Sigma DP2 finally takes some pictures, doesn't disappoint
By Donald Melanson 
posted Mar 7th 2009 4:13AM
posted Mar 7th 2009 4:13AM
Sigma has been doling out bits and pieces of information about its new DP2 camera since last fall, but it's just now letting folks use it to actually take some pictures and, judging from LetsGoDigital's results, it doesn't look to disappoint. Of course, the camera is still in pre-production form, but LetsGoDigital certainly seems to be plenty impressed with its abilities so far, with the "Super Multi-Layer Coating" on the lens helping to reduce lens flare, while the 41mm f/2.8 lens and 14-megapixel FOVEON X3 image sensor combine to deliver some great depth of field. Hit up the link below for some full-size sample images, plus another hands-on look at the camera itself.
好棒好棒
How do you mean?
Good bat, good bat.
;p
還早吧， 等有實際的測試我才會高興。
Nice camera. Sigam always does nice things at low costs.
BTW sorry for the offtopic, but in Portugal, Vodafone already stoped selling iPhones, saying that they will not receive more. Reason? New iphone on the way maybe....
One thing is for sure, iPhones 16GB are more cheaper (400%u20AC) and 8GB are not sold anymore, only the ones left in stock.
Well it is really 41mm equivelent, and not really 14 megapixels either.... I guess even Engadget falls for "marketing" material... every single time.
It isn't 14 mp, true.
But look at the pictures...
ZOMG!!!
Awesome pics. Looking forward to a review of the actual retail product.
Every other camera so-called 14mp camera is not a 14mp camera either, these guys (foveon x3 makers)get a lot of shit for building a true 5mp camera, and everyone else gets away with posing their 5mp cameras as 15mp camera (with each so-called "pixel" is "crippled" to only one color; yet they still chose to call it a "pixel", even though the definition of pixel is the smallest image unit, a single color is obviously not a pixel)
Even though this is not a 15mp sensor, nobody else's is either (except for the super expensive "50mp" sensors). Foveon is not the one that developed this scam, it's everyone else, and now Foveon has to present their product on this scam-ridden market.
This sensor produces TRUE 5MP image... all other makers of these so called 15mp sensors produce an extrapolated 15MP image from the same number of sensors that this sensor has.
Please do your homework, and look at the per-pixel quality of the images produced by this camera, and then look at the per pixel quality produced by canon 50d. You'll see a fundamentally different approach (lotsa fuzzy pixels vs fewer sharp pixels).
You've been brainwashed my friend, you've bought into the scam, and now you're judging based on your brainwashed belief of what a mega-schmixel REALLY IS.
It's just as much a 14mp camera as any other camera is. *You* have fallen for *everyone else's* marketing. Just because other manufacturers choose to measure megapixels in one way doesn't make that a standard and doesn't mean everyone else has to do it the same way. It's like saying you produce a "500 watt amplifier" - ok, measured how? Without more information, everybody who knows anything about audio knows that's meaningless. Megapixels is no different. It can be measured in different ways.
The way Sigma measures it is no more or less valid than anybody else. The sensor in this camera does capture 14 million (or so) distinct points of digital information. The fact that they're split over three layers and not spread across one layer is pretty immaterial to how many "megapixels" they can claim. And what really matters is the results.
The color rendition, dynamic range and noise levels from this camera frankly blow the doors of any other camera in this class. It's not even close.
@Pretol
What you say is *partially* correct. There are actuality two different "resolutions." Chroma and Luminosity. It is true that each pixel can only capture a third of the chroma (color) detail. But at the same time they all contribute more or less to the luminosity. Especially the green pixels, which comprise roughly 50% on your standard Bayer filter (with the remaining 50% split equally between red and blue).
Now human eyes work in a very similar fashion; We too are more sensitive to luminosity than color. Which is why a 14 MP Bayer can produce more perceivable resolution than a 14MP Foveon.
It is however worth noting that a Bayer Sensor's resolution drops significantly when the subject has a strong color bias.
