Hands-on with Sigma's DP1

Sigma was showing off their digicam slab of tech, the DP1. To the untrained eye, it doesn't look like much. But Sigma's put the same 14.1-megapixel APS-C sized Foveon CMOS sensor used in its SD14 DSLR into this tiny package. There's a f/4, 28-mm (in 35-mm equivalent terms) lens, but what really impressed us was the build quality on this unit -- not a creak or bit of flex to be had on the camera. It's definitely a high-end piece of photo jewelry.

Gallery: Hands-on with Sigma's DP1

• 
• 
• 
• 

[DennisVR @ Feb 1st 2008 8:13AM]

time to rename engadget.com to camgadget.com ?

[dosguy @ Feb 1st 2008 8:41AM]

No optical viewfinder. I'll pass.

[peayq @ Feb 1st 2008 8:47AM]

it is available as a snap-on option on top.

[roflercopterer @ Feb 1st 2008 9:20AM]

I am a big fan of the old-school, plain look on the high-tech digicams.

[alloneword @ Feb 1st 2008 9:31AM]

@DennisVR

No, it isn't. PMA is on. One of the biggest camera events of the year. Engadget are covering.

While many call for the same thing during MacWorld, or during one of the other focused events, they are doing their jobs and covering the news.

(I tried clicking reply, but the page wasn't working. So Ye-Old @Name will have to do :-)

[Joe_Templeman @ Feb 1st 2008 9:31AM]

They shouldnt have included video on it. Ditracts somewhat from the real aim of it.

I can see there being a real market for it though.

[jiganto @ Feb 1st 2008 10:01AM]

How does this compare to the PowerShot G9?

[DennisVR @ Feb 1st 2008 10:11AM]

@alloneword: ah, that explains :)

[tallgrave @ Feb 1st 2008 10:23AM]

A black camera is good-looking even if there is no memory card.
nnn..How much reflects?

[Guntis @ Feb 1st 2008 10:36AM]

If it only had F2.8 lens... With 4.0 lens it's of no interest to me.

[Guntis @ Feb 1st 2008 10:43AM]

jiganto wrote: "How does this compare to the PowerShot G9?"
First of all, this is Full color Foveon sensor. In reality it's 14 MP ÷ 3 = 4.5 MP.

[Jeff @ Feb 1st 2008 10:46AM]

Yeah the f/4 on the lens is kind of a downer. A prime lens is actually nice in a high-end pocket camera, but f/4 puts it in zoom lens territory as far as speed goes.

If not for that, I'd be really interested in this camera. I wish the Foveon sensor would show up in more pocket cameras. I can't see myself ever buying a Sigma DSLR (because I own a bunch of Canon lenses and accessories) but I'm always on the lookout for a new everyday walk-around camera.

[David @ Feb 1st 2008 11:42AM]

Sensors with co-located photosensitive elements like the Foveon behave fundamentally differently from Bayer-patterned sensors like those used in most cameras today. Simple math, just as dividing/multiplying by 3 doesn't actually create a meaningful equivalence. For example, for a typical real-world image, a Bayer-patterned sensor can capture about 80% of the luminance (gray scale) information as the sensor you'd obtain from replacing each single-color pixel with a full-color pixel. Color resolution suffers, but real-world images do not feature rapidly-changing color. The perception of detail is mostly in the luminance. So, by sacrificing some unnecessary color resolution, you get to drop 2/3 of the photosensitive elements (and thus 2/3 of the data you need to store/process) while retaining 80% of the detail capture. This is why Bayer sensors are so efficient and remain common.

Since consumers associate pixel count units such as megapixels with the image quality they've become used to through Bayer sensors, I think we should avoid using the same terminology to refer to a sensor type with completely different characteristics. It would be more honest to stick with calling the DP1 a camera with 4.5M full-color pixels rather than just multipying by 3 and pretending it's a 14Mpixel camera. It's a different kind of sensor that equates to Bayer sensors differently depending on the metric used.

