« New Canikon! | Main | 'The Red Chair' Print Offer (Three Days Only) »

Wednesday, 24 February 2016

Comments

Hello again quattrophonia and at the end its all about the beauty of photography!

I have a love/ hate relationship with the Foveon sensor and I must say that I do not have much faith that the bugs have been ironed out sufficiently in this iteration either.

I THINK I'd like to get excited, but THINK I'll withhold judgement for a year or so. Just kidding. Well, maybe not.

Mi uno peso

Kinda reminds me of "if a tree falls in the forest and no one is there to hear it, does it make a sound?"

If [presumably] great sensor technology exists but [practically] no one will use it, does it make a ripple in the photography world?

My analogy is a bit kludge-y, but so are Sigma's repeated stabs at the camera market. They should focus on the Art lenses and license the sensor technology for a reasonable fee to tempt other companies to use it.

"as for pixels...well, it's complicated."

Mostly because the terminology got messed up. All cameras have zero pixels. Pixels are picture elements. Cameras aren't pictures.

Cameras have a certain number of photodiodes. The pictures they make have a different (usually) number of pixels. Foveon takes three diodes to make one pixel in a picture. Bayer requires fewer, due to interpolation. It's smart interpolation, but it still involves guesswork.

Perhaps we should use two numbers for each camera, such as X/Y, where X = photodiodes and Y = pixels in the pictures output. Or maybe that would be even more complicated.

bless there hearts, as we say in the south, they can't hep it.
If these guys had a collaboration with one of the other mfgs in the business, it would have been the industry leader. Reminds me of Sony Beta vs. VHS.

I've owned both Merrill and Quattro cameras, both of which resolved more detail than any camera I have owned. But, there was a consistent problem with blowing highlights requiring underexposure in high contrast situations. Not sure if this is an issue native to Foveon sensors, or the horrendous Sigma software the files have to be processed in. Either way, until these issues are sorted out, Sigma is fighting an uphill battle. Why still no Adobe support for file processing? Getting that and sorting the highlight issue would be a tipping point for many, considering the resolving power of these sensors, which are fantastic for landscape work.

I'd be interested but there some major and some minor issues keeping me from getting too enthusiastic about it. The big ones: it uses the Sigma lens mount but there are no adapters to use other mounts with the camera. How many of us have Sigma mount lenses? They do offer Sigma lens mount conversions for a price so maybe they should offer discounted conversions for buyers of the camera.
No in camera stabilization. Some of their Art series lenses lack stabilization so in-camera would be of major benefit.
Minor...fixed LCD, odd placement of the EVF...
Still, I'll reserve judgment.

I wonder what their industrial designers smoke? That camera looks like a Ricoh GXR and the latest Quattro monolith-thing ran off to the woods together. But seriously, I've always been curious about their cameras.

I love the Sigma/Foveon "look" and do keep trying their cameras with each new iteration, but they flat-out don't work for the type of long-exposure photography I like to do. No way, no how, which is a shame. 8^(

Mike,

I have all three DP Merrills, and have just spent about a month in total with three of the dp Quattros (from LensRentals). The experience with the Quattros has convinced me to sell the Merrills. (The listing for the DP3M will appear this evening). I was about to jump into the dp Quattros, but now will wait and see about the sd Quattros. The addition of the EVF is a huge plus. Also the ability to use some of Sigma's highly regarded DSLR lenses. The non-H model appears to have exactly the same sensor as the dp Quattros, but with the addition of on-sensor PDAF, so presumably focusing will be a bit faster.

I second Jim Kofron's comment about the QS menu system being "actually quite nice." In fact, it's really nice, as are the camera controls. The sd models add a few more buttons, so may be even better operationally. Now, if Sigma would just make its software work at reasonable speeds....

Phil

Unfortunately, the new camera has a very odd design; with a lens mount that projects from the main body more than a pancake lens, and a bottom cover that does not even run along the entire length of the camera, it looks like a prop from Star Trek: The Next Generation".

Sigma has a bigger problem, though, And this gets to the essence of the problem for the Foveon-sensor cameras: None of their models were ever able to, in the words of Geoff Moore, "cross the chasm". I see no evidence that they will be able to do that with the advent of these new models, either.

