Sigma SD14 reviewed: promising, but not quite there

We've been hearing about Sigma's SD14 14.1-megapixel DSLR with that nifty Foveon 3-layer image sensor since last year, but it looks like all the hype was just that: Popular Photography got their hands on a review unit and wasn't too impressed. Although the cam earned high marks for color accuracy and image quality, the reviewers didn't see the full potential of the novel image sensor -- JPEG shots were rated on par with the 8-megapixel Canon Digital Rebel XT, while RAW images actually had less detail in them than shots taken by the 10-megapixel Nikon D80. The reviewers also knocked the slow image processor, which they found produced blocky JPEG images at higher ISOs and took 8-10 seconds to clear its buffer after shooting just six pictures in burst mode. Although the Foveon sensor is promising, the review concludes that you're probably better off spending your $1,600 elsewhere -- not exactly a ringing endorsement of the "fundamentally better technology" Sigma and Foveon promised.

Read - Sigma SD14 review
Read - Foveon X3 sensor review

[chris joseph @ Jun 14th 2007 9:53PM]

These could even be software issues, for all anyone knows.

[Foof @ Jun 14th 2007 11:14PM]

It could be, but at at this point, that's so speculative that if the issues concern you, you probably shouldn't take the plunge on the off-chance that they'll be remedied by some software/firmware update.

[Za @ Jun 14th 2007 10:14PM]

If by software, you mean firmware. And frankly, I doubt the firmware could significantly change how the RAWs look, although without question it would make a difference with jpegs.

[Shaocaholica @ Jun 14th 2007 11:29PM]

These are physical limitations of the camera/sensor. Too often tech writers spec this as a traditional 14mega pixel camera but its not a traditional sensor and its really a 14 mega 'sensor' camera with 3 sensors per pixel. Hence it really only has a resolution of 4 mega pixels (14/3) which is why it doesn't have the detail of a traditional 10MP camera. Its not a bad camera but it gets misleading impressions because Sigma marketing and ignorant writers keep throwing the 14MP spec around when its not really 14MP.

[t-bone @ Jun 15th 2007 12:54AM]

I hope you aren't calling McNamara ignorant. He knows exactly what he is dealing with here. I sense you may be one of those Foveon fanatics that are always making excuses for why your sensor doesn't ever live up to the hype. I found the reviews to be very fair and objective.

[Shaocaholica @ Jun 15th 2007 1:54AM]

I'm not talking about the pop photo review but the Engadget summary and replies to it.

[Ben @ Jun 15th 2007 4:13AM]

T-bone, Shaocaholica is exactly right. This is a 4.7 megapixel camera. Period. That's how many pixels it senses. (a) the camera is VASTLY overrated at 14.1 MP, (b) the reviewer should have explicitly called them on it, (c) comparing it to an 8 or 10 MP camera shows exactly what you'd expect, that the far greater number of sensors will eventually, with proper Bayer processing, overwhelm the RGB capability to resolve luma of the 4.7 MP camera. The obvious underlying truths here are (a) They want to sell this camera for a LOT of money (b) in order to do that, they're trying (laughably) to pretend that pixel-accurate RGB detail is equivalent to the number of sensor sites for luma in a bayer sensor (it isn't) (c) if you compare this camera honestly to something like an Olympus E20 (5 MP) then you're going to see exactly what you expect - the RGB sensor will eat the Bayer pattern sensor for lunch. (d) You're actually SEEING this in the review, with the camera being compared favorably with an 8 MP camera (which is probably losing 1/2 its resolution due to low pass filters designed to reduce moire patterns. When a 5 MP camera loses 1/2 its res, the RGB sensor will kick its fanny off the planet.)

Me, I'm all hot for the tech, you can even call me fanboy and I won't twitch, but at $1600 for the camera, and 5 MP for the images, EVEN though they are arguably the best 5 MP you can buy right now, I'm not buying. It's just that simple. If they want to sell me a 5 MP camera, I'll give em about $300. Maybe $350. Period. If it is a FABulous camera. Which the SD14 is, but which $1600 isn't (fabulous.) If you get my drift. Which is drifting.