Here's another way of thinking of it. Take the Foveon sensor and now spread the co-located stack of 3 photosensitive elements out so that they are side by side. Now you have the same number of elements but have traded off color resolution for spatial resolution. Using the same pixel count number for both cases doesn't capture the fact that each has different properties.

David

[Adam Aftab @ Feb 1st 2008 12:13PM]

opinions...

Sigma vs. Leica D-lux 3

[trumpton @ Feb 1st 2008 12:31PM]

The real competition for the Sigma is the Ricoh GRII and CX100 cameras - both have smaller sensors so image quality will lag in low light, but the user interface on the Ricoh cams is the best in the business and both cameras are substantially smaller than the Sigma.

The lack of video on the DP1 is a real pain too.

[bill @ Feb 1st 2008 12:41PM]

Ahh, lets see now, fake 14 mp , NO ZOOM of any sort, SLOW lens, No Optical viewfinder,

Yeah, what else is there to love on this one. Must be a slow day at Kmart camera dept.


Try a real camera....
Once you have used and held the Canon G9 in your hand and graced is METAL body, and felt the controls snap into place and seen the screen....these others are essentially disposable cameras.

[Daniel @ Feb 1st 2008 12:55PM]

Whilst I'm sure the g9 is a fair camera (Couldn't drag me away from my panasonic lx2 thought) I think this is more for the proffesional photographer to use as a holiday camera/take places a full size DSLR couldn't go. Whilst The G9 hast the control,this has the control and a DSLR size sensor which coupled with the fixed lens Should make amazing pictures. This is what should happen... Wether is actually will be the best compact in existence we'll have to wait and see :):)

[morgan @ Feb 1st 2008 12:59PM]

I think this is a great idea, but a 28mm (35 equiv) is too wide and f/4 a tad slow. Also, apparently the ISO only goes up to 800. So this will be great for sunny days, less so for anything else.

[Jeff @ Feb 1st 2008 1:11PM]

28mm is hardly "too wide". Most beginner photographers always want to get in close on everything, they think they need 600mm telephoto lenses on all their cameras. Most experienced photographers, though, know that the opposite is true. You can almost never have too wide a lens for general shooting. You can always crop later and you can always physically move closer to your subject, but you can never add to a picture in which you couldn't fit everything you wanted to. (The whole "physically moving closer to your subject" thing is something beginners take a while to get their head around - you need to build confidence as a photographer in order to do this a lot of the time.)

There are obviously certain situations where you want a long lens; if you want to compress the background for creative reasons, or if you're taking shots of flying airplanes or something. But probably 90% of the shots that all photographers take are shot at the widest focal length their lens will support. And the difference between a 28mm and, say, a 35mm lens is not going to enable you to do any of the things you'd want a telephoto lens for anyway. So you may as well just make the lens as wide as possible.

The nice thing about wide primes is that they have no distortion, which is the main reason some people don't like shooting wide with zoom lenses. So trust me, nobody's going to be taking photos with this camera and saying "man, I wish this lens wasn't so wide."

[dave @ Feb 1st 2008 1:27PM]

Ugh...f4 is pretty damn slow for a shooter of this size. I was hoping for it to be similar to a Ricoh GRD. Without a fast lens, this doesn't appeal to me any longer as a rugged street/low light shooter. Shucks.

[BigDaddyM @ Feb 1st 2008 1:29PM]

Slow lens, and lack luster ISO settings (f4 and ISO 800? Thats all? The chip is clearly able to do more because the same chip is available in their more expensive DSLR. The size is not a limiting factor. IT IS LIMITED BY SIGMA!!!)

Of course we know that which is why I am not buying one.
M

[SimonLok @ Feb 1st 2008 1:41PM]

You people dont seem to understand basic physics. When you put a full sized sensor in a compact camera body you need to make compromises. You obviously dont want a full size lens, nor would one fit, so f4 was probably the best they could really do. Also, you really cant physically have a wide angle lens either due to the size constraints. Since the megapixel race has led to compacts getting increasingly crappier image quality, this seems like a nice little camera when carrying an SLR is not an option.

