Sunday, October 28, 2012

Thoughts about Sigma and DP2 Merrill Review

Short Introduction

Sigma is an oddball company. Starting with its cultish following that keeps calling Bayer sensor 'blur' despite the fact that all great photographers of today manage to create great images with it; it is a company of hollow marking claims about Foveon sensor, with optimistically creative pixel count and unclear claims of color accuracy; finally, products such as SD1 which was priced in medium format range, but dropping to 20% of its initial price in mere few months.

But the way I see it, the biggest problem that Sigma has is the lack of significant user base among professional photographers. Even going to Sigma website leaves me perplexed - they could not even pay someone to use their camera? Most images showcased are shot with Sigma lenses, but not Sigma camera. Moreover, they are nice, clean images, but hardly the kind of stuff that captures people imagination.

Sigma DP2, San Francisco
But all of these is the evidence of manufacturer that cannot find a clean story to sell their product, rather than a fundamentally bad technology. Plenty of cameras, such as NEX-7, are not widely used by professionals. This does not make them substandard, just not professional grade. Cultish followers and pricing anomalies are even more removed from the question about the true quality of the sensor. And ultimately I and many other users are willing to put-up with bad cameras and clumsy marketing if ultimately there is something to show for the efforts.

Sigma DP2, Belmont CA

About all Sigma cameras

I have used some five Sigma cameras up until now - DP1, DP2, DP2s, SD14 and now DP2 Merrill. I found each one of those cameras awkward but straight forward to use - thanks to very few features. For me no Sigma would be the only camera to have for a shooting occasion. This is not as much for the limited usable ISO range (and limited it is) or lack of advanced features, but because of unpredictable, artifact full images. With most other mirror-less cameras I can predict what the result will look like, but I have never gotten to that state with Sigma. DP2 Merrill is no exception in that sense. Sometimes images will excel your expectation, but more often some crazy things will happen...

Sigma DP2, Florida

It would be easy to dismiss Sigma as the company who bet on the wrong technology and lost. I think in the long term this is correct, but at present it is too early to bury. This realization for me came from the analysis of my images that I choose to display on my website. When I count how many of my favorite images were made with Sigma count is about 20%. That is given that I could not have shot more than 1-2% of all the images with Sigmas. This is very unreasonable fraction for a failed brand. The other cameras that I use are not exactly budget point-and-shoots - they are some of the most recommended enthusiast/professional cameras - Nikon D700, Sony NEX-7, Fuji X-Pro1, Olympus E-PL1, Nikon D600 and many more, some 40 cameras in total.

Sigma DP2, San Francisco

So camera is odd, featureless, usable only at base ISO - there must be a reason why so many images look good. I find it unhelpful when people talk about '3D effect' but never bother to demonstrate it or even explain exactly what it is.

I think I have an idea what it is that looks good for me, and I will try to be more precise than - oh, that special je ne sais quoin. One very distinct case is color reflections in the water or glass. This sounds insanely specific and of little use, but this is where I see the most of the difference, and when those images come out good they are great. While lights reflected in the water look good with any camera, none look as good as with Sigma. Most of the time reflections loose color and sharpness, but not with Sigma. You get those thin, brightly colored lines as if drawn with a pensile directly on the water. Look at the image from Singapore for example.

Sigma DP1, Singapore

In general this ability could be formulated more broadly - ability to retain color of the bright light. My other cameras make color patches white as their intensity increases, while Sigma retains the color and tends to saturate color channel rather than making it white. This property can cause Sigma to blow red channel, but can also be very useful in other settings.

