You’re probably wondering why this DP3M doesn’t look anything like the press release photos – my friend attached a RRS grip and plate to it, and rightly so; without it, the camera is not very comfortable to hold.
I eventually caved to both pressure and curiosity, and borrowed the Sigma DP3M from master printer and good friend Wesley Wong – who has the DP Merrill in all three flavors. In case you’ve been living under a rock for the last few years, Sigma has been going their own way with the DP series of large sensor compacts; all of their cameras now share the same 14.7 MP (effective) three-layered Foveon sensor, with a 4.99 micron pixel pitch and true color/ true resolution information across all photosites. In a nutshell, the difference between Foveon and Bayer sensors is that the former records actual RGB values for each pixel, but the latter only records R, G or B, interpolating the other values from neighboring photosites. It’s difficult to determine precisely just how much resolution loss Bayer interpolation causes, but in my experience it seems to be around 50% or so. Sigma claims that the camera has the equivalent of 46MP (being 15.3 total MP x3 layers) – but this is really pushing it; images upsampled this far simply do not have the pixel-level ‘bite’ of a good Bayer file.
If you’re thinking Foveon technology is too good to be true, you’ll be masochistically pleased to know there’s a catch: because of the layered design; by the time light has been filtered through the first two layers, the signal from the final layer has to be amplified considerably. This of course results in higher than normal noise; the early cameras – I used a first-generation DP1 – were basically unusable above base ISO, despite having very large photosites. It’s nowhere near as bad in the latest generation, but it still won’t be competing with a Bayer sensor for low light performance – more on this later.
It’s important to take a step back and look at what the DP-M cameras represent as a whole: an alternative to the norm. They have significant compromises, but we should still give Sigma credit for innovating and sticking to their guns (though admittedly they do make some pretty lousy consumer zooms, and pricing of the SD1 was just insane). Lately, competition in the large sensor-fixed-lens-semi-compact market – I’m sure somebody will come up with a snappy acronym soon – has been hotting up; in the past, the DPs had this market to themselves almost exclusively. Even more so when prices started dropping for stock clearances; I still remember seeing DP1x and DP2x cameras going around the US$300 mark, though they had the much older sensor. Today, all of the DP-Ms are priced between US$800-1000, depending on the lens. The cameras are almost identical, except for the lenses – you get 28/2.8, 45/2.8 and 75/2.8 macro equivalents. Casual experience with the other two suggests that there really is almost no difference in normal use. It’s an interesting solution to the lens problem: the optics are very well matched to the sensor, and even carrying three cameras isn’t that bulky. I’m actually in favor of this in practice for several reasons: the optical results are better; you can optimize your camera setup to the focal length you’re using; there’s redundancy; and finally, there’s no need to fumble for lens changes when an opportunity presents itself. (Of course, you are carrying and paying for several bodies, so for this to be a workable option, there are size and price constraints.)
Having had the chance to use any one of the three, I opted for the 75/2.8 macro-equivalent DP3M for two reasons: firstly, fixed long-lens cameras are pretty much unheard of, making the DP3M unique; secondly, the very close focus distance made me curious to see how it’d stack up for product photography in the studio. I think the other three cameras are similarly specialized – the DP1M/ 28mm is probably the landscape pick; the DP2M/ 45mm for general-purpose use, and the DP3M/ 75mm is of course good for portraits and still life work.
Let’s get the bad things out of the way first: the minimalist design looks great, but is an ergonomic nightmare. Without the RRS grip my friend installed, it’s simply not comfortable to hold at all; the edges dig into your fingers and the camera just wants to twist out of your hand because the surface isn’t very grippy at all. Next, it’s not quite as slow as everybody makes it out to be, but it’s just slow enough that it will test your patience. The lack of speed is apparent in everything – focusing is similar to the first generation X100 or Leica X1, and that’s with the focus limiter set to a generous 1m; it’s worse without – but I suppose at least they let us choose the limit distance at all. Menu navigation is not as responsive as it could be; everything feels just slightly laggy. Playback is slow; write speed is excruciatingly slow – we’re talking 10-15 seconds with the fastest cards available – good thing it has a decently sized buffer. Seven raw files may not sound like a lot, but when it takes an age to shoot it anyway, I never found the need for more. Startup could be faster, too; the GR and Coolpix A leave it in the dust. Finally, there’s the battery life – or lack of. Even Sigma acknowledge this by including two batteries in the box; you’ll need them. The best performance I managed was about 120 images before the camera gave up. I’m sure there are good reasons for all of these things – large (up to 60MB) files; having three layers of sensor sites active, etc; the reality is that there are other cameras that shuttle around a lot more data – the RX100 comes to mind – and have larger active sensor areas, and none of them have such poor performance.