[m @ Feb 1st 2008 2:04PM]

david has explained it well, but a picture says a thousand words.

http://www.foveon.com/article.php?a=67

this is a new technology that attempts to mimic the superior color of tradional film. give it a chance, haters. and engadget, please keep covering anything foveon.

[zoyd @ Feb 1st 2008 2:39PM]

The problem with discussions about cameras, would seem to be that they about everything except the darn photo quality, you know, the whole purpose of a camera is to take pictures. They are not jewelry, or erotic devices. Tell me about picture noise thresholds, I rather here about that then see a friggin picture of the camera itself!

[simplebot @ Feb 1st 2008 2:40PM]

The Ricoh GR-1 film camera has a 28mm f2.8 lens, and is still very compact, so I don't know if it's a matter of basic physics. I guess maybe with the angle that light would have to hit the sensor with a fast lens doesn't work as well as with film. I think Leica had to deal with this problem with the M8.
I'll be curious to see what the interface is like on the DP1 though. It'll be tough to beat the ease of use of the Ricoh GRD.

[Guntis @ Feb 1st 2008 4:13PM]

If it only had F2.8 lens... With 4.0 lens it's of no interest to me.

[johan @ Feb 1st 2008 4:51PM]

yeah, I'm the target audience for this camera, and would have been all over it were it not for the lack-luster lens. Canon's 35 f/2 is teeny, and sigma have a stellar 30mm f/1.4 that would be awesome on camera like this. And yes, those focal lengths end up ~ 50mm equiv, rather than 28mm, so it's not apples to apples. However, I'd much rather have a slightly less wide but much faster lens than a slow wide-angle.

But at f/4, this is just not interesting. Maybe I'll just buy a second hand rebel xt and put the 30mm f/1.4 on that fulltime.

[Hans @ Feb 1st 2008 5:08PM]

How do you set the manual focus? via a menu?

[Bill Koslosky, MD @ Feb 1st 2008 8:23PM]

28mm, f/4, maxing out at ISO 800?

This is mostly an outdoor camera, since you'd be hard pressed to shoot many interiors, especially without image stabilization (IS).

BTW, the Canon G9 sensor measures 7.6 x 5.7 mm, and the DP1 is 20.7 x 13.8 mm, so that's 6x the area of the G9. I don't think you can get a good image at ISO 800 with the G9, but then again it has IS.

If it were 35 mm, f/2 with close focusing, and an acceptable IQ at ISO 1600, it might be doable for

[Bill Koslosky, MD @ Feb 1st 2008 8:24PM]

...

[Bill Koslosky, MD @ Feb 1st 2008 8:25PM]

... less than $800. (sheesh, you can't use the less than symbol)

[Dan @ Feb 1st 2008 9:50PM]

f/4 on a compact digital camera is faster than it is on an SLR. The sensor is a lot closer.

[dionysus @ Feb 2nd 2008 12:24AM]

DP1 is not your regular compact. It's pro grade. f/4 isn't a problem, a good photographer's hands sure can hold 1/2 seconds at least. 35mm f/4, 1/2 sec, ISO 400, and APS-C are perfect match for a handheld shot in dim light.

[Laurencd @ Feb 2nd 2008 7:01AM]