Anyway, seeing as how I can have (with a bayer sensor) 10 MP (luma) resolution and about 1/2 (5 MP) color resolution, and buy THAT for about $700 or $800, my message to sigma/foveon is "Back to the drawing board, boys and girls."

And by the way, I write image processing code for a living, including RAW conversion plugins, and I am intimately familiar with the bayer pattern, its limitations, and why RGB is better - given the same number of sensor sites. Which is not the case in the comparisons made in the review, and why the review, no matter your respect for the reviewer, isn't a very good review.

If *I* had done the review, I would have started off by explaining that the 14.1 megapixel claim is based on multplying sensor sites times color channels, for the SD14, wheras the 10 MP claim for the Canon (for example) is the *actual* number of sensor sites, so that for luma, (B&W detail) the Canon has better root technology, their sensor site density being considerably higher. Then I would have compared the SD14 to a 5 (or so) MP camera, it would have done brilliantly, and one could laud this appropriately, then point out the price disparity, and pan the camera based on price ONLY (well, and on hype - it isn't a 14.1 MP camera, period, end of story... but that's not the engineer's fault, that's the fault of the marketing dweebs.)

[VJH @ Jun 15th 2007 3:37AM]

I'm not too surprised by the PopPhoto review. The Bayer mosaic sensors are an example of a good technology that has held up well over the years and been able to scale in terms of pixel density and noise reduction. Superior or not, any other sensor technology is going to have a difficult time in a head-to-head competition with that of today's Bayer sensors. Of course, I also happen to think that the Foveon will appeal to a certain type of photographer; and that it may someday even make a good general purpose camera sensor. One thing I think the PopPhotog reviewers could have done differently was to use Sigma lenses for both cameras used when testing the sensor, since that would remove many of the peculiarities of the respective lenses on the results.

What's obvious though is that regardless of the benefits the Foveon sensor brings to DP, they are hampered by Sigma's R&D budget...and Sigma glass.

[Evan @ Jun 15th 2007 8:47AM]

Let's not forget Nikon (and I'll admit Canon) have vastly superior lenses compared to Sigma. Any way you slice it - or, pixelize it - it ain't worth no $1600.

[Russ @ Jun 15th 2007 11:01AM]

ben, you're exactly right.. this summary and even the comparison is retarded. comparing a 4.7 mp camera to an 8 and 10 is just funny.. see, the conclusion they came to actually gives props to the sensor, saying that it compared to an 8mp. and it's only 5.. what does that say for image quality. and they even mention image quality was great. and of course it has less data in a raw than a 10mb raw.. IT'S ONLY 5 MP... people are retarded.. especially engadget.. i'm not even necessarily a fan of foveon, but i'm not stupid enough to not know how it works. this entire review is bogus..

btw, i still shoot canon, but as soon as I can buy a 12.8MP (38.4) foveon sensor camera, I will.. it's logically the next step.. no more bayer.

[t-bone @ Jun 15th 2007 2:08PM]

If the SD14 is only 4.7MP, why does SIGMA call it a 14MP camera? What are reviewers supposed to compare it to? Are they supposed to dig up a Digital Rebel or a 1D MkI? How abouta How about a D1x? What would that say about it? I think if they did that the review would be even worse: "New Sigma Stacks Up Well Against Out-Dated Cameras".

[Russ @ Jun 19th 2007 9:45PM]

you're an idiot.. they call it a 14mp camera because that's how many photosites it has.. it is a 14mp camera.. it's just different.. they overlap.. so technically the images are going to have better individual pixels... but not be anywhere near the size of your common 14mp camera. just read,.. learn... try not to be ignorant. and in my opinion they don't have anything to compare it to. but the fact that a 5mp picture stood up to a 8 and 10 is a little impressive..

[Britton @ Jun 22nd 2007 12:04AM]

This "4.7" MP sensor's images are of good enough quality that when updcaled they compete favorabley with the high pixel-count Bayer mosaic sensors. Camera's using the Bayer sensors are just doing the interpolation in camera to reach the high pixel counts.
Here is a good article describing the resolution differences:
http://www.ddisoftware.com/sd14-5d/