Actually, that wasn’t too bad: there’s only one paragraph of bad things. If you’ve made it this far, there are many, many more positives to this camera. In use, the level of customization is excellent; as mentioned before, you can not only configure the minimum focus distance for the limiter, but you have a limiter in the first place! Setting shortcuts on the D-pad is easy, and you can have two sets using the QS button. In fact, almost all of the button and shooting behaviours are customizable; I didn’t find anything I wanted missing – which says a lot. Overall configurability isn’t quite as good as the Ricoh GR series, but it’s comparable to the Olympus OM-D. There’s only one thing missing from the controls, though – that’s another control dial. The DP3M’s sole rotary is a jumbo affair under the shutter button, and easily turned with your forefinger. The trouble with this is you have to use the left and right keys for exposure compensation or shutter speed in manual mode, and that’s not so fast in practical use. While we’re on the subject of buttons, I don’t like the power button placement – it’s a stretch to reach with your index finger, and you often land up hitting MODE instead – which feels the same, and is right next to it.
There’s also the build quality: it’s pretty solid, all-metal, and put together in a way that inspires confidence. This includes the lens barrels, too; they also have nicely-damped focusing rings that are a pleasure to use, even though they’re fly-by-wire; you have enough fine control over the lens movement to focus accurately and easily. If only they’d put an aperture ring on there, too; it’s not as though there isn’t enough space. The camera’s hotshoe is compatible with Sigma TTL flashes and centred over the lens for use with an optical viewfinder; I can see this being useful on the DP1M and DP2M, but not precise enough for the longer DP3M. I’ve left image quality for last, deliberately: there’s a lot to talk about here, especially because comparisons aren’t exactly like-to-like.
Let’s start with the lens: each of the cameras has a short-backfocus optic matched to the sensor; the DP3M has a 50/2.8 with a 10/8 optical formula, capable of focusing down to 22.6cm and achieving magnification of 1:3 (on APS-C; this is about 72x48mm minimum subject size). It’s multicoated, has a 7-element iris and a leaf shutter that tops out at 1/2000s – but you need to be at f5.6; wide open, you get 1/1250s. I don’t believe this is the same design as the 50/2.8 macro that Sigma also offers in a variety of other mounts; it’s much smaller, for starters. Maximum aperture may be moderate, but it delivers – on the extremely high resolution Foveon sensor – even wide open, and in the corners. I think it improves a hair when stopped down to f4 or smaller, but I honestly can’t see the difference most of the time – it’s that good.
The only time you see hints of the lens not quite matching up to the sensor is when shot wide open and near minimum focus distance; I don’t think it was optimized for close distances, as contrast reduces and we start to see small traces of uncorrected coma. That said, all of this is gone by f5.6 and things are perfect again. Lateral chromatic aberration is almost completely absent, even against high contrast subjects; there are traces of longitudinal chromatic aberration close up if used wider than f5.6. Bokeh is nice and smooth – both in the foreground and background. If there’s anything that I have to dock the lens points for, it’s mild pincushion distortion and some slight flare – but in all fairness, I didn’t have the lens hood, either. In short: the lens is really quite excellent, and supports my preference for matching optics to a sensor.
Time to take a look at the sensor. I should start by saying I’ve only seen this level of pixel acuity in a small handful of cameras: one is the Leica M Monochrom, that doesn’t have a Bayer sensor, and the others are all medium format. Even then, there’s a very slight but noticeable advantage to the non-Bayer cameras. There is no question that the DP3M is punching way above its pixel count; I’d say it’s probably resolving at a level not far off the D800E. Continuing the good news, color is mostly accurate; as one would expect from a sensor that samples each channel at every pixel location. However, I found it to look a bit ‘thin'; it’s hard to describe, but images from the Foveon cameras look as though the red channel is slightly desaturated compared to the others, which I suppose is a consequence of the sensor architecture. It’s not an unpleasant look, but certainly a very unique one; limpid clarity is what comes to mind – something akin to cold mountain spring water. I personally find most native Bayer output to be a bit too Disneyland and requiring some toning down anyway.
Comparison – click here for 100% version. From left to right: 100% crops of Olympus OM-D with 14-42 X (I didn’t have any other lens that could could closely match the FOV), Sigma DP3M and Nikon D800E with Zeiss-Hasselblad CF 80/2.8 (closest I could get to 75mm in a prime) and 24-120/4 VR. This represents the best of the 15MP Bayer cameras, the Foveons, and something else for benchmarking. The top row shows the cameras at their native output sizes; the bottom, upsized to match the D800E. No sharpening was applied other than converter defaults; I did my best to match contrast on all images without changing native color balance. It’s clear that as good as the OM-D is, it’s clearly left behind by both of the other cameras, which is to be expected. Repeating the test with the 45/1.8 didn’t yield any improvement – shown in isolation, this is a fair showing for the OM-D. The Sigma shows single pixel detail at native size; however, we’re clearly starting to see artefacts and general coarseness in the file upsized to match the D800E; note the ‘blockiness’ in the leaves compared to the D800E file. The DP3M doesn’t seem to be resolving any more detail – see the leaves again – but not much less, either – see the bricks. Interestingly, the difference between the 24-120/4 VR and Zeiss 80/2.8 is there, but not as noticeable as one would expect. I initially performed this test with the 85/2.8 PCE – but that gave too much of a magnification advantage to the D800E, and is incidentally a much superior lens to either optic used here.
