras must have been just about inbetween two 1/3 EV stops. More on this and on dynamic range follows right below.

5. Dynamic range in detail
   Differences in gradation and dynamic range visualized in curves, with a detailed discussion.

   In conclusion, regarding dynamic range, at least as long as we're disregarding all aspects of shadow range, there seem to be no differences worth mentioning between the five examined cameras when shooting RAW and converting with Silkypix. There's a slight highlight range advantage over the out-of-camera JPEG for all these E-System cameras, even if we're simply using Silkypix with default settings, and there's a really significant advantage for the E-510, the one camera out of the five which does indeed leave something to be desired in terms of JPEG dynamic range.

   On a side note, it looks like the E-1's larger pixel elements and its nominally higher signal-to-noise ratio, as Kodak officially states it in comparison to the E-300 sensor, do not seem to be of too much practical value here. In practice, the E-1 does not seem to deliver any more dynamic range than any of the later models with significantly more megapixels and accordingly smaller pixel pitch.

6. Colours by example, part 1
   Artificial colours, subdued (coloured sheets of paper), neon (markers), and strong (crayons).

7. Colours by example, part 2
   Natural colours (a fruit and vegetable display).

   While these examples surely do not cover every aspect of colour reproduction: Whoever might have expected vast differences here, maybe that the widely hyped Kodak FFT CCDs might somehow vastly outperform the newer generation of Panasonic NMOS type sensors, is bound to become disappointed. All five cameras seem to have quite similar characteristics in colour reproduction, which only means that Ichikawa Soft has done good work creating and adjusting their camera profiles. What differences there are to observe, are the somewhat less vivid colours of the E-300 and E-400, especially in orange and red tones, and the slight greenish cast of the E-330, which also used to be a characteristic of the out-of-camera JPEGs with firmware versions earlier than 1.2. The green tint can be easily corrected with the 'color deflection' controller in the white balance section of Silkypix Developer Studio.

   An 'interesting' aspect of the E-300 always was its tendency to cast red tones into orange when shooting JPEG, but also when certain RAW converters other than Olympus Master or Studio are involved. The E-500, which incorporates the same sensor (but was not included in this comparison), has solved this problem for the JPEG, but many RAW converters still have it. Silkypix does not; both E-300 and E-500 produce sensible red tones with it.

   So, colour reproduction in Silkypix Developer Studio seems to be nothing to particularly help anyone decide pro or against one of these five Olympus DSLRs, if there ever will be a reason for such a decision.

Even if there would be the potential for one, a final and definite conclusion is of course out of order, if only because this little study omitted lots of topics a complete DSLR review would need to cover. But if I would have to judge the five compared cameras by what was looked upon here, which were just several basic aspects of image quality when using RAW with Silkypix Developer Studio, I would say: Don't hesitate to go for megapixels, if you want or need them, but avoid the E-400, if you need high-ISO as well. Otherwise, go for features, viewfinder quality, build, continuous shooting performance, image stabilization, weight and size, whatever; if neither megapixels nor high ISO are of too much importance, this comparison won't help much, and dynamic range is really a non-issue between these cameras, shooting RAW and developing with Silkypix. Availability, though, might well become an aspect; of the five cameras only the E-510 is still being manufactured at this time (28-Feb-08).

And on a personal note, I'm quite happy with the E-510 since I bought it, although I'm missing the E-330's Live View Mode A and tiltable display. Olympus would do well to produce a replacement for the first and only DSLR (at least until the recently announced Sony DSLRs come out) with a working combination of Live View and DSLR-like, fast AF.

RS, 19-Oct-07, 28-Feb-08