The calls for a faster lens ignore the physics of light and how digital technology works. Of course we had fast lenses in compact FILM cameras, because film is far more tolerant of light coming in at narrow angles. And we pretty much ignored the vignetting taking place there, because that was the way it was.
Digital is different. Light coming in at tight angles causes severe vignetting. Some of that can be corrected through software; the rest has to be handled through technology. Leica chose offset microlenses with some modest success. These were tried on the DP1 but no real benefit was seen.
The second difference is that the size of the lens opening is directly proportional to the size of the imager behind it. Ricoh can easily do an f/2.8 lens on their GR cameras because the imager is so tiny. I bet that image circle of that lens has plenty of overlap. Here is a link with some comparisons of imager sizes:
http://www.dpreview.com/learn/?/Glossary/Camera_System/sensor_sizes_01.htm
One has to imagine that the front element has to increase proportionally to the size of the imager, by comparison. So either you calculate the comparison to the present f/4 lens on the DP1 or calculate the difference using the imagers in the GR and the DP1 as references. In either case, you will end up with a rather large clump in what was designed to be a pocketable camera.
Other points:
The camera has a fine optional OVF. It also has a very good optional flash (guide number 14). Video is ok for frame rate but not the best in terms of output resolution. It is great for U-Tube. There is image stabilization also. ISO to 400 is very clean and quite good at 800. A greater ISO range is certainly possible, as happened with the SD14.
The A0 images from the DP1 hanging on the walls look really good. Tiny bits of CA on a couple, but Sigma have said they will fix that in the software. The lens is sharp from corner to corner.

[Rich @ Feb 2nd 2008 8:50PM]

I tried shooting a 28mm,@f/4 with ISO 800 (without IS), and found I could handhold good shots in most cocktail lounges, theaters, pubs, museums and on most public transportation, including New York City buses and subways, even at night. Yes, moving subjects will blur some at such slow speeds, but often this proves more interesting than upsetting. I may consider this camera.

[Kendall Helmstetter Gelner @ Feb 3rd 2008 3:25AM]

First of all, lots of people are forgetting this is a 28mm equivalent lens - 16mm in reality. So if you are using a 28mm on most modern DSLR's, you are really shooting something more like a 42mm lens! That means the DP-1 has both a wider field of view, and an ability to handhold in lower light than people seem to be expecting.

The people that note the GR-D and similar ilk have an f/2.8 lens ignore that the sensor size of the DP-1 is six times larger than that of any other P&S camera around (including the GR-D). That is where the physics of the situation come into play, an f/2.8 lens for a sensor that large would be much more massive and put the DP1 out of reach of being pocketable. And if a camera s not pocketable - why not just bring a DSLR? For those that must have a bit more zoom as a possibility, either accept some further cropping or use the age-old technique of the tele adaptor such as those used on video cameras today.

That leads right into the discussion of image quality and the 14MP claim. You see a lot of talk when the Foveon sensor comes up that really layering the sensors does nothing, and thus the resolution is really 4.6MP. But that ignores the countless reviews of older Foveon bearing cameras like the Sigma SD-10 that show a sensor in that camera with even fewer pixels than the DP-1 is pulling at least 6MP of detail (and somewhere beyond), which means the DP-1 with a 14MP Foveon sensor therefore has significantly more detail than a 6MP camera would... so the fundamental conclusion reached by doubters is trivially shown to be wrong by working examples. It may not quite be 14MP in bayer terms, but to claim it's a 5MP camera in bayer terms is even more wrong.

To understand why this is you need to realize that the Bayer sensor really holds a variable amount of detail from any given scene whereas a Foveon camera captures a fixed level of detail regardless of color changes in a scene. An excellent source to understand that better is this comparison of the Sigma SD-14 (roughly the same sensor used in the DP1-1) with the Canon 5-D:

http://www.ddisoftware.com/sd14-5d/

Lastly, for those that don't think nature holds rapidly changing color data - trees against a sky? Hair against colored material? The world is full of sudden change and is of a fractal nature, where change repeats on many levels and different scales in the same scene. It's nice to have a camera that represents the change in colors and detail consistently no matter if the change occurs over many pixels, or just one.

[D-76 @ Feb 3rd 2008 7:34AM]

Sorry Kendall, but the number of pixels is a physical characteristic of the sensor and not something that is estimated based on how much detail people think it captures. There are 4.65 million pixels on the SD-14/DP1 sensor (2640 x 1760) and layering it does not change that. Each pixel may be better (or not) but the number of pixels remains the same, regardless of layers. Claiming it's 14 megapixels is deceptive and even fraudulent.

