Sigma SD1 Review
Including Comparisons with
Pentax 645D / Leica M9 / Sony A900 / Sony A55
By Michael Reichmann, with the assistance of Nick Devlin
and the participation of Laurence Matson
The Sigma SD1 is that company's long-awaited APS-C sized camera using a Foveon X3 sensor. The camera was first shown at Photokina in late 2010 and started to ship in June, 2011. Its launch created something of a furor because the company had telegraphed that it would come in at under $2,000, yet when launched the list price was USD $9,700. This quickly slipped to a street price of $6,899, but still much more than anyone had expected or wanted it to cost. I wrote about this at the time in an essay titled Rationalizing The Irrational. I'll have more to say about the camera's pricing later in this report.
The Foveon sensor story is a long and complex one, and if you want background this Wikipedia article is the best place to start. Briefly though, its claimed advantage is that instead of placing red, blue, and green filters over adjacent pixels using a Bayer Matrix colour filter array, there are RGB sensitive layers stacked one on top of the other.
|Bayer Filter Array||Foveon X3|
The argument goes that with a Bayer sensor only the green filtered pixels contribute significantly to luminance data (and it is this part of the spectrum where the eye is most sensitive to resolution and detail), therefore claimed pixel resolution on a Bayer sensor should be reduced (for purposes of comparison) by about a third. As long as every colour sensor in the world used a Bayer matrix, this was an academic issue at best. But a Foveon sensor collects luminance information from every pixel location and thus the comparative debate was joined.
Add to this the fact that just about every camera uses an anti-aliasing filter to reduce colour moire, and you have the countervailing claims that an X3 sensor records more information than its actual pixel dimension would indicate (about double), while a typical Bayer array equipped camera has less resolution than claimed (by about a third). Photographers and pundits have been debating this point ad-nauseam for years.
There hasn't yet been a reasonable comparison made using contemporary cameras, because till now the highest resolution available from an X3 sensor was 4.3 Megapixels. Few except Sigma employees and the company's most ardent supporters would claim that this was equivalent to a 14MP sensor, and since mainstream Bayer sensor equipped APS-C cameras are in the 14 – 18MP range comparisons were stretched.
But now the SD1 has a 15.4 / 46 MP sensor (depending on how you count), and so a comparison is more meaningful. I'll have much more to say about the X3 technology and how it works and compares to mainstream sensors further in this report.
Red Barn. Clearview, Ontario, July, 2011
Sigma SD1 with 8-16mm @ ISO 100
Setting Up The Test
When I first saw an SD1 prototype in late 2010 I told Sigma that I was very keen on evaluating one as soon as possible. Shortly after the camera's launch I was put in touch with Laurence Matson to organize this evaluation. Laurence has had a long-time relationship with Sigma, primarily through his printing, and had good contacts within Foveon and its architects prior to that company's acquisition by Sigma. Laurence is also a very experienced and talented photographer, and a frequent and respected contributor on Sigma-oriented online forums.
Sigma wanted Laurence to show me the Sigma SD1 / X3 ropes, and that was fine with me. Laurence was on-hand for a couple of days during initial testing, with comments, background information and his own observations. But Laurence did not contribute in any way to the writing of this report. He has been invited to write a separate commentary, if he wishes, which I will publish without limitations.
X3 is a fascinating technology and I wanted to be sure that I understood it well. Of course there were no other strings attached by Sigma, and this report is based solely on my own observations and those of Nick Devlin, who is a regular contributor / reviewer for this site, and who assisted me with the comparison testing and evaluations.
I should point out that it is not uncommon for companies to send along representatives when a brand new product is being tested for the first time. I've tested products from Epson, Sony, Canon, Hasselblad, Phase One, Leaf and others where an "expert" was available to provide background information and product orientation. Never was there any attempt to influence my opinions or test results. Just thought that you'd like to know.
Sigma SD1 with 120-300mm f/2.8 @ ISO 800
(NB: This image has been desaturated and toned in Lightroom)
SD1 Features and Specs
The SD1 is configured as a mid-market spec'd camera. Here are some of the main features...
- Magnesium Alloy Body. Light and strong.
- Weather Sealed. A must for a pro-grade camera these days.
- 77 Element Metering and 11 Point Cross Sensor Autofocus. Great specs.
- Separate Motor For Mirror with MLU. Clearly attention has been paid to mirror vibration issues.
- 460K LCD Screen. Decent, but not as good as many of the lower priced competition which have 900k+ screens
- 98% Coverage Viewfinder. Good, but much of the mid-market competition now gives 100% coverage
- Built-in Flash. Always welcome, even Hasselblads have them.
- Removable Dust Protector. No automatic dust shake removal, but an easily removable dust shield that's easily cleaned.
This so-called dust filter is also an IR filter, which all cameras have, and which is needed because silicon sensors are much more sensitive to infrared than is the human eye. Because it is user removable this makes the SD1 (and its Sigma predecessors) the only current DSLR of which I am aware that can have its IR filter user-removed. This makes it an ideal choice for anyone interested in doing IR photography.
- 100,000 Cycle Rated Shutter. Very good projected mechanical reliability.
- 5 FPS Maximum Shooting Rate. Good, but sub $2,000 competition offers 6–8 FPS with larger buffers.
Postoral Creemore. Clearview, Ontario. July, 2011
Sigma SD1 with 70-200mm f/2.8 @ ISO 800
Before we look at image quality, which is where Sigma has to hang its hat to justify an almost $7,000 price tag, we need to be clear about the SD1's deficits.