Also, a sensor can still only capture as much detail as the lens in front of it will allow.
I'll say it again. It's not 14 MP. And how does a f/2.8 on a small chip create "great depth of field"? Ever seen f/2.8 on a full-frame chip?
I think what engadget mean is that for a point and shoot style camera this has fairly good depth of field since its sensor is bigger than most other compacts. Of course its not going to compete with a full frame slr.
*yawn* Yet another know-it-all who likes to show off what he knows. Yes, we all know that these aren't 14,000 real pixels, you're not the only one around here with a brain. But being the photography know-it-all that you are, you must surely realize that more pixels does not mean better pixels. In a world of megapixel inflation, Sigma has no choice but to market the camera on "perceived" pixels - basically, to claim that it is as good resolution as a 14MP. Sheesh, you just can't please some people. The same people who whine that camera companies are cramming too many pixels on a sensor for the sake of marketing, whine when a camera company tries to make a sensor with better quality pixels, not more.
As for "great depth of field", every comment has to be taken in context. The context here is that this is essentially a compact camera. Have you ever seen (or rather "not seen") the depth of field available in compact cameras? Then you should know that this will BLOW THEM AWAY. It will be best of class, that's what I call "great depth of field".
However, neither is it equivalent to a 4.6mp bayer-pattern sensor.
"In a world of megapixel inflation, Sigma has no choice but to market the camera on "perceived" pixels - basically, to claim that it is as good resolution as a 14MP."
Sigma's pixels aren't perceived, they're imagined. The Sigma sensor is 4.6MP. There is no way that anyone can ever realize 14MP's worth of resolution with a Sigma camera and testing shows that.
"Sheesh, you just can't please some people. The same people who whine that camera companies are cramming too many pixels on a sensor for the sake of marketing, whine when a camera company tries to make a sensor with better quality pixels, not more."
In other words, you believe it's better for a company to lie about their product. As for better quality pixels, it's debatable whether Sigma provides that. Testing sure doesn't demonstrate it. Foveon has been trying for years and the best they've done is trade one set of problems for another.
"As for "great depth of field", every comment has to be taken in context. The context here is that this is essentially a compact camera. Have you ever seen (or rather "not seen") the depth of field available in compact cameras? Then you should know that this will BLOW THEM AWAY. It will be best of class, that's what I call "great depth of field"."
An apologist in every crowd. How many photographers who really care about this want a 41mm f/5 perspective in 35mm terms? DOF isn't that great even if it is better than a typical PnS, the lens isn't wide angle nor is it long enough for good portraits and it has no zoom. I'd say micro 4/3 with the 50mm f/2 lens is preferable though not as small (yet) and Samsung is demonstrating something similar with an even larger sensor. Sure, the DP2 is small but it's a dud. The DP2 is "best of class" only in a class of one.
I think the real problem here, judging not only from the attitude but the avatar, is that we simply have a Nikon fanboi. That Nikon apologists are the most obnoxious and vitriolic of the fanboi lot doesn't help any.
If Nikon decided to use the Foveon sensor in anything, he would simply hail it as the second coming and dismiss all other brands as obsolete. I the meantime, he has to slag on a simple point and shoot by comparing it to a D3? Get a clue. Please.
The chip in the DP2 is by no means small, it's almost APS-C sized.
@ bannedmonkey
Sorry bud, I use a Sony a900 with old Minolta glass. Just switched from Nikon in December. How could I possibly be a Nikon fanboy?
It's not a 15mp sensor, in the same way that NOBODY ELSE'S "15MP" SENSOR is actually a 15mp sensor.