Good for Sigma. I agree that our outsiders' rear-view-mirror perspective suggests that Sigma might have been smart to open the Foveon to other manufacturers years ago. I have, however, read that there are certain design attributes of the Foveon (power issues?) that made it undesirable to other brands. Not being an electronics engineer with first-hand knowledge of the Foveon I could not verify this. (And don't really care.)

My own history with Sigma's Foveon cameras has arced over the years. My first encounter was with the original DP camera (which I reviewed here in 2009?) was a horror. I had never encountered a gadget that provoked me so close to mayhem. But some of the image files were wonderful.

Fast forwarding a few years I bought the DPx Merrill cameras. Better design, or at least more tolerable. Last year I decided to spend dedicated time seeing what these guys could really do for me. After only a couple of weeks using the DP Merrills exclusively (I felt like Alfred Eisenstadt) I ended up with two outstanding sets of images, two of which have become among my all-time favorites. One is superb a street collage and the other is a rather surreal park/construction scene. Of course you cannot possibly discern the distinctive qualities of these images on a web site; you must see them printed large. But you can get a sense of their extraordinary detail from this tiny section of that park scene. You can nearly tell what these guys had for lunch! (By the way, I shot both of those as in-camera JPEGs.)

Is this new Sigma in my future? Probably not, as I'm trimming kit and becoming more selective in my needs. But those of you who are curious about the legendary Foveon sensor it may soon be an excellent time to buy a used DPx Merrill to see for yourself.

Your Sigma DPx Merrill Fun Fact!
The Sigma DPx Merrill battery (a Sigma BP-41) is the same as Ricoh's DB-65 battery, used in their GR and GR II cameras!

Excited I was, until I saw they kept their SA mount and its not so mirror-less flange to sensor distance... It surely makes sense for Sigma, I understand they want to capitalize on their existing SA lenses... but who has SA lenses? It looks to me as if they trap themselves with their existing user base instead of expanding it ?
A more sensible move to me would perhaps have been to go for a shorter mount and possibly proposing adapters for their A-Mount DN lenses and their SA lenses? And well, a plethora of third party adapters would start too... ? or am I missing something?

Greetings,
S.

(addendum to my post)
I mean...
http://camerasize.com/compact/#656.549,669.402,380.305,ha,t
:-/
I really want them to succeed !

Even the images from 1st generation Sigma compacts still amaze me with their richness of colors and subtle nuances of rendering the light.

Here is a gallery of some of the best DP2 images on flickr:

https://www.flickr.com/photos/-fs-/galleries/72157664297050855/

What a lens!

I will be purchasing the H through your links when they are available. I own a couple of SD1M bodies, all three DPM cameras, a Quattro 0, and about eight or so SD mount lenses. Do you think I am a loyal Foveon photographer? :D

PS: In response to the 'MFA in Alaska' post the other day (sorry, always late to the party), and you implying you may have missed the boat becoming a teacher -- you ARE a teacher. Your classroom differs from mine geographically, but you do on an occasion visit my classroom and photography students via the big screen. I think your teaching perch is more relaxing than mine, but I do not have to edit blog comments or go through heaps of email. What your missing on my ground floor is dealing with young adults that say things like: "I was homeless last week," or sadly watch another bounce off the walls while dealing with a tragic personal loss. My mode of discipline is forbidding a student from borrowing something from our gear closet and it hurts me almost as much as it hurts them. But no Mike, you did not miss the boat, your just above my ground floor. BTW, I started out teaching in private schools, but in retirement, I decided it was time to go public and glad I did. A different set of needs, but the same results.

And here I just sold my SD1 Merrill and a clutch of Sigma SA-mount lenses. It was a glorious experiment while it lasted, but after doing a 1:1 comparison between the Pentax K-3's 24MP sensor and the 15MP Merrill, it was a visual wash (using the same settings in Iridient Developer), and the Pentax is just so much faster and easier in every sense.

Sometimes the SD1 felt a bit like owning a classic MGB; exhilarating when the weather's right and the road is open, but more the opposite sitting in traffic or the garage. Thankfully, the SD1 didn't require service with every season!

If only they'd in parallel be working closely with the 2 or 3 most popular image processing software vendors, such that full support would be available at initial launch.

I am an happy user of the three Merrills and one has to see the prints to really realize the quality and unique signature of the Foveon sensor. This is a surprising and very welcomed announcement from Sigma, waiting to see how they will improve upon the existing models and fix some of the issues of the Quattro series sensors.
One thing for sure they need to improve is the Sigma Photo Pro software that is still incomplete and slow even if it is usable with some pain and patience.