- No live view. This is surprising since it is a capability that just about every CMOS sensor equipped contemporary DSLR has. Live View has become an important feature for many photographers since it is a great assist in assuring precision focus.
- No video. There is nothing in the Foveon X3 technology that precludes video support, so its absence on the SD1 puts it at a competitive disadvantage to even some $1,000 cameras. In fact, in an interview with Sigma executives at the last Photokina by dpreview, it was suggested that this sensor's video capabilities are outstanding. It's just not there in the production camera.
- No tethering. Being priced at a pro level, it is surprising that the SD1 doesn't include a pro feature much in demand; tethering shooting. In studio shooting situations this is now de-rigeur. (Note that the Pentax 645D, also included in this report, does not support tethering either.)
- No Pro-level service program. Pros demand support. Not tomorrow. Not next week. Now! When you make your living with a camera (or any product for that matter) dealer and manufacturer support in terms of technical assistance, loaners, fast replacements and quick repairs are expected. Nikon and Canon have Pro programs to address this, and even Pentax instituted a Pro support program when it launched the 645D. Once Sigma priced the SD1 in Pro territory it needed to consider pro-level support, and this it thus-far has not done.
- No access to third party lenses. There is an irony to this, because Sigma is primarily a third party lens company. But its proprietary SA lens mount does not have support or even adaptor-based compatibility from any other lens maker. Pros who currently shoot with Canon, Nikon, Sony/ Minolta or Pentax lenses will find themselves in the position of having to replace all of their glass if they want to move to an SD1. They, as well as new owners, will only have Sigma lenses to choose from (though the selection is large, and many of their top lenses are very good indeed).
- No Dual card slots. Cameras in the price range typically have two card slots and allow saving raw to one with JPG to the other, or copying between slots. (The M9 does not have this).
- No Third Part Raw Support. Sigma Photo Pro 5 software is underfeatured and flaky. According to several developers Sigma does not appear to be in any hurry to provide technical advice so that they can add raw support to their software. Virtually every other camera maker is highly supportive of third party developers. It helps sell cameras.
River of Grass. Clearview, Ontario. July, 2011
Sigma SD1 with 24-70mm f/2.8 @ ISO 100
Let's put aside image quality and price for the moment and look at the SD1 simply as a device for taking photographs. It's a mid-sized, medium speced APS-C sized sensor DSLR. The 98% viewfinder is decent, but for anyone used to a full frame camera such as a Sony A850, A900, Canon 5D MKII, IDS MKIII or Nikon D3 model, the view seems smallish, dimish and distantish.
The grip is large and deep and I found it extremely comfortable to hold, even with a large lens such as the Sigma 120–300mm f/2.8. The control layout is different than that of other cameras. Not better. Not worse. Just somewhat different. There will be a battery grip available, but it won't be ready till later in the year. Same with the new flash gun designed for the camera. There is a small pop-up flash which seems to function well for snapshots.
There is no top LCD (as is the trend these days), instead the rear LCD is used for all control settings. The screen itself is "OK", but in these days of 1 Million + dot screens, 460K seems bit chinzy. There are direct control buttons for Metering Mode, ISO, and Focus Point, along with a FUNCtion button which calls up a couple of screens, allowing rocker pad access to just about every major setting. This is similar to the system used on the Sony A900, which I am quite familiar with, and find comfortable to use.
There is a button marked QS (for Quick Set) which again provides direct access to a couple of screens worth of settings. This means that there are in some cases up to three different ways to adjust various controls and settings, which is actually a good thing, since each user will likely find the combination that appeals to their own style of working.
I was not happy though with how QS settings are activated (this is not new to the SD1). The four rocker pad buttons each activate a control (for example, Up is for ISO on the first of two screens) but then the ISO is actually changed by continuously pressing that same button. The settings then cycle from one end to the other. If you pass a setting that you want, you have to cycle all the way round again. Frankly, this is lame, and as a consequence this method of making adjustments was one that I avoided.
The top left dial has an OFF position for powering down the camera, and also positions for single frame advance, high speed frame advance, two second and 10 second self times, Mirror Lock with delay, and Auto Bracketing. (Note to Canon – check it out. MLU is a good thing. Don't be afraid of it).
On the top right is another dial with the usual PASM settings as well as three custom positions which can be programmed with ones favourite combinations of camera settings. There are no amatuer "Scene" modes, as befits a camera intended for the more advanced photographer. A vertical control wheel is located just behind the shutter release, and there is a second horizontally oriented one where ones thumb naturally falls.
None of the dials have interlocks, but all are reasonably stiff, and I only found the camera to be mis-set coming out of a bag a couple of times due to inadvertent handling.
Over-all the SD1 breaks no new ground in user interface design, but it's a very competent design and clearly photographer oriented. Too many cameras these days look like they were designed by engineers, and were just shown to photographers for input toward the end of the development cycle, almost as an afterthought. Not so the SD1, and so kudos to Sigma for this effort.
The SD1 takes a CF card, loaded in the usual position under the right palm pad. The battery is 1500mAh, and of typical size, though of course not compatible with any of the 1,000 other Lithium Ion camera batteries that come out of a couple of factories in the Far East. Battery life seemed good, and a single battery lasted an entire day of shooting (a few hundred frames at warm July temperatures). I would say that one extra battery would cover most situations.
As mentioned earlier, there is no video, no Live View, and no tethering.
