A Compact Conundrum
Is That a DSLR in Your Pocket...?
The two seem completely at odds with each other – image quality and size. It's been a rule of thumb since the invention of photography that a larger camera with a larger recording area (be it silver or silicon) will – all else being equal – produce superior image quality. And so photographers continue to schlep around cameras and lenses that weigh them down; all in a search for the optimum image.
On the other hand, we all would like to be free from encumbrances. To walk around a foreign city (or a dangerous one) without the weight and sometimes attention drawing of a DSLR would be nice. To take a hike in the woods with ones spouse, able to photograph the mists of early morning or a chance encounter with a deer, without the baggage (both literal and metaphorical) that a DSLR involves is something many of us wish for.
The digital revolution has provided us with no end of cute, small, inexpensive and well featured little pocket cameras that would seem to do the trick. For some users they do. For snapshots at family gatherings or vacation pictures which will end up as smallish prints from Walmart, they're often just fine. But for the more serious and critical photographer, no matter how appealing these cameras might be in terms of features they usually disappoint in terms of image quality.
The reason for this is simple: small sensors. For example the Ricoh GR2, a very popular pocket digicam among more serious photographers, has a sensor that measures 7.6 X 9.5mm. Take out a ruler to see how small this is – about the size of your smallest finger nail. And, as we know, smaller sensors, especially those with high pixel counts (which we mostly all want) will have very small photo sites. Net results – noisy images, even at the lowest ISO settings, or, if the manufacturer has used aggressive noise reduction on-chip, reduced resolution.
An APS-C Sized Sensor
Which brings us to the Sigma DP-1, the subject of this report. For the first time we have a shirt-pocket sized digicam using a roughly APS-C sized sensor. This is approximately the size used in popular DSLRs such as the Canon 40D and Nikon D300, and slightly larger than the 4/3 format used by Olympus and others in their DSLRs.
To make the comparison a bit easier to visualize let's compare the diagonals of these formats. The Ricoh GR2 and Canon G9 (and similar digicams) have sensors with about a 10mm diagonal. On the other hand a 4/3 format DSLR, like those from Olympus, has a 22.5mm diagonal. A typical Nikon or Canon or Pentax DSLR with a roughly APS-C sized sensor has a diagonal measurement of about 28mm. Godzilla was right – size does matter.
Sigma DP1 @ ISO 100
The Foveon Factor
But wait! There's more that's unique about the Sigma DP1. It is one of the only cameras in the world to currently use a Foveon sensor. Its big brother, the Sigma SD14, is also thus equipped. And to confuse matters, both the Sigma DSLR and the new DP1 not only have essentially the same sensor, but it's just 4.6 megapixels in resolution.
What? – you say. No one has offered even a sub-6MP digicam let alone a DSLR in years with such low resolution. This is the resolution of digicams from 2003, not five years later on, here in 2008. But a Foveon sensor is not a typical one, and if you're not familiar with this unique technology, where each photosite location contains stacked light sensitive layers for all three colours rather the the usual Bayer matrix array, you should give a read to the Wikipedia entry on this, and then the sidebar below.
Foveon and Bayer Compared
It really isn't possible to compare megapixel count between a Bayer matrix based sensor and camera, and a Foveon based one. The numbers just don't jibe, or tell you much of value. A 10MP Bayer sensor, for example, because of its antialiasing filter and matrix colour recording, both of which reduce resolution, has less of an advantage over a 4.5MP Foveon sensor as used in the DP1 than the numbers would suggest. The 10MP Bayer has, many experts agree, about a third less resolution than a comparably sized monochrome sensor would.
A 4.5MP Foveon sensor, on the other hand, has neither matrix decoding nor an AA filter to reduce resolution, and thus is considered by some experts to be equivalent in terms of spatial resolution to a sensor one third more to double its pixel count.
So, arguably, and based on this analysis, a 4.5MP Foveon should reasonably be compared to an 8-10MP Bayer based camera.
The problem is that Sigma and Foveon have long been pitching their wares as having triple their actual pixel count. So the DP1 and the SD14 DSLR are advertised as having 14.1MP. This may make good advertising copy, but it just confuses the hell out of people who are trying to understand what this is really all about and to arrive at an intelligent purchasing decision. It also leads to endless punch-ups on web discussion forums, where the Foveon Faithful have to defend their turf against often uninformed naysayers.