I find it interesting that they've added a slightly bigger sensor with the APS-H, but didn't do a m4/3 one. The have m4/3 mount lenses (though designed for APS-C), and I'd have to think that the telecentricity designed into the Olympus & Panasonic lenses would help with some of the issues the Foveon sensor seems to have with incident angles.

But I probably just want a way to try a Foveon sensor without a huge commitment. Maybe I should just pick up an older DP or DP Merrill! Any suggestions as to which model?

I've been shooting off and on with a Sigma DP3 Merrill for the last couple of years, and love the look of the file. As you say, there is an integrity in them that I do not see in the best available Bayer cameras, like the D810 that I also use.

Your Kodachrome analog is apt too: exposure requirements of the Merrill remind me of very much of slide film. They have a pretty limited dynamic range, and it's easy to blow out highlights, but when you get it right, there's really nothing else like it. I think because of the exposure demands of the sensors, there is a certain look to Merrill photos, for better or for worse.

For me, they are the ultimate cameras for taking pictures of clouds, despite what I said about highlights above: spot meter the brightest whites at +0.7 EV is a good rule of thumb. (D810 would be at least +1.7, if not more.)

Note that I use the term Merrill instead of Foveon above, because in the Foveon community, there's some disagreement about whether the Quattro sensor is an actual advancement over the Merrill.

I picked up a Polaroid x530 with Foveon sensor a number of years ago, just to have a Foveon sensor camera to play with. Awful low light performance and a very clunky camera and interface but in good light its handling of colors is simply amazing. Beautiful smooth and rich reds and blues especially.

Sigma has been on my studio shopping list for a while now. I'd heard rumors of a new camera in the offing and have held off to see what unfolds over the camera release season. We are spoilt for choice for sure.

I've played with Sigma RAW files and they are brilliant. As far as I'm aware, the top full resolution layer of the Quattro sensor is used in black and white conversion. It's wonderful!

Excellent print offer. Sorry to hear it's tinged with sad emotions. Thoughts go out to you, having been there.

Best. David.

On a good day, the various flavours of Foveon can be excellent but there are lots of limitations which may be a inherent part of using silicon depth filtering.

The strange colour space which can sometimes mean colours go a bit weird; a tendency to colour casts in the corners; excessive noise even at lowish ISOs; desaturation of shadows; dramatic loss of saturation as ISO is raised, weird big magenta blotches. Not to mention that the lack of an AA filter leads to rather obvious stairstepping aliasing artefacts on diagonal lines.

However, on a good day, you can achieve spectacularly crisp shots with some subtle colour differentiation and lots of micro contrast. And pixel for pixel, the detail beats Bayer sensors (although perhaps not by quite so much as Sigma claim).

Maybe it would be good to ring in Ctein for this discussion, because he knows a lot more about it than I do, but my view of sensors at this point in their development is that photos are products of software, not hardware -- that if any big photo company wished to do it, they could develop a "Foveon" look as one of the "art" selections on their Bayer cameras. IMHO, there's a lot of nonsense circulating about the inherent "fuzziness" of Bayer sensors because of the necessary interpolation of the RGB sensors, but when you realize that the sensor sites in, say, a Nikon D800e are about 4.8 millionths of a meter center-to-center, you've gone way beyond any perceptible fuzziness. (For comparison, the finest grains of sand in standard classifications are about 60 micrometers across. So 12 or 13 of these pixel sites would fit across the finest grains of sand.) And you're not really guessing about interpolation anymore because of the overwhelming numbers of sample sites -- it's not like you've got nine sensor sites and have to decide how much of each color to mix.

So, basically, I think Foveon is a dead end. If given as much initial development money as Bayer, Foveon may or may not have equaled the output of Bayer sensors, but I find it highly unlikely that it would have exceeded it. And if it doesn't exceed it, and if the "look" can be replicated with software, what's the point?