Allignment. Clearview, Ontario. July, 2011
Sigma SD1 with 70-200mm f/2.8 @ ISO 400
Like any unfamiliar camera the SD1 takes a bit of time to become familiar with. It's fairly straightforward though, and I didn't bother looking at the manual, or even need to, until the third day, at which point I did study it to make sure that I hadn't missed anything important. I have to add that having Laurence Matson available for the first couple of days did speed up the familiarization process. Thanks Laurence.
The only control feature that I found badly designed is auto bracketing. One sets the number of frames (up to 5) and the order of shooting in terms of under and over exposure. The problem though comes when choosing to activate auto-bracketing, because all you've done so far is to set the increments that will be accomplished once activated. Here are the steps needed...
- Turn the top left mode dial to AB
- Looking at the rear LCD, turn the rear control dial to select the exposure increment, up to +/- 1.7 EV
- Turn the top left dial to one's preferred shooting mode, most appropriately High Speed
- Set the shooting mode dial to ones preferred setting, preferably Aperture Preferred
- Take the shots
- To exit auto bracketing mode turn the top left dial back to AB
- Using one of the control dials, set the AB increment to 0. If you don't, the camera continues to auto bracket
This is tedious and time consuming. But, it gets worse. Because of the SD1's very lethargic write times and unresponsiveness during file writing, doing something as simple as checking how the shots look, or switching modes becomes an exercise in frustration.
A five frame bracket takes just over a minute to write to the card (16 GB Sandisk Extreme IV – UDMA at 45 MB/s). By way of comparison, the 24MP Sony A900 can write seven files to the same card in 7 seconds. During those 60 seconds it's hit or miss whether you can get into any of the main menus, the FUNCtion display, or the QS display. Sometimes they work. Mostly they're locked out. The Sony, as with just about any other camera, allows full access to menus while files are writing.
This brings us to the whole topic of write speed. Just about every other DSLR made over the past five years can walk and chew gum at the same time. The SD1 can't. I've had it explained to me that the problem isn't so much the camera's write speed as it is the in-camera processing that X3 files require. So be it. But, this doesn't change the fact that as a consequence the SD1 becomes a much slower camera to work with than its competitors.
For the amatuer working slowly and patiently, doing landscape work for example, this may not be a big issue. But for someone shooting weddings, fashion, reportage, etc, not having access to the camera's main settings and controls during extended card saves can make the difference between getting the shots, and not.
One other annoyance. The battery indicator is inaccurate. It can be showing two bars, and then 20 frames later flashes red. Even though the SD1's battery life is pretty good, always have a spare on hand.
Otherwise there isn't much to criticise about the SD1's handling, and as I've written, there's a lot to enjoy, including external control buttons for most major functions, something that I appreciate on a camera. I do wish though that the SD1 had a proximity sensor. On cameras like this that no longer have a top panel LCD for observing how the camera is set, it is annoying to have the rear LCD remain illuminated once the camera comes up to the eye. On the Sony A900, for example, the sensor blanks the display when the camera is brought to eye level, which is most welcome, and something that while using the SD1 I wish it also featured.
Standout. Clearview, Ontario. July, 2011
Sigma SD1 with 120-300mm f/2.8 @ ISO 200
(NB: This image has been desaturated and toned in Lightroom)
A great sensor deserves great lenses. It should go without saying that one will be wasting ones money buying an SD1 and then attaching an inexpensive under-performing lens. I don't have enough personal experience with Sigma lenses to recommend which ones to buy and use. But, the ones that I was loaned and did my testing with all seemed up to the task, and indeed were very good indeed. These were...
– Sigma 8–16mm f/4.5-5.6 DC
– Sigma 24–70mm f/2.8 EX (suffers somewhat from chromatic aberration with the SD1)
– Sigma 70mm f/2.8 DG Macro
– Sigma 70–200mm f/2.8 EX
– Sigma 120–300mm f/2.8 EX (an exceptional lens)
I was particularly impressed with the 8–16mm and especially the 120–300mm f/2.8. The latter is a world class-lens. Quite extraordinary. If it was available in Sony Alpha mount I'd place my order today. (Hint, hint to Sigma).
Creemore Farm. Clearview, Ontario, July, 2011
Sigma SD1 with 24-70mm f/2.8 @ ISO 400
IR Filter and Dust Shield
Unlike just about any other camera (excepting its siblings) the SD1 has a removable Infrared filter just behind the lens mount and in front of its sensor. This acts not just as an IR blocking filter (which all digital cameras require) but also as a dust shield. Because it is right behind the lens mount rather than laminated to the sensor surface, it is much easier to reach for cleaning, and can even be removed.
When removed one can attach an IR-only pass filter to the camera's lens and shoot colour or B&W infrared. Some people spend many hundreds of dollars to get their Canons, Nikons and other brands modified for IR shooting. With the SD1 it's as simple as removing the behind-the-lens filter, a task that just takes seconds and can just as easily be reversed.
Sigma Photo Pro 5
Sigma's proprietary raw processing software, Sigma Photo Pro, is currently the only choice available for the SD1. This is regretable, because as I wrote in my 2008 review of the Sigma DP1 "Unfortunately this is very primitive software; simple to use, but very slow and under-featured". Three years later little has changed. This is still not a raw processor that I would want to spend too much time with.