Without taking sides on the issue, I'll just add, as is indicated below, that the DP1 is able to make extremely sharp and very low noise 13X19" prints when ressed up slightly (more on this below), that are far superior to what you might expect from the 4.5MP numbers, and which appear to me to be quite comparable in IQ to those from a typical 8 – 10MP DSLR. Note that I write DSLR, not digicam, because the overall image quality of the DP1 is on quite a different level from that of other pocket cameras. This is, I believe, due to the size of its sensor, not necessarily any special properties of the Foveon technology, but therein lies a tale, so read on.
The Sigma DP1 is quite small, though of manageable size. Some pocket digicams have become so small that they sit nicely in a shirt pocket, but are fussy to work with. I find the DP1 to be just about right. It's a bit smaller (actually, mostly thinner) than the Canon G9, but when you add the need for an accessory viewfinder to the Sigma the extra bulk of the Canon diminishes in importance. The Ricoh GX100 is thinner and actually easier to handle because of its contoured grip. Like the Canon the DP1 is quite boxy, though Sigma has placed dimpled gripping surfaces where needed, which helps a lot.
Built quality appears high, with a metal body and mostly quality fit and finish.
A snap-on lens cap is provided, but unlike the Ricoh it isn't tethered, and thus easily misplaced. The Canon has a self capping lens mechanism, which is my preference in any event in this sized camera. I also was a bit frustrated with the Sigma's lens cap, because it will only fit on in a certain orientation, making put it away in ones pocket a process that requires visual inspection.
Even though the DP1 has a fixed focal length lens (28mm equivalent) rather than a zoom, it still extends when the camera is turned on, somewhat slowing down the shooting process.
A neckstap is provided, but why anyone would want to put so diminutive a camera on a strap is beyond me. This size camera wants to have a wrist strap, and that's the first thing I did, stealing one from a pocket sized tape recorder that I no longer use. (I can foresee a huge online business in unused factory wrapped Sigma DP1 neck stapes in the days ahead).
Sigma DP1 @ ISO 100
The DP1's menus are a mess. There are poorly laid out and have 15 items in a scrolling Shooting menu and another 17 items in a Set Up menu. There are no shortcuts such as there are in most other cameras, where one can access frequently used settings either with a dedicated wheel or control. Everything on the DP1 is controlled via the main LCD menus. Tedious and not conducive to any form of rapid and involving interaction with the user. It's almost as if Sigma hasn't noticed the advancements which its competitors have made in the last half decade in user interface development.
The Sigma has the usual slow contrast detection autofocus that all digicams offer; though seemingly a bit slower than some others. But again, in terms of poor user interface, though it has multiple focus points available there's no way of getting to them other than though a menu item buried within one of the scrolling lists. Not at all good.
Curiously, there is a manual focus control wheel on the top left panel. Combined with a button which allows a magnified view on the LCD screen, quite accurate focusing is possible. But (you knew there was a but, didn't you) it's a free turning wheel and easily moved out of position with the brush of a finger. Also, incredibly, there is no lock on the infinity position nor is there a hyperfocal setting such as provided by Ricoh on the GX100. I would gladly give up manual focus to have Infinity and Hyperfocal lock positions. These are far more useful in a pocket digicam than magnified manual focusing at arms length.
My approach to shooting with the DP1 has been to frame with the Voigtlander optical viewfinder (see below) and trust the autofocus, shooting when the beep is heard, indicating that autofocus has locked. I even do this when shooting distant objects because there is no way to be sure that the camera will stay at Infinity because of the lack of a lock in the focus wheel. What was Sigma doing during the 18 months since product announcement? This stuff isn't rocket science. Just buy a few competitor's products and see how they do it!
It is possible to set the camera by either pre-focusing and then switching to manual, or by guess focusing. This is almost mandatory when trying to work quickly and with rapid shutter response. It all would work a whole lot better though if Sigma had emulated Ricoh's hyperfocal lock setting.
Be aware as well that because of the larger sensor, depth of field is quite a bit narrower than on the typical digicam. This makes accurate focusing even more important, which, of course, exacerbates the camera's focusing issues.
Features and Functions
The DP1 has a fairly modest list of features for its price. US $800 is at the very high end of the scale, if not the very top, and if it weren't for its sterling image quality (as we'll soon see) it would be seriously out-featured and out-priced by hundreds of dollars by several other pocket cameras that also shoot raw.
While there are some fixed lens pocket cameras (those from Ricoh come to mind) most of this breed have zoom lenses. The Sigma has a fixed 16.8mm f/4 lens, which is equivalent to a 28mm lens on a full frame camera. This focal length will be found to their liking by many street photographers, but will be too wide by many. My personal preference would have been 35mm, but I understand that many folks like something wider. The f/4 aperture is a trade off with physical size. F/2.8 would have been very welcome, but I appreciate that it would have been very tough to design a small enough lens with such an aperture. If they'd gone to 35mm it would likely not have been as tough though, but might have produced reduced sales. Trade offs, I know.