Sigma DP2 Merrill, Los Altos Hills, CA

If one wants to have amazing experience with Sigma he/she just needs to seek out opportunities for Sigma to excel, and avoid all other situations where Sigma is likely to fail. In general it could be formulated as a simple set of rules:
  • Do not be caught with only Sigma camera on you.
  • Always shoot in low light with a tripod. 
  • There are two usable ISO settings - 100 and 200.
  • Always shoot RAW - JPG from Sigma are useless.
  • Make sure there are bright colored lights in the picture.
Sigma DP2 Merrill, Bay Area, California

Another application which Sigma sensor seems to be good for is Black & White portraits. It is kind of hard for me to review, since it is very rear for me to shoot B& W. But with few shots that I took, Sigma seems to be pretty good; though to be honest many other cameras are very good at this too, and in general camera is far less important in this case than the subject and lighting.

Sigma DP2, Stanford, CA

Impressions from DP2 Merrill

When I opened the camera box I was surprised that it is quite a bit bigger than non-Merrill version. I, just like everyone else, like smaller camera and this was unpleasant surprise, especially that the difference in sensor size is pretty negligible. Camera also became heavier. I hope this is not just a way for Sigma to convey that you are getting more camera for money.

At the same time there was a noticeable improvement in construction, or should I say feel of the construction since I have not opened either one of them. No more odd sounding motor extending the lens. The buttons feel solid and definitive. The control wheel is very comfortable and easy to use. Quality of the monitor is a substantial improvement over the previous models. If old DP cameras had LCD from approximately year 2000 or so, now there was a quantum leap to approximately 2006. Still far, far away from the bight, vivid LCD's of todays cameras. 

Sigma DP2 Merrill, Half Moon Bay, CA

Merrill inherited excellent quick selection menu of later DP series which makes control, if not close to pro cameras, still quite comfortable to use. Overall menu is also pretty clear thanks to not so sophisticated feature set.

Sigma Dp2 Merrill, Foster City, CA

Not much happened to the operation speed of Merrill. Perhaps there is some change, but not the one that is clear from using it. Focusing is slow. With contrast detection focusing speed is closely tied to high ISO performance and video performance. As those did not improve, there is no reason to think that focusing would either. Generally when it locks on it is pretty accurate, but more often than some other cameras it will select wrong object to focus on. 

White balance, as before, is the worst in the industry. How bad it is defies expectations. Colors sometimes are truly crazy. At the same time custom WB is not too bad and comfortable to use. Oddly enough start-up speed is pretty decent - much better than Fuji X-Pro1 for instance; formatting speed is one of the best ones I have seen! This was odd. 

At the same time speed at which images are processed by the camera and dumped to the card are distinctly class trailing - if there is such a word - in that respect it is just like medium format camera. I have tried to use it with the fastest card that I had - 95MB/s - no help. The only time I was forced to be more careful before pressing a shutter button is when I used film medium format camera and each exposure was $3 to process.

It does not look like there was a huge improvement in artifacts realm as well. In fact it took just one try to see the purple spots that Sigma is infamous for - F16, point, shoot and ...

Sigma DP2 Merrill, Half Moon Bay CA
In addition to magenta patches that show-up if you close the aperture, there is also green/magenta glare that shows up about 50% of the time when sun is in the picture. 

Those effects are common enough, and repeatable that sometimes this would make me abandon the shot due to the persistent problem at a particular angle. 

Basically the improvements are confined to just the resolution. In this case we got a crazy tripling of pixel count! Not unlike jump from Nikon D700 to D800. Given the amazing sharpness of Sigma lens/sensor and the fact that you truly get 15MP this time, one can do quite a bit of cropping and still have image perfectly usable for the web. Wave image is the example of that. To get to that image I had to crop to about 90% and was still left with quite usable 1500x1000 pixel image.

While brilliant in the conditions described in the previous section - light reflections in dusk - it is no better than its predecessor. I used Merrilll for couple of day trips and found general quality of images during the bright day light below those produced by Fuji X Pro or Nikon full frame cameras. I do not mean in terms of resolution, but rather in terms of color cast and how much can one squeeze out of a RAW file. Though generally Sigma file allows for good range of corrections, it is very hard to keep image looking natural while making this adjustments. Of course, the fact that Merrill cannot be used with Photoshop is a significant drawback - hopefully to be corrected at some point soon. 