None of the high quality raw programs from third parties, such as Photoshop's Camera Raw, Lightroom, DxO Optics Pro, or Aperture support the SD1. More significantly, Sigma appears not to be in any rush to work with these other companies, which is a real shame for potential SD1 owners. All the major camera makers now work with Apple, DxO and Adobe as soon as new camera models become available (or even earlier) and provide them with both sample cameras and technical data so that support can be written in a timely manner.
The solution for SD1 users is to soldier your way through the limited features and flakiness of SPP and do the bulk of your work in your favourite post-processing software. The way to do this is to set white balance, white point, black point, overall gamma, and reduce sharpening to as low a setting as possible within SPP. Then, export the file as a 16 bit TIFF in ProPhoto RGB.
Now you can import the file into Lightroom or whichever other program you prefer, and continue the processing task. Image quality is unlikely to suffer as long as you're not doing anything too extreme. An annoying but soluble situation.
Followup: I wrote the above after a few days of using SPP. A week and several hundred frames processing later I've decided that it's more than annoying. It's a royal pain in the butt. I used to think that Silkypix was a dog. Now there's a new winner – Sigma Photo Pro 5. Woof woof.
Note to Sigma: Since you are pricing the SD1 as a Pro-level product, assume that your customers want and need something with significantly more power and flexibility than what SPP currently offers. And, if you won't or can't bring it to an appropriate level of performance, then please allow others to support the SD1. This camera deserves better, and so do its users.
The Question of Sharpening
There is a sharpening slider in SPP and the correct setting has generated quite a bit of debate with regard to the SD1. Firstly, because the camera does not have an AA filter images appear quite a bit "crisper" than those from a typical Bayer camera that does have one. This means that quite a bit less sharpening is needed.
Also, because of the special resolution characteristics of the Foveon / X3 sensor images seem to have a lot more microcontrast than typical Bayer images, which also mitigates against too much sharpening.
I discussed this with Laurance Matson, and experimented at some length on my own. Laurence is of the opinion that somewhere between -0.4 and -1 is an appropriate setting. Some early users are reportedly happier with -2, but they don't then say what they do afterward.
What I have concluded is that what one should do is set SPP at -2, which seems to be the equivalent of no sharpening. The reason for doing this is because sharpening of SD1 files seems to vary considerably depending on the subject matter. Some images require more or less than others, just as with Bayer sensor cameras.
My suggestion therefore is set sharpening to zero (-2 in SPP) and then sharpen to taste in your post processing software depending on the subject. I also advise using both Input Sharpening (which we've been discussing) and separate Output Sharpening, depending on whether the image is going to the web, offset printing, or an inkjet – and even the type of paper being used. This is the workflow invented by sharpening guru Bruce Fraser, and is what Jeff Schewe and I recommend in our Lightroom, Photoshop, and fine art printing tutorials.
Growing. Clearview, Ontario, July, 2011
Sigma SD1 with 8-16mm @ ISO 100
Both the camera settings and SPP have something called a Color Mode. These include Standard, Vivid, Neutral, Portrait, Landscape, B&W, and Sepia. These are nothing more than metadata tags on the raw file. They can be changed in camera before the fact, or in SPP after the fact. But, I find Standard to be a bit too punchy and Neutral to be too insipid. Therefore if I set the camera to Standard all the files come in looking like Velvia and with Neutral like Astia on an overcast day.
But there are other problems as well. Overall gamma is flat and even in Standard mode images look really dreadful in SPP. To compound the problem SPP is such simplistic and primitive software than customizing the "look" isn't terribly easy, which means that without loading each file separately and adjusting it appropriately it's tough to judge what you're working with.
If Sigma were to adopt industry standard ICC or DNG profiles this wouldn't be an issue, but I wouldn't bet on that ever happening.
Until Sigma fixes and enhances SPP, do yourself a favour. Set white balance, black point, and white point within SPP and then export as a 16 bit TIFF in Prophoto RGB to a decent program such as Lightroom or Photoshop's Camera Raw to do the balance of your processing. (As we will soon see, there are other problems. The colour palette of the SD1 is way off norm, and without the ability to use profiles there's no easy way to fix things).
The IQ Evaluation Set-Up
The testing and evaluation of the SD1 took place at my print studio near Creemore, Ontario during the second week of July, 2011. A Mac Pro was used along with a NEC Multisync PA271W monitor, calibrated and profiled with Spectraview II software, using an Eye One Spectrophotometer. All files except those from the Sigma SD1 were processed from raw to print using Lightroom 3, while SD1 files were first raw processed and white balanced with white point and black point set in SPP, and then exported to Lightroom in 16 bit ProPhoto RGB. Sharpening in SPP was turned down to -2 (off), with subsequent Input sharpening and print sharpening done in Lightroom according to my normal practice.
Evaluation prints were made on Canson Bartya Photographique (one of the widest gamut papers I know of) using profiles that I made myself with the Eye One spectophotometer. Printers used were the Epson Stylus Pro 9700, Stylus Pro 3880 and R3000. Visual evaluations were made on a D65 balanced GTI viewing station. This is all the same gear used in the brand-new edition of our Camera to Print and Screen video tutorial, done with Jeff Schewe.
Our studio desktop test suite was lit with three Elinchrom strobes using a mix of grids and softboxes.
The Eyes Have It
The world is divided into two main groups when it comes to the testing and evaluation of photographic equipment – those that insist on numbers, and those that prefer the evidence of their eyes.