The LCD screen of the DP1 is a disgrace. It's simply awful. Resolution appears low, and it simply appears dim and coarse. Yes, I know that it can be made brighter in the menus, but that still doesn't put it anywhere near on a par with other screens. The one on the Canon G9 is not only larger, but brighter, of higher resolution, and is much crisper. The Ricoh's screen isn't that hot, but it's streets better than the Sigmas.
I also was distressed when shooting in low light (say a living room at night) to note that the Sigma's LCD looses colour. Shots made are in colour, but the screen is totally washed out and monochrome, and this isn't a special effect. I haven't seen this behaviour in years. Suddenly it's 2002 all over again.
Also surprising for any digicam in 2008 is the lack of a live histogram. There isn't even a provision for flashing overexposed highlights. There is a histogram on review, but it's small and poorly defined. Almost useless. Frankly, for a camera shooting raw, and which is designed for the more serious photographer, the lack of a live histogram is a serious omission.
In addition to ISO settings from 100 to 800 there is an Auto-ISO mode. But it only moves between ISO 100 and 200, not including any of the other speeds. Seems to be kind of pointless to me, and another example of strange oversights and omissions that Sigma has allowed to exist in this camera.
With the DP1's LCD being as poor as it is, a separate optical viewfinder would be a good idea, and Sigmas offers one. I wasn't able to get my hands on one during this test period, but I had a Voigtlander 28mm accessory shoe mount finder on hand and this did the job very nicely. The only problem with using this particular finder is that it blocks the pop-up flash from fully opening.
By comparison the electronic viewfinder on the Ricoh GX100 isn't as bright as an optical VF, but it does show highly accurate framing and technical information. The optical viewfinder built into the Canon G9 is pretty small and dim, and lacks accurate coverage, but it's better than nothing, and when combined with that camera's big and bright LCD makes it an acceptable combination.
Flash & Stabilization
The pop-up flash on the DP1 is anemic. There is a proprietary Sigma accessory shoe mount flash available, but it was unavailable for testing. Fortunately the DP1 has pretty high image quality at ISO 400, and is acceptable at ISO 800, so available light shooting with its f/4 lens should not be much of any issue except in the dimmest lighting conditions.
More to the point is that lack of any form of stabilization on the DP1, either optical or electronic. Again, in 2008 this is a real omission, with IS now standard on virtually every camera on the market, including those at the lower end of the price range in this segment.
JPG and Movies
Note that I did not test either JPGs or video output. It seems to me that this camera, because it is so pricey and yet feature poor, is all about image quality, and so therefore to shoot JPGs with the DP1 would be for snapshots only, and that doesn't interest me, at least not for this test. Neither do I think that many people will bother shooting video with the DP1, and I know I won't, so I didn't bother doing any shooting in these modes.
Be aware that because of the low spatial resolution of the Foveon sensor videos produced are only 320 X 240 pixels, barely big enough for use on Youtube, and hardly watchable on TV let along on a computer monitor. In any event, please refer to one of the popular digicam review sites for their insights into how the DP1 does in these areas if they still interest you.
The Sigma DP1 shoots raw, and that along with its large sensor and Foveon technology is a key component in its appeal. But, just as with all other digicams which shoot raw, the DP1 is slow to save its files to memory card. Why camera makers can't put in ASICs that can write and read at the same time, as they do in DSLRs, is beyond me.
The DP1 is provided with raw processing software for Windows and Macs called Sigma Photo Pro. Unfortunately this is very primitive software; simple to use, but very slow and under-featured. Apparently Sigma has changed the compression algorithm on their raw format from that of the SD14 camera, even though these use the same sensor, and so neither Adobe Lightroom nor Camera Raw yet support the DP1 as of mid-April, 2008. Hopefully soon though, and not soon enough for my liking, because raw processing is currently torture with the Sigma software.
A curious aspect of the Sigma raw software is that there appears to be a fairly high amount of sharpening applied to raw files at the default (zero) position. In fact, I found it necessary to dial down sharpening to -4 before exporting to get a level of sharpening that I found acceptable. But, if I ran sharpening all the way down I also found that the image because so lacking in accutance that no amount of sharpening later on would bring it back to life. I simply don't get what Sigma is up to with this. Don't mess with the sharpening of raw files Sigma. That's our job!
Though a minor point, the supplied charger has a cord rather than a built-in plug. This makes travel with the charger more awkward than it should be. Most camera makers now get it – that small cameras need small chargers. Sigma doesn't.