In response to Shaun's comment about blown highlights in high contrast situations, I completely agree with regard to the Merrills. However, i think the Quattro goes a long way towards solving that problem. I'm just finishing up an extensive comparison (but not yet posted on the web). The short story is that the Merrills have a strong bias toward the blue channel in overexposure. In many cases, the red and green channels become significantly underexposed when holding highlights in the blue. This problem seems much ameliorated in the Quattros. I assume because the red and green-sensitive photosites have about 4x the area of the top-layer blue-sensitive sites. Also, the Quattro top-layer (blue) seems to absorb less light than the Merril's - thus allowing more light to penetrate to the lower green and red-sensitive layers. The net effect of this, is that when overexposure occurs with the Quattro, it's more likely to involve both blue and green channels - and the red channel may be not far behind. Still, when using matrix metering with Quattros in high contrast situations, a little negative exposure compensation is a good idea. But that's really a metering issue.

I've been so tempted at times to try a Foveon. And I don't even mind taking the financial risk of a full purchase. What I do mind, and cannot tolerate, is crappy software. Even the most ardent fans of these cameras never have anything positive to say about the software.

I still find it amazing at this point that companies such as Sigma (and Sony to an extent) are able to produce such great hardware but are unable to solve the software/UI stuff. How hard can it be to hire the right people to get it fixed?

A big thumbs up from me on the industrial design of these. Like the Quattro it is foreword-looking and lovely.

Sigma gets a lot of flak for their camera designs, which I think is mostly unjustified. I've got a DP2 Merrill and it's a surprisingly well thought out camera. It seems spartan compared to even some modern compact cameras, but somehow, Sigma made it very easy and comfortable to use, despite limited set of physical controls. And it's got the best shutter button I have ever used. It was almost an agonizing experience going back to my E-M10.

I don't know, maybe I'm weird, I really enjoyed shooting with DP2M. What I was not enjoying, was processing RAW files in Sigma's software. I've got gigabytes of unprocessed images from it and I don't even feel like going through them to select a dozen or two to process. I'm sure I'll get to that one day, maybe when I upgrade my PC to some kind of monster machine.

Anyway, those new sd cameras look stunning. I know they will be a joy to use. And I want one of them. Sigma's announcement triggered the most severe GAS attack I've had for a long time. The only downside I see is the lack of tilting/articulating screen. I guess you can't have everything :-)

Now, I'm secretly hoping my enthusiasm will not deflate once prices are announced.

Is that really a bit missing at the bottom? Won't it tip over when you put it down?

Have to applaud Sigma for trying new things. The styling of this camera looks to have potential for decent ergonomics and the APS-H sensor is interesting.

I would have been ok with the fixed lens Sigmas were it not for the workflow issues and ergonomic issues on the cameras themselves, such as high battery consumption and slowness. The new cameras are probably improved, but I doubt enough to warrant a serious consideration from me and that Sigma mount isn't all that interesting either.

Still, if the wrinkles are worked out then I wouldn't mind to try one.

You have to love Sigma's bold approach to industrial design, that is unless you shoot with your left eye. Ouch.

Files from the Foveon sensor are effing mind blowing. Yeah, Kodachrome is a good analogy. I haven't seen anything else like it. I do hope Sigma has sorted the intensive power requirements these sensors seem to have. In my limited experience with the dp2, it went through batteries like nobody's business. Anyway, regardless, it'll be really interesting to see these cameras.

Big step to the right direction. But I think with two major problems. Why introduce two identical bodies at the same time? For a small company even one new body takes effort to build and distribute. How many dealers want to stock two quirky cameras? Just do one first and then have a mark 2 version a year or two later. But you cannot announce the new, better version now. Otherwise some will wait and you lose sales. Stupid marketing.
And why the long, fixed tube to replace the mirror box so that slr lenses can be used? Shorter mount would have allowed new, smaller system lenses to be developed and many other lenses mounted via adapter. You could have an 'extension tube' to use slr lenses like some Leica visoflex lenses , which is basically a lens adapter. This mount limits lens selection to Sigma's own. No other manufacturer has ever made lenses for Sigma cameras, because the market is not there. This will limit their camera sales and it will limit their body sales. So it will be another failure.
Sad. Because the basic technogy is really very good.

Meant to say: it will limit their camera body sales and it will limit their lens sales.

Sigma's Achilles Heel is the lack of support for the Foveon sensor in the major raw converters. It turned out to be too much of a barrier for me with my DP2. If I am going to go through that much effort I might as well scan film. If they could find a way to work with Adobe, say, then I would seriously consider the H even though the last thing I need is another camera.