The SD-14 met with numerous poor reviews, with its slow operation, poor autofocus and images that were not as good as less expensive 10 megapixel cameras. Each time, the Sigma fans said that the reviewer didn't take the time to figure out how to use the camera (why is it so difficult? there aren't that many features in it). Mike Chaney comes right out and states that he never got the hang of the Canon 5D nor could he obtain sharp images from it, yet he goes on to compare the (possibly defective) camera with the SD-14. Why is it not OK for a magazine reviewer (who probably sees more cameras in a week than most people do in a lifetime) to 'not understand the camera' but it's perfectly fine for Mike Chaney to flub with the 5D? Just about anyone who has used the 5D knows that it's an excellent camera and capable of outstanding images.

So why do Foveon images look different? One major reason (there's a few but I'll focus on just one) is a heavy dose of sharpening in the raw processing (0 sharpening isn't really 0 with Sigma). Most Bayer cameras do not sharpen the image anywhere near as much (or at all in most cases, leaving it to the user to decide, and 0 really does mean 0). Invariably, the 'sharp' Sigma images are compared to the 'soft' Bayer images. Apply the same amount of sharpening to both and the difference vanishes, with the Bayer image often looking much better, as the many reviewers found.

And even if the world was full of rapidly changing colour (it isn't), the human eye can't resolve it (refer to any physiology book on human vision). Sigma fans love to use colour resolution charts (Bayer's weak point) to prove that it can resolve more than Bayer, but that isn't anything that humans can see. Hang a colour resolution chart on the wall and see how well you can see it compared to a b/w chart (not very). Ask an eye doctor why eye charts are always b/w, except to test colour blindness. Requiring the camera to record things that can't be seen is just silly.

I'm also puzzled as to why Laurencd (and it's obvious who it really is hiding behind that nick, LM) claims that there is stabilization in the camera when Sigma makes no mention of it whatsoever. And this is from someone who has worked with Sigma during the development of the camera! Very strange. Also, both he and Kendall beta test for Sigma.

And finally, what's with that digital zoom? A camera that is supposedly all about image quality, but it has a digital zoom?? Does Sigma really think that people are going to be fooled with a fixed focal length lens and digital zoom instead of a true optical zoom? CNET is claiming that the camera will sell for $999. Crazy. The SD-14 costs less, even with a lens! That price won't last long. Anyone considering this camera should wait a few months until the price is slashed, just like with the SD-14.

[Mike V @ Feb 3rd 2008 5:00PM]

I really wish they would stop calling this 14 Mega pixel.
It is 4, get over it Sigma.

I will be buying this camera on principle.
If this camera succeeds, then hopefully someone like Canon will bring out a similar (but killer) camera.

[Riley @ Feb 13th 2008 10:37PM]

conventional bayer principle sensor of 14Mp would have 3.48 mill red, 3.48 mill blue and 6.9 mill green photosites (twice the green register). The foveon has exactly the same number of photosites stacked on each pixel position. The only difference is it is organised spatially differently so the images are smaller.

since there is no AA filter stack on foveon it naturally appears sharper, there is no sharpening effect or processed sharpening. this is well known and well documented, one could even remove the AA filter from a conventional bayer array and see that the images are sharper. You would have to endure moire effects in around 5% of images, something that foveon doesnt do.

An APSC compact would have as much if not more difficulty in organising a lens in so compact a design, the slow F4 is a product of the confines of the lens and sensor relationship.

Riley

[tyty @ Mar 5th 2008 3:21PM]

[quote]Brand-new “TRUE” image processing engine
The DP1 incorporates “TRUE” (Three-layer Responsive Ultimate Engine), the world’s first image
processing engine suited for Foveon three silicon embedded layer direct image sensor. The unique
image processing algorithm, which has been developed throughout the development of the SD9,
SD10 and SD14 cameras, is incorporated into “TRUE”. It enables high-speed image processing and
high image quality.[/quote]

Does this mean the sharpness is in fact processed?