Image Quality Assessment
Compared with the Canon G9 and Ricoh GX100
Over the past few months this site has seen reviews of two high-end pocket digicams, the Canon G9 and the Ricoh GX100. Image quality comparisons will therefore be made here with these two models, if for no other reason that because they are on-hand, and relatively popular alternatives. Both are also at the higher end of the digicam price scale, though neither is as pricey as the DP1.
Here's the summary up-front, to save you from the tedious chore of examining the images below and needing to read the text. The raw image quality of the Sigma DP1 is excellent – truly far and away superior to that of any digicam that i have yet seen or used. Given that the camera uses an almost APS-C sized sensor this should not be all that surprising. Combine this with what appears to be a very high quality lens (Sigma has some small experience in this area) and you have a winning combination.
The fact that the camera uses a Foveon sensor seems to be both good news and bad news. The much vaunted (at least by its fans) advantages of the Foveon chip just don't seem all that apparent to me. For the most part colour rendition looks good, though it's not all that exceptional as to be obviously superior to decent mid-range Bayer sensors I am familiar with. Detail resolution seems high, especially given the sensor's size, but the very fine lens that the camera is using may also be an important factor in this judgment.
The downside of the Foveon sensor in the DP1 is its low pixel count. As explained above, the camera may in fact be comparable to a Bayer-based camera with some 8 Megapixels, but if one wants to make larger prints one has to be willing to work a bit at processing and resizing.
I tried the Sigma raw software's provided image size doubling capability and found it to be awful. You're much better off up resizing in Photoshop. My preferred approach for making prints larger than 13X19" from the DP1 (if smaller than that I simply leave output resolution as-is) is to res up in Photoshop using Bicubic Smoother, or if printing from Lightroom 2.0 to allow that program's printing output upressing capability to do the job.
As will be seen in the various samples below, in every area which I've tried (and I've done more than what appears here), the DP1 either equals or whacks the completion handily. There simply isn't a digicam that I've seen that can compete in just about any area of image quality comparison.
Is the Sigma DP1 the equal in image quality of a DSLR? This is something that that I'm deliberately not going to comment on, for several reasons. The first is that I don't know what to compare it with that won't have people asking why I didn't compare it with something else. Also, lens choice plays a major role, as does raw processing. Pitfalls abound, and I simply am not interested in falling into any of them at this time.
Is the DP1 in the ballpark when it comes to image quality with DSLRs. Yes, it is. That I can comfortably say. If you are happy with the shooting constraints that working with a DP1 entails, don't need fast frame rates or clean high ISO above 400, then yes, the DP1 can be a pocketable substitute for a DSLR. But one has to keep in mind the ultimate output size and cropping limitations that this camera insists on.
Samples and Crops
In each instance below I will include one image showing the overall scene photographed by the DP1 and then 100% crops from each of the three cameras, comparing the Sigma DP1 with the Canon G9 and Ricoh GX100. In each case images were shot raw and white balance and basic exposure processed in Lightroom 2.0. The Sigma files were first minimally processed in Sigma's raw software for white balance, basic exposure and black point, then exported to TIFFs as 16 bit files in Adobe RGB. Sharpening was done to the Sigma files in the Sigma software and not at all in Lightroom. Sharpening, while balance and exposure adjustments were done to the Canon .CR2 and Ricoh .DNGs in Lightroom.
Sidebar – Which Side of the Mushroom to Bite?
Alice was told that taking a bite of one side of the mushroom would make her taller and the other side would make her smaller. That's our dilemma as well.
We have three cameras on comparison; one that's 4.5MP (sort of), one that's 12.1MP, and a third that's 10MP. What to do? Ress the smaller one up, or res the bigger ones down. Or, leave them as is? The problem is that none of these approaches is ideal, and each is subject to someone crying foul. Remember, as described earlier in this review – Foveon files are able to be ressed-up to a much greater extent than can images from a Bayer camera, but are smaller to begin with.
After much experimentation I decided to reduce the Canon and Ricoh files to the same on-screen dimensions as the Sigma files using Bicubic Sharper. This allows them to display conveniently here beside each other, and isn't a factor when any aspect of IQ other than resolution is discussed.
Daylight – ISO 100
Even at ISO 100 the Canon and Ricoh (especially the Ricoh) show a bit of luminance texture. The Sigma is completely clean. Not that this is a big deal, since it's all luminance noise and can easily be eliminated with the Luminance Noise slider in Camera Raw or Lightroom, without any real effect on resolution. But it is an indication that high pixel count / small sized sensors do have their limitations.