Definition of pixel - is the smallest image unit. And in today's camera that means that 1/3 of these "pixels" are green, 1/3 are red, and 1/3 is blue. Cameras extrapolate those colors onto these neighboring differently colored pixels, and then present it to you as a 15mp picture. Arguably this is better than having a fewer, but fully colored pixels (life Foveon does), but in no way it gives them a permission to use a "14mp sensor" moniker (because there's only 5 mp of information for each color). I guess if you're color blind you can call it a 15mp sensor. It's been a scam all along, and now that these guys produce a true 5mp imager (equivalent in actual colored pixel count information to a so-called 15mp sensor), everybody flips out how it is not a 15mp.
Well, I'm here to tell you that it is. It has 5 million green pixels, it has 5 million red pixels, and it has 5 million blue pixels, the same as any other 15mp camera. The only difference is that these are truely aligned to be in the same location, and there's no extrapolation that's going on.
And if you look at those pictures, you'll see per-pixel quality that you won't be able to get with any canon 50d, nikon d90 or any other bayer-pattern camera (even though a 50d and d90 outresolve this camera).
Problem is that everybody else's 15mp sensors are NOT TRUE 15MP. And now these guys have to compete in the field of sensors that rely on color extrapolation to get the advertised resolution... and these guys make an equivalent sensor that doesn't do any color extrapolation, and yet it gets the bashing for misadvertisement.
craig:
both you and David (and everybody else who's brought this up) need to educate yourselves about camera and lens resolution. Resolution is not measured in megapixels. It's measured in lines. (Per something, but usually per inch, occasionally per horizontal or vertical frame.) It is a measure of the amount of detail a camera can capture. You can have a 20 megapixel camera with a lens and sensor that are so bad that they turn every scene into a big black blob. Or you can have a 2 megapixel camera with a lens and sensor so good that they render every scene in fine detail. The latter camera has higher resolution.
It is a huge mistake to look at two cameras and say "this camera has more megapixels, therefore it has higher resolution." That's frankly a nonsensical statement - as anyone who knows anything about cameras will tell you, one simply has nothing much to do with the other. It's like looking at two amplifiers and saying "this amplifier is more powerful, therefore it has wider dynamic range." There's just no logic there.
There are a huge number of factors that affect a camera's resolution, with the number of pixels spread across the sensor being way down the list.
@pretol Actually all Bayer Filter cameras are 25% blue, 25% red and 50% green. This is because much of the luminance information comes across in the green channel.
So, ya know, everyone is wrong now :)
That Sigma actually markets the lens coating as a selling point is pretty dumb. Asahi/Pentax marketed their lenses as "SMC" (super multi-coating) in 1972, with everybody else shortly after (Zeiss = T*, Fuji = EMC, etc.). This 70s shtick is fun marketing, though. How about, "Games: now with video graphics!", or, "Apple computers: not just a pile of parts".
Maybe put out some low light shots? Most cheap point-and-shoot cameras do fine in good light.
this isnt just a point and shoot, its a 1.6 crop FOVEON X3 sensor. So while its smaller than full frame its the same size as what you find in the smaller canon dslr.
also while its not a 14mp sensor, because it dosnt use a bayer pattern and interpretation to create the image it can actually resolve more detail than a full frame canon 5d m1 with much better dynamic range. In high contrast area's (like some landscapes) you would have to take muti exposure on a normal dslr to grab the same amount of information int he shadow/highlight areas that this can capture in one exposure.
Sure its not perfect.. it sucks at high iso for example but ppl should really read up more about what they are talking about before thay just diss it because its not a canikon.
I shoot pentax so im not a sigma fanboy either.
"this isnt just a point and shoot, its a 1.6 crop FOVEON X3 sensor. So while its smaller than full frame its the same size as what you find in the smaller canon dslr."
No, it's a 1.74x crop. A Canon has 16% more sensor area than this. A Nikon DX about 30% more.
"also while its not a 14mp sensor, because it dosnt use a bayer pattern and interpretation to create the image it can actually resolve more detail than a full frame canon 5d m1 with much better dynamic range. In high contrast area's (like some landscapes) you would have to take muti exposure on a normal dslr to grab the same amount of information int he shadow/highlight areas that this can capture in one exposure."