I sit astride the two camps. I used to be a numbers kind of guy. I was one of the beta testers of DxO Analyzer which is now widely used by magazines and web sites for their reviews. For a couple of years I used it here on LuLa for quite a few product reviews, but in the end stopped doing so. This was for two reasons. Firstly, it was a lot of work (time which I would rather spend shooting and writing), but more importantly I found that there was a disconnect between what I measured and what I was seeing. And, since in the end photography is a human visual experience, I decided to trust my eyes rather than charts and graphs.
Now, don't get me wrong. I still find value in reading technical analysis of sensors and lenses. I just don't like doing them myself anymore, and in the end I always trust the evidence of my eyes over anything else. I would point out as well, that while measurements have their place, I wouldn't judge a fine wine by means of a chemical assay, or a great musical instrument by looking at an audio-scope waveform. The eyes indeed have it.
Rusted Out. Clearview, Ontario, July, 2011
Sigma SD1 with 701-200mm @ ISO 800
In Its Own Right
We will get to a comparison with other cameras shortly. But first, we need to look at the SD1 on its own merits.
I spent just over a week working intensively with the SD1. I must say that it was a very enjoyable experience. Not because it's a particularly great camera. In fact it isn't. It's simply "OK" as a camera. A decent enough mid-range APS-C sized DSLR that's nicely built, has capable ergonomics and control interface, though a bit frustrating because of its slow write times.
What made it so enjoyable though was loading the files on the computer and making prints. After working out a workflow that got me out of the dreadful Sigma Photo Pro as quickly as possible, I loved working with the files. They are that good.
Describing how the SD1 differs from other cameras (except its Sigma siblings) involves the use of similes and adjectives. Its akin to describing the taste of a meal, a new wine, or listening to a musical performance. That the steak was at 140F, the wine had a PH of 6.2, and the music peaked at 16,000 Khz, tells you nothing of the actual experience.
So, with that said... here are some observations from the notebook that I kept with me during the week of testing...
- Resolution is remarkable.... Surpasses (by a small but noticable margin) the Sony A900 and A55 on-screen and at typical print sizes...
- Not quite in Leica M9 territory though...could be the lenses as much as anything else
- The Pentax 645D is in another league. The SD1 is no competitor to even the lower end of the current medium format market
- What impresses most about the SD1 is micro-detail... clearly a result of not having at AA filter, as much as anything else
- This may be what the so-called 3D effect that people talk about is caused by. There is detail within detail
- These files can really be ressed up.... prints ressed up by 2X show little to no deterioration
- Auto ISO sucks. Don't use it.
- Sigma Photo Pro 5 sucks. Don't use it except when and as necessary. WB, Black and White point only.
- Some people say that the SD1's colour palette looks like Kodachrome. No. It's actually broken. Looks to me like Agfachrome from the 1970's.
- Warmer and somewhat unnatural.
- There is a smoothness to colour transitions that is surprising given that it's only a 12 bit pipeline. Looks more like 14 bit. Part of the "Foveon look"?
Sigma SD1 with 70-200mm @ ISO 800
You may have noticed that most of the sample shots shown for the SD1 are of rural and farm landscapes. The reason for this is that's where I was during the week that I had available for testing. But skin tones are very important, and so lacking a pretty model, Laurence Matson was pressed into service as a skin model.
As you can see, the SD1 did very well, in large measure because of its warm tone bias. A problem arises in dark areas such as his forearm and the top of his head where three quarter tones can take on a nasty redish cast. This can also be seen in some of the landscape shots. Until Sigma fixes their SPP or allows third parties to do raw processing of the camera's files and thus allow profiles, this can only be fixed on a local basis.
Above is a 100% crop for those who wish to pixel peep. Needless to say that a large print of this portrait looks extremely good, and do note that this was taken at ISO 800, with only SPP's standard noise reduction applied. Excellent performance.
Full Frame of Test Set
The frame above was set up to show ISO performance. It contains a mini Gretag Macbeth colour checker, some fine detail and some out of focus shadow areas. Exposure was set so that the right hand white square was at about 200; just short of clipping. White balance was set on the third gray square from the right. Sharpening was optimized within Lightroom after having been exported from SDD at -2. Below are 100% crops at each available setting.
As can be seen, ISOs 100 through 800 are excellent. 1600 is marginal, and 3200 is pretty much a write off. Not bad high ISO performance, but not up to the current standard of mainstream sensors. For all its IQ strengths, this simply isn't a low light camera.
In Comparison With
We decided to include four quite different camera systems to use in comparison with the SD1. These cameras were not in any way to be compared in terms of operational features, but simply with regard to image quality. Sigma has made some pretty significant claims with regard to the SD1's image quality, and priced the camera accordingly, so it seemed worth having at least four points of comparison. We chose the Pentax 645D, Leica M9, Sony A900, and Sony A55. Here's why...
A medium format camera might seem like a strange comparison bedfellow for an APS-C sized DSLR like the SD1, but maybe not. Sigma calls the SD1 a 46 Megapixel camera. In terms of the rest of Bayer cameradom, it isn't. But then a Bayer sensor camera doesn't have the effective pixel count that it claims when compared to a sensor without colour matrix decoding.
If we assume (and I did going into this test) that a Foveon sensor actually has comparatively about double its linear resolution (so about 30MP for the SD1) and a Bayer equipped camera such as the 645D has about 1/3rd less effective resolution than its actual pixel count (again about 30MP for the 645D), this could actually be a competitive IQ horse race. For, as Sigma themselves have written...