Sensitivity test. Click here for 100%.
We get resolution in exchange for noise, it seems. Though the camera isn’t as bad as most people make it out to be, ISO 800 is the limit if you value color accuracy; anything above that starts to show noticeable channel-specific noise, depending on the color temperature of the ambient light. The images aren’t actually that noisy – the color is just isn’t at all accurate, and at the two highest sensitivities, some very odd large-patch artefacts can be seen. I’d be comfortable with ISO 1600 and perhaps even ISO 3200 in black and white. The good news is that the files do take noise reduction fairly well; even at ISO 6400 there isn’t much loss of resolution (with the default settings in SPP) since there was so much to begin with; resolution at this point is comparable to a good Bayer camera at low ISOs. Dynamic range at base ISO appears to be comparable to or slightly better than the 16MP APS-C Bayer cameras; individual channels don’t seem to clip as quickly. However, you quickly start to lose the shadows as you go up the sensitivity scale; it’s easily a stop or more per increase in sensitivity.
Up til now, I haven’t really talked about SPP – Sigma Photo Pro. This is the free software that you need to convert the DP3M’s X3F raw files; it’s similar in feel to Lightroom, but nowhere near as powerful or fast; in fact, the software is downright clunky, and has a lot of odd glitches – if you open a file to process and close it, you lose your place in the thumbnails and go back to the start, for instance. You have to render files at full size before you can preview at 100%. The adjustment latitude for the highlight and shadow recovery sliders is very limited, and there doesn’t seem to be any curves function. Overall operation just feels mired in treacle; it’s not for want of processing power on this end – I have no problem opening a dozen 16-bit raw files from the D800E in Photoshop and working on them with 2000-pixel wide feathered brushes. At least the files come out of camera looking remarkably natural and with a very pleasing native tonal map; they don’t require that much work. Monochrome conversions are excellent, perhaps the best and most flexible digital solution I’ve seen to date – the M-Monochrom doesn’t give you any post-capture flexibility for channel mixing, and medium format digital still lags in tonal response.
Since nobody else seems to support the DP-M’s files, we have no choice but to suffer the intermediate step of working the files in SPP, exporting a 16-bit TIFF, opening them again in ACR to make some other specific adjustments like extra highlight/ shadow recovery or gradient filters, and then processing in Photoshop – needless to say, this is an extremely slow process. At least there won’t be too many files to process, since you’ll either have run out of battery or been limited by the speed of the camera. I would love to see just how much latitude the camera’s files have when run through a proper piece of software like ACR; that said, given what Adobe have done with every other non-Bayer file – Fuji X-Trans comes to mind – I don’t have high hopes here, either. It’s such a shame, because the poor workflow further cripples the camera’s outstanding imaging potential. I could happily use this for commercial studio photography – including watches – image quality is more than sufficient, and in the studio I use live view on a tripod anyway with the D800E. What gives me pause is not the speed, not the battery life, not the ergonomics – it’s workflow for dealing with large numbers of chunky files.
With every iteration of the DP series, Sigma improves a bit – but what it really needs is more of a radical change; make one body, make the lenses interchangeable, don’t worry about exposed sensors, and leave the software processing to Adobe. All of the other things – operation speed, ergonomics and battery life – are easily solvable; changing the camera’s shape would make it more comfortable to handle, allow installation of a larger battery, which could in turn support a more faster and more powerful processor – and you’d have a killer mirrorless system. As it is, this camera challenges the very best of the 35mm full frame DSLRs on image quality; it’s probably better than the previous generation of medium format digital backs, too. And it still beats the current ones on high ISO performance. If you have to have the ultimate image quality in as small a package as possible, there’s no other option – and I think the compromises simply aren’t that much of an issue when compared to the alternatives. I might actually pick up one for myself…MT
Visit the Teaching Store to up your photographic game – including workshop and Photoshop Workflow videos and the customized Email School of Photography; or go mobile with the Photography Compendium for iPad. You can also get your gear from B&H and Amazon. Prices are the same as normal, however a small portion of your purchase value is referred back to me. Thanks!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved