I was looking for some time for an additional small camera for situations where I don’t want to carry or use my Leica stuff. This could be weekend trips with my family, birthday parties of my daughter, business trips, or shooting in an unsafe environment.
This camera has to be small, light, not too expensive, with an unobtrusive look and most important with image quality comparable to the Leica M9. As you can imagine these requirements are not easy to fulfill.
In the past, I used a Sigma DP1 for these cases and was satisfied with the result (despite some clear shortcomings of this camera). The IQ of the DP1 was comparable to cameras like the classic Canon 5D but is clearly a level below the Leica M9. The result was, that every time I got some good pictures out of the DP1, I know that they would have been even better if I had taken them with the M9 (or 5D Mark II) instead. This was a bit unsatisfying and I decided to sell the Sigma and use my M9 in most situations. But this solved the problem only partly.
There were still situations, where I felt not very comfortable carrying or using the M9. For example on business trips I often have to enter R&D areas where all kinds of cameras are completely forbidden, meaning I have to leave my camera in some dubious places. I would be much more relaxed with a P&S than with an M9 kit in these situations
Therefore I decided, that I really need another camera in addition to my Leica.
There are not many options that fulfill my requirements. I had a close look at the following cameras:
Sony RX100: I used this camera for 4 weeks. It is very small, very fast (in all aspects like power-on, AF, etc.), has a great display, a very good sensor with good pixel sharpness, good high ISO performance, and great colors and a lens with good center sharpness and nice bokeh. But my lens was not sharp on the right side, even stopped down. Therefore I returned the camera. Although I was a few days over the 30-days return period when I discovered the decentering Sony accepted the return, which proves that Sony has good customer service. Since this decentering problem seems to be rather common with this camera (many people complain about decentering of the RX100 lens on the net) I decided not to try another one. Besides this problem, the RX100 is a great camera. If you are willing to buy several Rx100 and send them back if the lens is decentered until you get a good one, this could be a great second camera. I was not willing to do this.
Sony RX1: According to the specification (FF, high-performance Zeiss lens, size, etc.) this could be the perfect second camera for me. But for my planned use-cases (leaving it in dubious places, shooting in unsafe environments, etc.) the price is too high. In the same situations where I would feel uncomfortable with the M9, I would have the same feelings with the RX1. Besides this, I fear that the RX1 is (intentionally) optimized by Sony for high ISO performance and video performance at the cost of reduced low ISO IQ. The first sample images seem to confirm this, but nobody uses RAW-images at the moment (due to the lack of useable RAW-converter). Therefore take this comment with a grain of salt. I could be completely wrong. In addition, the price in Europe for this camera is ridiculously high compared to the price in the US.
Sigma DP1M: Since I used the Sigma DP1 in the past and often prefer wide-angle lenses the DP1M (with a 28mm equivalent lens on FF) was a natural choice. I preordered this camera some months ago. But when the first sample images appeared on the net I was not so sure anymore. The center sharpness looked great, but corner sharpness was rather weak (stopped down ok, but not great) and the lens suffered from CA. I would probably accept all this if I had not seen the DP2M images.
Sigma DP2M: The DP2M uses the same Foveon sensor (see below) as the DP1M, but a different lens: a 2.8/30mm lens (equivalent to 45mm on FF). This lens has a nearly perfect performance with great corner sharpness wide open, nearly no CA, nice bokeh, and is a perfect match for the sensor of the DP2M. Although I would prefer the focal length of the DP1M there were a lot more sample images of the DP2M with a difficult to describe wow effect, that I’ve not seen in the DP1M images. Therefore I decided to cancel my preorder of the DP1M and buy a DP2M instead.
Compared to the Leica M9 the DP2M feels like a toy. It is quite light for its size and there is some plastic used in the construction. Besides this, it is well built and clearly a level above typical point and shoot cameras.
It is quite a bit larger than the first generation DP cameras, but the size and weight is still acceptable for this kind of camera. I like the concept that it has a fixed and stable, non-retractable lens with a 49mm filter thread (and no need for any filter adapter tubes).
The display is ok, but not great. Clearly not in the same league as the newer Sony displays but better than the old DP displays.
Writing files to the memory card is very slow, but since this is not an action camera I can live with this limitation.
Battery life is very short. Some people reported on the net that they get only 20-40 images out of one fully charged battery. My experience is, that I get at least 100 images out of one charge. Not great, but since spare batteries are light and cheap, this is not a huge problem.
AF of the DP2M is very slow but in most situations also very precise. MF is compared to other modern cameras with lifeview terrible. If you use the 12x zoom everything is blurred and it is nearly impossible to focus, at 8x zoom it is a bit better, but clearly not as good as with current Sony cameras. I did a comparison between MF with 12x zoom and AF (both from a tripod) and in most situations, the result with AF was sharper.
One of the special characteristics of the DP2M is, that it uses a rather unconventional Foveon sensor. The sensor size is APS-C as in many DSLRs and high-performance compact cameras.
But normal digital cameras use a color filter array consisting of a pattern of red, green, and blue patches (Bayer-Pattern) in front of the grid of photosensors to generate the color information. This means that one pixel can only detect either red, blue, or green light and the other colors are generated afterward in the postprocessing by interpolation of the neighboring pixels. This leads to a reduced resolution compared to a black and white image sensor without the color filter array. In addition, this could also lead to moiré artifacts, which are often avoided/reduced by using an anti-aliasing filter in addition to the color filter array which further reduces the resolution. The Leica M9 has no AA filter but suffers from color moiré artifacts in certain situations.
The DP2M uses no color filter array, but instead a special sensor design where every photosite records the complete color information. To realize this, the effect that different wavelengths of light (and therefore different colors) penetrate silicon to different depths is used. Each photosite of the Foveon sensor consists of three vertically stacked photodiodes, which responds to a different wavelength of light.
If you read out the energy of these photosensors and combine this information you get the complete color information at one photosite. No interpolation is needed and therefore no antialiasing filter.
The current Foveon sensor used in the DP2M has 14.8 MP, but delivers a resolution comparable to a good standard Bayer-Pattern imager with ca. 25-30 MP. This means the resolution of the DP2M should be better than nearly all current 35mm FF DSLRs and only slightly below the state-of-the-art 36 MP Nikon D800.
Software and Tools
One problem with the Foveon concept is, that the conversion of the raw data to a viewable color image is not easy and completely different from the processing of Bayer-Pattern images. The main problem that ideally the photodiodes responding to the lower wavelength light should not respond to light with higher wavelength. The photodiodes sensitive to the blue light should be blind for the red light. But this is not the case. Since the light with higher wavelength also has to pass the upper regions of the silicon it has, of course, a significant influence on the photodiodes responsible for the lower wavelength light. This means the photodiodes responsible for the blue light also see the green and red light and the photodiodes responsible for the green light also see the red light. To get the true color information for a photosite a complex matrix operation is therefore necessary.
This means you have to use the proprietary software Sigma Photo Pro for the RAW file conversion. The format of the DP2M raw files is at the moment not supported by Adobe Lightroom. Even if it will be supported in the future it is very likely that the quality of the raw conversion is not as good as with the Sigma software. This means you can’t use your normal LR or PS workflow with the DP2M images but instead have to convert the X3F raw files with SPP to 16-bit TIF files (luckily batch processing is possible) and then import the TIFs to LR and make the final adjustments. You also have to archive both the X3F files and the TIF files to be sure, that you can do any PP changes later. And both the TIF files and the X3F files are huge.
This annoying workflow with the DP2M is the biggest disadvantage of this camera for me.
The question is, whether it is better to do the sharpening at the RAW conversion in SPP or later in LR.
Theoretical the Foveon sensor has the characteristic that colors of the longer wavelength which penetrate deeper in the silicon suffer from some diffusion of the electrons leading to a slight loss of sharpness. Therefore the red channel should need a stronger sharpening compared to the blue and green channels. Since this is not necessary for Bayer-Pattern images only the SPP software will provide this function.
On the other side, the sharpening algorithm in the Adobe software (LR and PS) is optimized over many years and is quite refined now.
To answer the above question I did several tests with different sharpness settings in SPP and LR.
The result was that even if I turned the sharpening in SPP completely off (-2) my standard sharpening setting in LR (50 / 0.7 / 100) was much too strong. I had to reduce the intensity significantly to get pleasing results.
With no SPP sharpening I got the best results with the following setting:
SPP sharpness=-2 and LR sharpness=25/0.7/100
The best alternative using the SPP sharpening was the SPP default setting 0 (which is not zero but a medium strength sharpening) combined with a very slight additional sharpening in LR (10 / 0.7 / 100).
Comparing these two alternatives I don’t see any real benefit for either of the two.
For the following comparisons I’ve used the following settings for my DP2M images: SPP sharpness=0 and LR sharpness=10/0.7/100 and for my Leica M9 images: LR sharpness=50/0.7/100.
I did some test shots to evaluate the sharpness of the DP2M compared to the Leica M9 with the Summilux 1.4/50mm lens. All images were shot from a very stable Gitzo tripod using the self-timer. For every setting I did two shots: With the DP2M one shot using MF with 12x zoom and one shot using AF, with the M9 using two independent MF procedures. Afterward, I chose the sharpest of the two images. The DP2M RAW images were converted with the default settings (colors were set to vivid in the camera menu, everything else to zero) using SPP to 16 bit TIFF files and then imported in LR where the above-described sharpening was applied in addition to a manual WB correction. When necessary, white point, black point, and brightness were adjusted to get comparable images (between M9 and DP2M). With the M9 similar adjustments in LR were done. Sharpening was done with my default LR parameters for the M9 (50/0.7/100). In addition, contrast was set to +35 and saturation to +20 to get comparable images to the DP2M.
I tested two different apertures, f/2.8 (wide open on the DP2M) and f/5.6 (the sweet spot on most high-performance lenses), and two different test subjects. The first subject was a flat wall at a medium distance (5-10 m).
This is an overview of the wall, shot with the DP2M at f/5.6, followed by the 100% crops from both cameras:
DP2M, f/2.8, 100% crop, center:
DP2M, f/5.6, 100% crop, center:
DP2M, f/2.8, 100% crop, upper right corner:
M9, f/2.8, 100% crop, upper right corner:
DP2M, f/5.6, 100% crop, upper right corner:
M9, f/5.6, 100% crop, upper right corner:
From the wall images, it can be clearly seen how good the corner performance of the DP2M lens is, even wide open and compared to a high-performance lens like the Leica Summilux 50mm. Regarding the center sharpness, there is not a huge difference between the M9 and DP2M images.
The next test subject was an overview of a city at infinity distance, shot from a hill. In addition to sharpness, this image is also suitable for evaluating the contrast and colors of the lens (more precisely the lens/sensor-combination).
First, two images of the whole subject, cropped to the same field of view (50mm), followed by 100% crops of the center and edge from both cameras.
DP2M, f/5.6, cropped to the field of view of 50mm lens:
DP2M, f/2.8, 100% crop, center:
DP2M, f/5.6, 100% crop, center:
DP2M, f/2.8, 100% crop, right edge:
M9, f/2.8, 100% crop, right edge:
DP2M, f/5.6, 100% crop, right edge:
M9, f/5.6, 100% crop, right edge:
The DP2M crops show a better sharpness, higher resolution, higher contrast, and fewer color artifacts compared to the M9 crops, both in the center and edge crops and at both apertures. Keeping in mind that the M9 has an extremely good pixel sharpness due to the lack of an AA filter this result says a lot about the IQ of the DP2M.
To identify the best aperture of the lens I did some test shots (tripod with self-timer, a subject near infinity, and non-changing light) at each aperture. I evaluated the visual impression of the images at 100% and also converted the RAW files (with identical settings) to JPEGs. The idea behind this is, that images with less detail can be better compressed by the JPEG algorithm and therefore have smaller file sizes. Here is the result:
f/2.8: 12.5 MB
f/4.0: 14 MB
f/5.6: 13.5 MB
f/8.0: 13.5 MB
f/11: 12 MB
f/16: 11.5 MB
These numbers correlate as expected perfectly to the visual impression. The best aperture is f/4, but the differences between f/4, f/5.6, and f/8 are very small and can be neglected. At f/11 we start to see some significant diffraction effects and f/16 should really be avoided.