All those claims are preposterous. Check out DPReview's test of the Sigma DP1: http://www.dpreview.com/reviews/sigmadp1/page10.asp
This camera is 4.6 MP while the Canon 5D is nearly 13MP. Even adjusting for the Bayer sensor there is no way the Sigma can outresolve the Canon---it can't even outresolve a Nikon D60. As for dynamic range, a Canon 5D is near the top of the class even today. Meanwhile, Foveon sensors are not known for great dynamic range.
"Sure its not perfect.. it sucks at high iso for example but ppl should really read up more about what they are talking about before thay just diss it because its not a canikon. "
I agree, and one of those ppl should be you.
"I shoot pentax so im not a sigma fanboy either."
Saying it doesn't make it so.
Well, craig, it clearly outresolved the d60... the images produced by dp1 are smaller but have superior detail. d60 is blotches of meaningless pixels in comparison... compare the image of the blue chick on the two sensors.
You've gotta be kidding me. 40 f/2.8, and those are their test shots? No wide aperture portraits? C'mon, we need some shots of those Tokyo ramen stalls.
It's *equivalent* to 41mm. Which arguably still makes it a little wide for portraits.
wow a "P&S" that has the quality of a DSLR...cool...if only the batteries weren't so hard to find in Canda
Are you daft? the DP2 has a DSLR size sensor...no way a crappy sony with normal size sensor could beat that...even with zeiss optics
Actually those shots are not impressive, except in the sense that the don't actually look real.
I downloaded the first shot of the semi from further back. The colors look a bit weird and there seems to be jaggies all over. The overall effect is more like a video game image than a real world image.
Next I looked at the closeup of the tire. WTF. The black tire is blown out to pure white. All over the place there nothing but blown out white where the camera lost it, and more jaggies.
Nothing screams artificial like jaggies and big pure white patches. This is looks like an beta product.
I downloaded the large files and didn't see any jaggies at all. The amount and nature of the details are impressive, but that's no surprise as the the DP1 has similarly impressive performance. The white patches indicate that the photographer overexposed the photo in RAW and couldn't quite rescue those parts. If he'd done the same thing with any other compact camera, the whole frame would have been pure white, as the Sigma has a larger dynamic range, at least in RAW.
I know this camera has great IQ; what I want to know is how fast it is to record the images, how fast the shot-to-shot times are and what the 1600ISO shots look like.
I don't see the any "jaggies" either. And please tell me which video games you are playing, as I would most certainly like to see some that look like this. The color fidelity is outstanding and the only thing that is "unreal" about these shots is that they hardly look like digital — which is also a factor that impressed me very much when browsing DP1 pictures: even if there are lots of high-ISO noise, that noise doesn't give the pictures a "noisy digipic" look. It has an organic, film like quality to it.
You have your opinion, I have mine. Still using an old blurry CRT? In the first photo on my LCD, I see obvious jaggies in the shadows on the hood. I see them drivers side mirror bracket and the windows frame on that door as well. That combined with whatever is strange with the color, make the picture look artificial.
As far as the over-exposing the black tire to white. The camera did that, it was choosing the exposure.
This camera certainly be better than small sensor cameras, but these pictures don't impress me.
Upsized for the vision impaired. Upsizing didn't add these, it just makes them obvious to those on blurry monitor or with blurry eyesite.
http://i41.tinypic.com/2iw60px.jpg
Ah, now I see what you mean. Clumsy sharpening applied when making the JPEG?
Welcome to the world of computers. Yes, pixels are square. So when per-pixel quality is high you see these "jaggies". What all other camera makers have done is - up the resolution and smudge the pixels for a smooth effect.
Here's a solution for you (the non-technical-imager kind):
Take a pair of glasses, smudge them in fine sand for about 5 minutes.
Now put them on, and look at these "jaggies".
You'll now notice how smooth the image is, and the pixels will blend into infinity of megapixels.