"The luminance resolution of this sensor is, in fact, equivalent to that of a 30MP CFA sensor as measured on the standard B&W resolution chart used in conventional digital camera resolution testing." – Sigma Web site
One of the claimed benefits of the SD1 is that its sensor doesn't require an anti-aliasing filter to block moire. Well, the M9 is a full-frame 18MP camera that also doesn't have an AA filter. It is priced in the same ballpark as the SD1.
The A900 has a 24MP full frame with a Bayer filter. In this comparison it therefore is a counterpoint to the lack of an AA filter with the Leica M9, though the A900 has a full increment higher resolution than the Leica. Features aside (as is the case with all of these comparison cameras), the A900 is priced well under $3,000, so quite a bit less than half the price of the SD1.
I wanted to include a moderately priced APS-C camera. Canon, Nikon, Pentax, all have offerings that could be used, but I happen to own a Sony A55 and a selection of top Sony and Zeiss lenses, so it was an appropriate choice. Also, the 15MP sensor in the A55 has almost the same linear resolution as that in the SD1, and this same Sony sensor is used in quite a few current DSLRs from other makers, (Pentax K5 and Nikon D7000).
Are these four the most appropriate cameras for comparison with the Sigma SD1? I think so, for the reasons given above. The A900 serves as an IQ surrogate (at least at low to medium ISOs) for cameras like the Canon 5D MKII, 1Ds MKIII and Nikon D3x. The A55 has a state-of-the-art Bayer equipped sensor and delivers similar overall IQ at 15MP to other leading mid-level DSLRs.
In any event, these were what we had available for image quality comparisons, so are what we used. Let's see what happened, shall we?
Sigma SD1 / Sony A900
Let's start this with something that helps identify what it is about the Sigma X3 processor that makes it different, if not special. The above shots were taken at the same time as the ISO tests shown above. The one on the left is the SD1 and the one on the right the Sony A900. The reason that I chose the A900 for this comparison is because it has a 24MP sensor with a moderate strength AA filter. These were shot at ISO 200. The SD1 shot was with the Sigma 24-70mm f/2.8 and the Sony with the Sony/Zeiss 24-70mm f/2.8.
Of course there are unavoidable differences. The larger and higher resolution sensor on the A900 produces a higher magnification image at the same frame size. There are therefore obvious depth of field differences and also each camera renders a different palette.
What I see is that the SD1 is more contrasty and has more saturation in some colours and less in others. As I've said, it's a different palette. The A900 rolls off more smoothly in the highlights. But, when it comes to resolution the lack of an AA filter shows though. The SD1 clearly resolves more micro-detail. This is one of the reasons it is able to punch above its weight when it comes to making large prints. There is simply more detail in the file.
Now, the advantages of not having an AA filter are well understood, and we see a similar resolution advantage with the Leica M9. But the advantage that the Sigma's X3 sensor has in this regard is that its lack of an AA filter is not compromised by colour moire, the way that it potentially is with the Leica M9 since it uses a Bayer filter array. Is this benefit worth the price – both literally and figuratively? That's for us each to decide for ourselves.
Comparison Set Up
All shots were normalized as much as possible, with exposure equalized for the lightest white square and white balanced set in SPP and Lightroom for the third lightest square.
When it comes to the SD1's palette it's clear that it is radically different than the other four cameras. Whether this is the palette that Sigma intends the cameras to have or is caused by the SPP software being flaky, it's hard to know. But, since this is a shipping camera and not a pre-production sample one can only assume that it is intended.
The SD1 is clearly heavily biased toward a warmth not seen in the other cameras, which are noted here for their similarities rather than their differences. This goes a long way to explaining why the sample images that I posted on the site's Home Page in the week prior to publication of this review, and which are also now seen on this page as well, were described by some as having a very appealing look. It's the soft warmth, which I see as being reminiscent of Agfachrome from the 1970's. But, it certainly isn't accurate.
Below are 100% crops from each of the cameras. It should go without saying that exposures were equalized, focus was double checked, the cameras were tripod mounted and a studio flash setup was used. The images below are of different sizes because resolutions ranged from 15MP to 40MP. All files were visually optimally sharpened in Lightroom. (Sharpening in SPP was set to -2 and then proper sharpening was done in Lightroom, since 0 sharpening in SPP is too strong).
The three cameras without an AA filter (SD1, M9, 645D) clearly produce sharper files. The A55 and A900, less so. Of course there are lens differences, since it was impossible to use the same lens on all cameras. But, in each case the lens used was a prime, and the aperture used was optimal. No surprises.
Sigma SD1 / Sony A900 @ ISO 6400
Some people may not be aware of it, but Color / Chroma Noise and its associated removal tool in various processing programs is there to remove noise introduced by Bayer Filter Array decoding. One would therefore assume that a Foveon type X3 sensor, which doesn't use a Bayer Matrix, wouldn't have Color Noise, and indeed it doesn't. I had not looked at this before, but when I did I was curious to see how it would clean up at high ISO.
Above is a 100% cropped segment from the SD1 and (arbitrarily, though it is the noisiest of the non-X3 test bunch at high ISO) the Sony A900, both at ISO 6400. I have cleaned them both up as best I know how using the new noise reduction tools in Lightroom 3, which are about as good as it gets outside of a dedicated NR program.
The results are interesting. There is more perceived detail in the Sigma frame, (due to shallower DOF don't judge the colour chart for sharpness) but the nasty colour noise in the white pussy willow can't be removed, and seems to be some sort of X3 specific artifact that I've not seen before. This test doesn't prove much of anything, other than that I really wouldn't want to use either camera for low light shooting beyond ISO 800.