I’ve not tested the dynamic range of the DP2M thoroughly but my impression is that the DR is comparable to the M9, which means it is good but not great. Clearly better than nearly any P&S camera but not as good as for example the Nikon D800. With the SPP tool, some additional highlight recovery is possible (compared to LR) but at the cost of an HDR look of the resulting image.
Due to the special sensor layout colors of the DP1M images are different compared to other digital cameras. Often they are not very accurate but a lot of people find them very pleasing. Often the images need some color adjustments in PP, but this is also true for the Leica images. Situations with mixed or certain artificial light can sometimes lead to weird colors and the Auto White Balance function is not very good. But with some PP effort, the result is often very pleasing.
This is clearly not a perfect camera, but I think it is so far the best compromise for my requirements:
– small and light
– discreet and unobtrusive
– IQ comparable to M9 (or current FF DSLRs)
The drawbacks of the DP2M are:
– annoying workflow (using the two different tools SPP and LR simultaneously)
– very short battery life
– colors need more correction in PP compared to other cameras
– very slow AF
– bad MF implementation (especially at 12x zoom)
I will keep and use the DP2M until something better arrives on the horizon. At the moment there are no alternatives with the same IQ in this class.
I am surprised by resume. Sharpering image looks more sharpen. DP2M & M9 has different software microcontrast algorithm and level! One can change it, for example M9 is more microcontrasted than SPP done.
What is the resume about cameras now?
My summary regarding the DP2M is: it’s a nice camera with great resolution, limited DR and a terrible workflow (if you use LR). The last point was my main reason to change to the Ricoh GR as my compact go-everywhere camera.
Thanks for your tests, am getting this camera dp2 the day before leaving on trip, and don’t have time to test it. Excellent tests, very helpful.
Big files from a compact camera.
Save an .x3f file with double output size. This gives you an 337 mb tiff! Save this in the Adobe rawconverter as 35.35 mb jpeg.
Gives you a perfect photo sized 66.38 x 99.57 (portrait) centimeter, 240 Pixels/inch.
My first digital camera, about 13 years ago, was an Canon PowerShot A5. With a compactflash card 8 mb, I could make 75 photo’s. 🙂
Thanks Kendall and Leo-Jan. When I have finished processing my Tasmania images I will do some tests regarding Standard vs. Neutral settings.
Leo-Jan, I agree, that the useability of SPP ia not very good and in the instructions of SPP at lot of important information is missing. Therefore I try to do as much as possible with LR.
Boris, lately I use SPP when there are some highlights are blown out. There is much more room to correct in SPP with the original X3F file then with the normal workflow directly to tiff. It is not necessary for all your files, but some will benefit from working with SPP before convert to tiff.
Yes, you can often get a better highlight recovery with SPP compared to a TIFF/LR workflow. But I’m still not very happy that I can’t use my usual LR workflow. If Adobe won’t support the DPxM in their next LR release I will probably sell my DP2M, although I still love it’s IQ.
SPP is far from ideal Boris, but for me no reason to sell de DP2M. I even bought the DP3M.
And I learn some things in SPP. Corrections that you use often, save those corrections, for example sharpness -2. You can save all the color modes, so you can save them to tiff in a batch. Let say that I use the batch process to tiff in batch for 90 %. The other 10 %, highlight recovery mostly, I do one file at the time. The older x3f files from the earlier Sigma DP1 and 2 are supported by Adobe, those files can not be that different from the new ones. So I think it will be a matter of time. Still, no dealbreaker for me and if I see your photo’s, oh boy they are fantastic.
Thanks for your tips Boris ans Kendall.
I already set my camera at standard, I always do, but every tip about SPP is welcome. I hate that program. When I make a preset, for example +0.3 more saturation, -0,2 sharpness (strangely that means zero sharpness…) and Croma, Luminance and Banding noise on zero ( or -3 ?), the preset remembers the first two but the Noise correction stays the same (3). And I do not want to correct the noise in SPP.
Another strange behaviour of SPP I do not understand is the fact that when I open an .X3F in review mode a couple of second the photo looks nice and then it gets much darker. This darker “version” is what you get when you directly save the file as tiff.