Point which I did not think there is a need to touch is the battery life. It is not good and this fact is recognized by everyone, including Sigma itself which includes two batteries with every DP merrill. I am sure there a hot debate to be had about exactly how inconvenient it is to have several batteries. The reason why I mentioned this point here is due another unexpected problem - battery performance in the cold. It is well understood that batteries need to be specifically designed to operate in cold weather. During my last trip to Tahoe lake for skiing, I found out that fully charged Sigma battery last for about 20 shots at freezing weather. At the same time my Sony RX100 did not seem to think much of the weather. 

Another strange thing about Merrill is the fact that even at base ISO while in focus parts look crisp and clean, out of focus area has some strange patterns and things that appear as noise. It is odd that it did not come-up in other reviews. Out of focus area looks quite rough. 

Overall, as a user well conditioned to use Sigma products, I was pretty pleased with the camera. A bigger question is the price tag - $999 - which seems a bit high for such a special purpose camera. At the same time I have $2,000 lenses laying in my camera bag and used perhaps once a year. I could see more use from this camera. If you can get DP1X for $299, does it make sense to pay $999 for DP2 Merrill? This is a hard question, and I guess the objective answer given my use for the camera is - no. But for people who like to make 30"+ prints answer may be different. 

I believe that this is the last generation of Sigma when someone could really try to justify using it over other cameras. The chief advantage - resolution of Merrill is no longer above Full Frame cameras such as D800/E, and it is quite comparable to other APS-C mirrorless such as Fuji X Pro1. But besides the resolution, Sigma product is becoming more and more outdated - slow focusing, no image stabilization, bad battery life, no optical viewfinder, larger than other APS-C cameras, more expensive than most mirrorless, not the best dynamic range, poor ISO performance, poor video, poor white balance and color reproduction. The only step that Sigma could make without any technological improvements is making DP full frame, this will buy perhaps one more generation. It is sad for me to see Foveon to become irrelevant. I cannot help but think that Sigma as a company gave-up on Foveon by now and is not willing to spend R&D resources on further related technologies, such as sensor image stabilization. 

Merrill and flash

I am a flash photographer. If weight and set-up time was not an issue I would always use flash. I cannot wait for that special time when natural lighting is good, especially that it may never come. To me portraits almost always look better with active lighting when it is done well.

That said, there is flash and there is flash. When Sigma took flash out of its DP line, that was only a minor issue for me. Though convenient in a pinch, onboard flash of DP* cameras was very underpowered and of limited use (like all onboard flashes). I do not mind using external flash - this is my set-up for ad-hoc camera bag which I have on me most often:

Generally it is hard to find a camera that does not work well with flash. In its basic form there is nothing to support, so every camera works well. Well, not quite...

I mounted my manual flash on Merrill DP and took few images. I really wanted to see what was the maximum shatter speed at which camera and flash would synch. So I first tried 1/500 sec, and of course, a portion of the screen was covered by the shutter. So I dropped the speed to 1/60 sec - the same thing! What? Basically I got the same effect at every speed (UPDATE: it appears that compatibility with manual flash was resolved with subsequent firmware update). 

So I went back to the manual where I read: "When you pop-up built-in flash, you can use camera in this mode." Apparently Sigma forgot to tell the manual people that flash was out in this model. Fantastic! Aside from confusion about the fact that camera has no flash, the only piece of information that was relevant - you cannot change flash mode when flash is not attached. Of course, this is Sigma talk for "Sigma TTL flash is not attached." 

The same occlusion effect did not happen when I attach the flash via wireless trigger. So I was actually quite confused, and did not know what was happening... I tried the experiment with 3 different flashguns (Nikon, Metz, Sunpak). Nikon SB900 yielded the most interesting experiment. When flash was set to full power - effect was unnoticeable; and when I dialed the power down effect became more pronounced. 