At ISO 400 the Sigma does very well against both competitors, showing a clean and natural looking image with hardly any noise. The Canon holds up quite well with regard to noise well but isn't very good at holding highlight detail. The Ricoh fares least well, exhibiting chroma as well as luminance noise at this speed, and similarly doesn't do well with strong highlights, as seen along the banister in all shots.
Given how impressed I've been with the Ricoh's handling, it's quite disappointing to see how poorly the image quality holds up at ISO 400.
ISO 400 vs 800
As good as the DP1 is at ISO 400, it starts to lose grace at ISO 800. It's not terrible, but chroma noise definitely starts to become visible. I would say 400 is a very useable speed when needed, but 800 should be reserved for those times that quality is secondary to getting the shot.
The subtitle of the review called the DP1 a "conundrum", and indeed its design is a bit of mystery. Firstly, it was announced at Photokina in September, 2006 yet didn't begin to ship until March of 2008. Elephants don't gestate that long.
Sigma commented publicly during that long wait that it had some technical design and IQ issues to address that prevented its anticipated earlier delivery – and that's fine – but what this 18 month hiatus makes all the more curious is the question of what on earth the Sigma DP1 engineers who weren't fixing sensor / processing issues were doing during that time. The camera has so many small design deficiencies, that they weren't addressed in all the time that the company had available to it is quite surprising.
It doesn't take long for one to see though that the DP1 is a special animal. Controversial though its sensor technology may still be, and as confusing as the pixel count issue still is for some people, the clean, sharp, and high resolution images that this little pocket camera can produce are nothing short of remarkable. It's clearly a noticeable step above that from any other pocket camera.
Though I don't have any other experience with a Foveon-equipped sensor camera, what time I have spent with the DP1 shows me that a great deal of the nattering on on-line forums about the almost magical properties of the Foveon sensor are just that – nattering. There simply isn't any marked advantage that I can see over a Bayer matrix sensor, other than that the chip's lower spatial pixel count does translate into higher apparent enlargability than the numbers alone would indicate.
Part of the issue is that dematrixing algorithms, the use of micolenses and other advances, have improved Bayer-based matrix technology substantially over the past 7-8 years, and every other digital sensor on the market, from the lowliest point-and-shoot to $30,000 medium format backs uses Bayer technology.
Does Foveon offer some advantages. Yes, it definitely appears to. But it also has downsides, such as reduced red channel sensitivity, and poorer low light sensitivity than contemporary cameras from makers that use non-Foveon chips.
When Foveon technology was announced back in the 90's there were those that forecast that it was so far superior to Bayer that it would spell the death knell for matrix colour decoding. Time has shown that this has not been the case. Indeed with the exception of some specialized technical cameras, and Polaroid a couple of years ago in a small point-and-shoot, the Foveon designed chip has only appeared in cameras from Sigma. If the technology was that good one would have to imagine that some other cameras makers would have at least dabbled in it.
I can't speak to the business side of things, as I have absolutely no information on any negotiations and testing that may have taken place with major camera makers other than Sigma. But truely winning technology has a way of rising to the top (just look at CMOS vs CCD before Canon showed what was possible with the original D30).
And speaking of CMOS, since the Foveon sensor is itself CMOS technology, one has to ask why Foveon has found itself so limited in its ability to produce higher resolution sensors. The original Sigma SD9 DSLR of 2002 had a 3.4 MP chip and it wasn't until four years later in 2006 that the SD14 raised this to a whopping 4.6MP. In that same period other DSLRs went from about 6 MP to 10 and even 12MP and 14MP. Not to say that megapixels are any sort of measure of absolute goodness, but it really does beg the question of why Foveon sensors have been so slow in scaling up.
The bottom line then is that the Sigma DP1 is able to produce very high quality images; possibly the highest from any current pocket-sized camera. But it's my belief that this quality stems less from its use of Foveon technology that it does from simply using an almost APS-C sized sensor and a high quality lens, and when (and if) other camera makers decide to follow this route, and assuming that they pay more attention to basic camera features and functions than Sigma did, it may be very hard for Sigma to compete unless they really pull up their socks. Hey Canon – how about a G10 with one of your 1.6X sensors and a decent optical viewfinder? Now that would be a competition killer, and would have serious photographers flocking to it in droves.
On the other hand, if Sigma were to bring out a next-generation DP1 style camera which had more attention paid to its operation as a camera than simply proving its technology, they might really have something as well.
Sean Reid at Reid Reviews has online Part One of a two part review of the Sigma DP1.