I also own the Sigma DP1 and DP2 and if you work with those .x3f files in SPP you can zoom your photo in review mode. This is not possible with the DP*M files.
Does anyone of you know an decent tutorial about SPP? It frustrates me every time I work with it, even the help function!
It’s not necessary to mention that I love my Sigma’s to pieces and that in the end I still am amazed with the results. 🙂
A nice review with lots of great test images, I have a few comments from owning both a DP-1M and a DP-2M
1) If you want color accuracy, shoot in Neutral color mode – you had said you were using Vivid. My experience has been that the other modes shift colors at times in ways you don’t want (though some people I know do like to use Landscape mode sometimes). If you find the colors a bit dull increase saturation in SPP to +0.3, there is an interestingly larger shift from +0.2 to +0.3 than you would think that seems to bring back in some colors to where you think they should be (like strong reds) and seems to yield more the image you would think of when you use the term “vivid”.
2) The dynamic range is really good, but is based on highlight recovery more than bringing up shadows. If you shoot at ISO 200 the DR is greatest and the detail is nearly identical to ISO 100. Usually in SPP you can bring exposure down as much as two stops with full recovery of anything hidden, and then use other tools to bring back up the rest of the scene – if you export a 16-bitt TIFF file from SPP with -2.0 exposure you can import it into a tool like LightRoom or Aperture, bring the exposure back up anywhere in between 0 to +2.0 to bring the scene back, but also use the highlight recovery tools as if you were working from RAW.
3) I’ve read that Adobe is working on X3F support. They do have X3F support for older Sigma cameras, so it’s not unlikely that ACR will support Merrill X3F files at some point.
4) I’m getting similar battery life to you (100+ shots per battery), I’ve found it helps a lot to turn off the camera when done shooting for the moment. If you just leave it on all the time even if you are not shooting, the battery drains rapidly.
5) I usually start by evaluating the JPG’s for rough sharpness and framing, when I have the images I want to use I batch convert just those files into 16-bit TIFF files from X3F. That saves a lot of storage space and also conversion time.
Thanks Kendall for your comments.
1) I now use STD color mode on the DP2M and do the further adjustments in LR. Color accuracy is much better compared to VIVID. Since I prefer rather saturated images, I’m not sure that NEUTRAL would make any sense for me.
2) I still think that the DR of the DP2M is rather limited, at least compared to modern cameras like the D800 or Nex 7. Both highlight and lowlight recovery are often difficult or impossible (even with using both SPP and LR combined). But I’ve mostly shot at ISO 100. I will try ISO 200, maybe the DR will improve slightly.
3) Could happen, but I wouldn’t bet on this. And the other question is: will the IQ with LR be similar to using SPP. With the old DP1 I used only LR at the end, although SPP had slightly the edge, but LR was good enough. The first version(s) of LR which supported X3F-files of the DP1 were clearly not as good (and I therefore prefered and used SPP at that time).
4) That’s absolutely true. Turning off the camera after shooting as quickly as possible is very important for battery life.
5) I thought about this, but decided to just shoot RAW. If saves space on the memory cards and writing time to the card. After a photo-trip I just do a batch conversion of the all RAWs to 16bit-TIFFs and do the further evaluation with the TIFFs. The final images will somtimes be processed again with SPP.
And sorry for the late reply, I was in Tasmania during the last 4 weeks for landscape shooting (with my DP2M).
Thanks for the response, I understand as I am just back myself after being gone a few weeks.
I have one other thought to add, that you may prefer Neutral color mode with saturation set to +0.3. To me that seems to delver the most accurate colors while still being pretty vibrant. You can save the +0.3 saturation as a preset for batch export from SPP.
Very nice review! I own the DP2M myself and the results are stunning. I use the same workflow, but I delete the tiff’s after work, because the tiff’s are ridiculous huge. And you can always make a new tiff. I have read that it is recommended to correct the color balance in SPP before making the tiff’s.
Sometimes I do the color balance in SPP but only if sigificant adjustments are necessary. For the typical slight WB corrections there is in my opinion no gain in IQ by using SPP for WB instead of LR.