What We See
There has been a lot of nonsense promulgated over the so-called 3D qualities of Foveon / X3 images. I now understand (I think) what people have been talking about, but there really is no magic involved. There are also some issues that are relevant to the current version of SPP software.
- The SD1 does not have a blurring (anti-aliasing) filter. When used with a very good lens this allows extremely fine micro-detail to be recorded, creating prints and on-screen images (sometimes) with a feeling of greater depth and dimensionality. This isn't unique to the SD1 or other Foveon / X3 cameras because it isn't a function of this sensor technology; it is simply a result of not having a softened image caused by an AA filter. This is also seen with the Leica M8 / M9, which similarly do not have an AA filter, and which many users claim have a comparable 3Dish quality to their files. Indeed the absence of an AA filter is part of the appeal of medium format cameras and backs, and in the above comparison series is seen as well with the Pentax 645D.
- The X3 technology of the SD1's sensor means that there is no colour aliasing. Fine if you're shooting fabrics, but not that critical for most users. Where it does seem to play a role is in not requiring any Chroma noise reduction, since Chroma noise is an artifact of the Bayer sensor. Even at moderate to low ISOs there is some chroma noise, possibly even below normal visual sensitivity. But (and this is a conjecture on my part) its lack may play a role in the "look" that Foveon / X3 fans enjoy.
- It's worth noting that while a Bayer filter camera interpolates its image data, so too does a Foveon X3 sensor when one wants to make prints larger than native size. For the first time with the new SD1 model though, with its usable native 15MP size, up-ressing may not be needed when making all except large prints.
- Sigma Photo Pro 5.0.1 for the SD1 (the current version as of the camera's release, and of this review) is seriously broken. Some people are complaining of crashes (Mac), but I only experienced this once in a week of daily use. There are other more serious problems, including over-zealous sharpening (turn down to -2 and sharpen elsewhere), and a much too-warm colour bias. The ability to use either .ICC profiles or .DNG profiles is completely missing from SPP, and since no other raw software can handle SD1 files, there is no way that one can make one's own profiles. No amount of manual tweaking of colours can fine tune the camera's current palette to anything resembling normalcy, so users who like the 1970's Agfachrome look will be pleased, while others are SOL (Seriously Out of Luck) for now.
Let's be frank. The SD1 is ridiculously overpriced. It's a decent APS-C camera with mid-level built quality; at about the $1,000 level. It has neither the features nor specifications of a Pro camera such as a Nikon D3x or Canon 1Ds MkIII, let alone a 5D MKII or Sony A900. And with new 30+ Megapixel versions of these three company's flagships coming in the months ahead, along with not having video, no Live View, no tethering, and no pro support program, the SD1's price is simply untenable. The camera's only calling card is potentially its image quality.
As we've seen there really is something attractively different, even somewhat special about the SD1's image quality. People have been claiming this for years with earlier models, and now we see it on a more mainstream model.
Is it magic? No, of course not. Is the SD1 better than any other mid-range DSLR. In terms of function and handling – no. In terms of image quality, it might very well be. And therein lies the quandary.
Its current retail price of $6,900 is clearly indefensible. The SD1 simply isn't, to my mind at least, worth that much money on any objective basis. This price makes it a $1,000 camera body with a $6,000 sensor. Fail.
If the camera were priced at under $2,500 it would sell by the tens of thousands, worldwide, and would likely generate significant new lens sales. At its current price Sigma will be luckily to sell a couple of hundred a month. Sadly, this is also likely an unsustainable level from a manufacturing point of view, particularly with regard to the sensor. No fab can produce such small quantities of a bespoke sensor at competitive prices.
Canola Field #1. Malancthon, Ontario, July, 2011
Sigma SD1 with 70-200mm f/2.8 @ ISO 800
The Bottom Line
I like the SD1 – a lot. It's quirky and annoying, has truly awful raw software, and the slow write-times and unresponsiveness during card writing cause a lot of teeth gnashing. But, the image quality is superb (colour accuracy - not so much at the moment). No, it's not a 45MP or even a 30MP camera, as some claim, but it does out-resolve several 24MP cameras on large prints due to its lack of an AA filter. On smaller prints and on the web its "goodness" can still be appreciated. The camera's flawed colour palette and current inability for the user to easily fix it is annoying to say the least, but hopefully will be addressed by Sigma soon.
I'll simply conclude by saying that if the SD1 had been priced at about $2,000 I likely would have bought one. I find its image quality appealing, and though the last thing that I need is another camera and set of lenses, I find it capable of producing a look in prints that I really like. But unrealistically priced as it is, no sale.
Whether you should buy the SD1 at its current price is a matter between you, your heart, and your bank manager. I would certainly suggest though that you find a dealer that has one for you to test if an SD1 is giving you the hots.
In the end I cannot give the SD1 a Tested and Recommended ranking, because though it has much to offer, its current price precludes it from being a competitor. Your seven thousand dollars would likely be more productively spent buying a Leica M9, Nikon D3x, or Canon 1Ds MKIII. For a couple of thousand more you can even buy the medium format Pentax 645D, which with its 40MP sensor and no AA filter runs rings around the SD1 in terms of IQ, and is a far superior camera in every other respect.
Let's hope that Sigma gets the memo and re-prices the SD1 so that it can enjoy the broad audience that it deserves. Its image quality really is (potentially) that good.