What does that mean to increase flash power - increase duration during which the bulb is active. So what's going on is that Sigma shutter is not perfectly synchronized with the flash signal. When flash burst is very short, you can see effect very strongly. When flash is long, the effect is minimized the amount of light produced after shutter is fully open. This also explains the fact that effect is not present with the wireless trigger - trigger introduces a delay which eliminates the problem. Why would those be not synchronized - most likely TTL preflash is what setting the flash off. Unfortunately, Sigma manual is correct there is no way to change any settings. 

Wow! Now, I've seen everything - camera that does not support manual flash.  Once again I marvel at the fact that Sigma users are not up in Sigma's face about this after paying $1,000 for a compact. One explanation is that I could be wrong in this subsection and my camera is defective. But I do not think this is the case... 

Why is this review different from others

I cannot help but puzzle over the fact that my review of DP2 is so different from other people's. I am not a professional reviewer, and there are some things that I simply would not be able to comment on since I do not run any technical tests. Yet, even with that caveat things that I am discussing here are somewhat different from the points that other reviewers mention.

For instance, I was reading Steve Huff's review of this camera, and we clearly differ in few points. He says only nice things about white balance, I think it is extremely poor. He says nothing about the artifacts. For me those are the major drawback. Of course, everyone agrees about the amazing per-pixel resolution. 

But looking at the sample images that Steve includes in his review provides some idea about the difference in perspective. All, or at least most of them are snapshots of people shot in daylight on a bright sunny day. This is dramatically different from images that I use here - non of which is shot in a bright sunny day. Also most of my images are landscape style images, which are quite a bit different from pictures of particular subjects. Looking at his images, I also would not expect WB issues and artifacts; but those are not the types of images that I am mainly interested in.

Postfactum Clarifications

This review became more polemical then I intended. So I will repeat the generally understood truth that any blog represents a personal opinion and people should feel free to disagree. I thought that the fact that I am stating at the end that I was pretty pleased with the camera despite numerous shortcomings would qualify this as a positive review, though with many caveats. But it was not enough for most vocal Sigma users... OK, but this is not what I wanted to clarify... Here are some points that I would like to iterate over:

Some people were wondering why stating that my set of favorite pictures is some 20% Sigma, while I shoot Sigma only 1% of the time is not enough to conclude that Sigma is the superior camera.  This is very simple - I go out of my way to use Sigma camera in the way, that in my opinion, makes it shine. If I was not to do that and use it as a general camera for a variety of situations it would frankly suck. It so happens that light reflections in the water is a great subject, and is Sigma's strength. Fuji and Nikon do not enjoy such special considerations - they are more of a general camera. Some canonical shots, such as sunsets for instance, do not come out well with Sigma - I find color over saturated and blown. 
People took an issue with me commenting on Sigma community referring to Bayer sensors as "blurring". The point I was trying to make is not that Sigma is not sharper - it is, which I thought would be clear from me stating that the camera demonstrates "amazing sharpness of Sigma lens/sensor". The per-pixel sharpness of Sigma is quite unique. What I meant is the comments by some Sigma users that they are tired of Bayer blur, making it sound like there is no way to make a decent picture with a Bayer sensor - which I think is insane, given the quality of the best images produced by the rest of photographic community vs Sigma. 

Finally, my big disappointment with Sigma not having too many photographers that I would consider exceptional. After all camera is just a tool. If one can shoot a picture with an iPhone that would be great, clearly they can do much more with Sigma cameras. I would love to see someone who is better than me, spends more time behind camera and spends necessary time to learn how to optimize camera's capabilities to consistently produce outstanding images. I remember some 3 years ago I saw a website of a photographer shooting with SD14 whose images I thought were generally exceptional - I was unable to locate him/her since. I do not know why this people are not there...