You're looking at what every other manufacturer has done with their sensors (and it only cost you: glasses & 5 minutes of sandblasting).
tons of chromatic aberration in the 2nd big test photo.
Did engadget actually look over the photos carefully? The results are very disappointing.
For a second generation product this is great progress. I'm really impressed by Sigma's initiative to use a bigger sensor inside a relatively compact body. The reason why I love the 5d mk1 so much (will never get rid of mine) is because it has amazing (low) pixel density.
What I am really waiting for is a shootout between this and the upcoming Olympus camera using the Micro 4/3rds system chip.
I've never had a DP behave like this.
OK, it's semantics, but it's not "great depth of field", but "shallow depth of field". Every compact camera on the planet has "great depth of field", because that's all they can offer. What sets the DP2 apart is its ability to give us less depth of field, not more of it!
Secondly, while the details are sharp - these pictures are awful - they look like heavy-handed tone-mapping (HDR) has been applied to them. While I mostly hate this particular set of pictures, I'm still optimistic that shots taken with the real, production camera in the hands of real users won't look as bad as this. If the DP2 performs better than the painfully sluggish DP1, then I'll probably buy one. Unless of course it turns out that it really does produce fake-looking HDR pictures like these.
"Foveon sensors are not known for great dynamic range..." " All those claims are preposterous. Check out DPReview's test of the Sigma DP1: http://www.dpreview.com/reviews/sigmadp1/page10.asp"
I followed that link, this is from the review you linked to;
"These figures are actually quite impressive. In the Dynamic Range area the DP1 is unmatched by any other compact cameras and even leaves some entry level DSLRs behind it"
Specifically in the test, the D40 was what got left behind, and it matched the D60 in range but the Sigma had a full stop better highlight detail.
Next time link to an article that SUPPORTS your point of view...
"As for dynamic range, a Canon 5D is near the top of the class even today."
Again with this, however now its a Canon fanboi comparing a simple point and shoot to a pro level DSLR. WTF, over?
I would like to see them work on getting a micro 4/3rds lens on that camera. It would solve a lot of lens problems that people have.
Other than that, I like what they have done with the sensor, but haven't done enough research to see how much detail is lost/gained compared to a DSLR & 4/3rds sensor.
I assume that you meant APS-C and 4/3" sensors?
The problem is that APS-C doesn't refer to just one sensor size. There are at least three different "standards". The 1.5x crop used by Nikon, Sony, and Pentax. The 1.6 crop used by Canon. And then the 1.7 crop used by Sigma.
A 1.7 crop sensor isn't that much bigger than 4/3".
Thank you LetsGoDigital...
I would like to see some more difficult shots in terms of high ISO too. Show me something from a pub or so at ISO 800+. Show me some moving objects. Show me MORE SHOTS!
http://www.ddisoftware.com/sd14-5d/
as you can see in some situations the foveon x3 sensor can resolve more detail than the full frame d5 mark1.
ok craig so i was wrong about the crop factor but i was pretty close, some ppl just think that this has a small sensor and that isnt quite true as it has a bigger sensor than the 4/3 format and is so much larger than any point and shoot.
while the sigma only has 4.6mp, it gathers around 3x more info per pixle and also dosnt require an AA filter (which reduces details to).
Any guess on price on this thing? Less, more, same as DP1?
The problem with you arguers is that the number of pixels in a file is not indicative of how much detail the camera can resolve, so the megapixel discussion doesn't even matter. You can have a 1000MP file that's blurry and worse looking than a 1MP file. Stop arguing about all this nonsense and just compare the output to other cameras.
Plus, the DP1 had plenty of problems other than the sensor. The camera has low-light AF problems, a weak flash, slow lens and slow operation (check out DPReview). If the DP2 is going to get it right, they'll have to fix all that too.
At f2.8, the DP2's lens is twice as bright as the DP1's. And the new processor is supposed to speed it up performance wise just a bit.
I wouldn't expect miracles from a dinky built in flash. That's why they included a hotshoe.