By Nick Devlin
Nick was involved in the first couple of days of our SD1 testing and then subsequently
spent a week working with the camera's files together with those of the others cameras used for IQ comparison.
Sigma has long marched to its own drummer in what has become a rapidly homogenizing photographic market. Consistently pioneering the design of ever-more exotic lenses, Sigma has also become the sole custodian of Foveon technology – a technology that should, in theory, put the inherently improvisational Bayer-matrix into its grave. As such, I was very excited to join in testing the SD1. It’s 15MP Foveon chip represents a more than three-fold increase in pixel count from the previous Sigma cameras, and has been touted as a match for even the lower range of medium format. Heady stuff indeed. So what is the reality?
Image quality is obviously the ultimate question. (I must qualify my observations by stating that I had a shorter time with the camera than did Michael, and thus my observations carry less weight than his.) To my eye, the SD1 essentially matched the performance of the 24.5 megapixel A900, yielded slightly to the Leica and was clearly bested by the 645D. On-screen, I agree with Michael that the SD1 files had a slight edge in fine to very-fine detail over the A900. In print at 13x19, however, that difference was hard to perceive. Equally, however, the 645D prints contained notably more detail when viewed on screen but this advantage was again largely lost in print at 13x19. In truly large prints (the kind everyone on the forums claims then ‘need’ but likely almost never make), the 645D pulled well-ahead, while the SD1 and A900 remained within a lick of each other to my tastes.
Put in the context of an APS-C sensor, this is a terrific performance. Its significance is particularly acute when one remembers that APS-C sized sensors now inhabit some truly compact cameras, such as the much-vaunted Fuji X100. Sigma adapted its last generation sensor into the DP1, and a similar effort with the X3 chip could create a fiercely powerful compact camera, and a serious bestseller.
This next paragraph should contain the conclusion that a +/- $2,000 APS-C camera can now compete with the very best of what full-frame 35mm dslrs have on offer. Alas, for reasons we all know, that sentence will not be written. As Michael concludes, this is a really fine imager. It is a fine evolutionary step in the development of digital camera technology. Now for the cold water. This performance still falls well short of the three-times megapixel multiplier claimed by Foveon’s evangelists. This is not a 45MP camera, and no one should say it is. Nor is it a competitor for medium format in any application where the difference of MF is likely to be seen (truth be told there aren’t that many). I would also speculate that a lot of unrealized potential lies within the SD1, locked there by the inadequacies of the Sigma Photo Pro software. This software is functionally unusable (like every other camera makers’ crappy software), and makes getting the best of the camera challenging. The sample prints produced by Laurence − arguably the most experienced and proficient Foveon file-wrangler – are better than what I think we saw from this camera, suggesting that workflow is especially critical. Sigma simply must get on board with Adobe (and perhaps Phase One) to bring their files into the mainstream of processing. I venture to say the commercial viability of the product depends on it.
The other half of the equation is, of course, the camera qua camera. The best description I can give the SD1 is that it’s a bit ‘rough hewn’. There’s nothing overtly bad about the camera. Indeed, it functions very competently. But the User Interface [UI] is in the middle of the range for comparable offerings.
It fits nicely in the hand and focuses very well. It also has a beautifully dampened shutter. If I were committed to the sensor, I could get used to it fairly easily. But, unlike the Pentax 645D, it would never give me joy to use as a camera. It’s ‘ok’. Clearly, Sigma’s engineers were devoted to ‘weaponizing’ the x3 chip, and only peripherally concerned with building the best camera they could, and fair enough.
That should be the end of the discussion for now, because ‘ok’ is about all you get in the APS-C DSLR market. For $1,200 to $2,000 that’s par for the course. But the SD1 is not being sold as such. Rather, it is sold, or rather attempting to be sold, as a premium product. That is a whole other game.
A premium camera must be both rugged and lush. It must feel like it is largely indestructible, but at the same time rich in its textural and sensory rewards. The D3x epitomizes this. So does the 645D. These cameras feel GREAT in the hand. Their controls are not merely functional, but feel good to use. Their screens and finders are a crisp delight to the eye. They bespeak quality at every touch. People who pay that much for a camera expect this.
The difference I am describing is found in the comparison between a $2,000 Canali suit, and a $400 suit from a good department store. From ten feet away, they look roughly the same. But to the wearer, the fine garment is unmistakable. It slips on like a silk glove and hangs weightlessly. Its buttons are rich and luxurious to close examination. It has an obvious and intentional style. The SD1 is the $400 suit: perfectly serviceable, but not luxurious. EVEN IF the camera’s quality rivaled that of medium format, I doubt many purchasers would be satisfied with the product at its present price, simply because it feels like a Ford with a Lexus price tag. I repeat, there is nothing wrong with this camera. It is just fine. But it is just fine.
Bottom line: Sigma has done something special and interesting. They must, however, realize the product’s place and potential, and price it appropriately. If they do so, the SD1 will be a success, and the next generation Foveon camera will be even more exciting. If not, their venture will die here.
Nick Devlin is a barrister and photographer in Toronto, Canada. He works as a Federal Prosecutor, specializing in major drug and terrorism cases. With almost twenty years behind the lens, Nick worked extensively as a photojournalist and pro sports photographer before turning to the law. Presently, his main visual interests are urban landscape, portraiture and travel photography.
For an additional perspective on the Sigma SD1 you may wish to read
Sean Reid's insightful review at http://www.reidreviews.com/.
Sean's site is subscription-based and well worth the modest investment.