  1. Overall I thought it was a fair review, but I have a few disagreements:

    1) Overall use. You seem to think it's a pretty specialized camera, but it really does a great job even in strong midday light as this set shows:

    2) High ISO. ISO above 200 is pretty usable, without even a lot of noise up to around ISO 1200. This is especially true in conjunction with your theory about strongly lit colors being captured well by the sensor, as in this ISO 800 concert shot:

    Also good though is this late afternoon shot where I simply needed ISO 800 for the shutter speed:

    3) White balance. I shoot the DP-2M on auto WB pretty much all the time. You claim that the WB is inferior to other cameras, but from my experience shooting models side by side with users of many other cameras this simply is not so. The Sigma auto-WB is pretty good, and better than most cameras - but if you want truly exact color with ANY camera you have to use custom WB, that's just how it is. So it's not like the Sigma cameras are really at a disadvantage here.

    For example is this ISO 200 model shot using only Auto-WB and no color correction. I did shot using Neutral color mode, your issue might be that you left it at the default of "standard" which I think tends to throw the real colors off a bit in order to gain saturation.

  2. Thanks you for your comment. Clearly we are exchanging experiences here and they are subjective to a large degree. We have to acknowledge that the is no "correct" way to evaluate any given picture. For some sharpens is the only thing they look at, for others (myself including) this is a feature but there are others that are just as important.

    My biggest disappointment with DP2M is that it had poor color in the best possible lighting - bright day light. I really like the images that you are linking to here. But look at this image for instance in the set you are providing:
    and compare it to this Fuji X-E1 images:
    You may think this is just a matter of simple saturation. But it is not. In every instance when I shot DP2M next to X-E1, Fuji won - every single time. I was not able to fix this in PP. And I see the same color effect in every image of the set that you provided a link to.

    This is generally not a problem with night shots, since it is hard to tell what is the correct color and hence these effects are not noticeable.

    As far as High ISO I will agree with you, but also this is not what I meant. Images are usable perhaps up to 1600. But "usable" is not a very strong compliment. If the clean sharp image is what one is after with Sigma, you can see negative effects of ISO after 200. And by 800 there is a good amount of detail smearing. I guess, what I was trying to say is that above 200 advantages of the sensor are quickly becoming negligible.

  3. The image you selected to compare:

    Is way more comparable to this image of mine:

    The image of mine you selected was taken at mid afternoon; the comparison Fuji image was from morning, and probably shot with a polarizer (mine was not). You can see another example of how the sky is more pale in the afternoon where I was shooting the mountain in this shot:

    It may be a matter of taste but I find my image much more natural. The image you posted is, to me, over-saturated - mine I could increase saturation if I liked, but when an image starts out over-saturated de-saturation just makes it look faded.

    As I said, I shoot models off and on and that provides the best direct comparison of colors - and I find there the Sigma is usually the more accurate camera in terms of color. That's inside or outside, flash to natural lighting.

  4. I am not sure how you get your information, but the image that I shot was shot around 5PM, no polarizer; though I do not think this is central to the argument.

    As I said, this is the matter of taste how saturated you like your images. Some people shoot B&W and there is nothing wrong with it.

    What is not true is that you can just crank up saturation get the same result. It does not work with Sigma.

    In terms of portraits which I also do, I get a somewhat complex picture. In some instances color is quite nice, while in others images lack magenta hue. I see less variability from Nikon and Fuji cameras. I can pretty much predict where those will fail and how. Under correct exposure Nikon does a great job, but when overexposed color saturation is being lost. Foveon is better in keeping saturation under those conditions, but I find that fidelity of the color is worse.

    Here are two examples Nikon vs Sigma:
    Notice the desaturation in Nikon image and yellow shift in Sigma.

  5. The evidence for the brilliance of the foveon cameras is right on your article, just look at each and every image, no camera could produce the sharpness and sparkle of those images, and at 100% view! The images are nothing short of spectacular and it is puzzling as to why you don't see that in all the images, not just reflections. I've had a Dp2 for years, and even considering it's low resolution, it outperforms any other digital camera I have, in terms of image appearance.

    1. i ve a 5d and dp1 and i have to say the sigma pictures has always something more special, like a "soul" . it's a 4 millions pixels but i m able to make a nice 60x80 cm with it. With the 5d, pixels doesn't look so nice after the enlargement.