I have and use the original Sigma DP 1 and 2. I also have and use the Merrill 1 and 2. They work in completely different ways. The former are lovely street / travel cameras especially used with an OVF and pre-focussed with their nifty little wheels. Lovely prints up to 8 x 11 1/2 inches or so without much up-ressing. The latter are like mini-medium format cameras. I only have capacity to print to 17 x 22 inch and they more than are capable of prints that size. Since suffering double shoulder injuries, I use them for landscape work at lot because I can carry both easily. They do look rather silly on a large tripod though. All have significant limitations but so do all tools. If you want one size fits all buy a small sensor superzoom. I have been thinking about buying an SD1 which is available new in Australia for A$1250. But I have been put off by the SA mount - I don't want to replicate my Nikon glass. I suspect the same will be true for this new version but it will depend on price. Having the same lens mount with both a SLR and a mirrorless (especially the slightly larger H sensor) may be enough persuade me to set up a new system. It's funny how that works. I'm won't buy one body and lenses (too expensive) but two bodies and lenses (more expensive) - no worries, Mate.

I have to say that I am intrigued and appalled. I use the Merrills as tripod-based landscape cameras, but I find the software to be slow, very buggy - computer crashes frequently. The software is THE LIMITING FEATURE for many users. Now I haven't the most recent version that handles Quattro files, but until I get some indication from other users that the software has improved, I am not likely to buy another Sigma. The problem is that the user base for the Sigma cameras is very small and the processing pathway too different from Bayer and Fuji (Bayer-like single-color R, G, B photodiodes, but arranged in a different array pattern). The Foveon sensor R, G, B values are all determined from one photodiode-equivalent site, with the colors actually being the distance the photon travels within the silicon layer. Quattro complicates this further by having the blue read out from "giant" pixels, and red and green from standard pixels. It isn't profitable for Adobe to go after such a tiny market that requires so much work to rewrite the ACR module. So, it has to be either Sigma making their own software, or a small concern making a very focused RAW editor without bells and whistles - RAW to tif - then pass on to some other software to do more peripheral functions. Iridient Developer supports Merrills and earlier, but not Quattros. I don't think that anyone other than Sigma supports Quattros. Correct me if you know better, please.

I noticed this a few days ago: https://sigmaphoto.com.au/cameras?gclid=CIan9M_akcsCFVcRvQodFygKvg

A$1249. That's US$899 at today's exchange rate. Probably take off the 10% GST as well. Compare: http://www.bhphotovideo.com/c/product/846208-REG/Sigma_C26900_SD1_Merrill_Digital_SLR.html Save yourself $1,000. I'd guess that Sigma Australia might be wanting to clear their stocks. It's tempting when you look at the Art lenses and the upcoming 50-100mm f1.8 Art. My favourite zoom.

I wonder why Sigma made the grip shorter than the rest of the camera. How would that work for people with large hands?

As a pro I look et the ergonomics. The viewfinders is 4-5 cm too far to right. A photog. has to keep his head in a terrible angle, or this is meant for the left eyed photogs. And. The screen will be greased by the nose. Saw this and thought, hey - at last something new...then saw the back face. Oh no!!!!!!!!!

I'm waiting for a Foveon camera the size of a Sony RX, with a zoom lens 28-100. I have no doubt the sensor would be better, but after years of lugging around heavy cameras and heavy lenses I am done with that gig.

I've not dealt with images from a Foveon sensor so I can't comment on the IQ differences between it and Bayer or Fuji matrix type sensors, but...

... your analogy to color film tech is a bit off. Comparing the Foveon to layered dye-coupler films like Kodachrome is quite reasonable. Comparing Bayer matrix sensors to the 3 shot cameras or their 1 shot 3 film descendants is poor. These 3 shot and 1 shot cameras are good analogs to the tech used in color TV from its inception until the early '70s.

The Bayer matrix is a digital replication of the many one shot films and plates that used a colored matrix in front of a panachromatic b&w emulsion. These "processes" include Krayn (both line and mosaic), Dufay Pioptichrome, Omnicolore, Thames, Paget, Leto, Baker Duplex, Finlay, Dufaycolor, Autochrome, Afga Color (c.1916), Lignose, and Aurora. Autochrome was the most successful on the market and is the best remembered, but the Bayer matrix most closely mimics the Dufay and Baker Duplex films.

An excellent reference on this topic is Colour Photography, The First Hundred Years, 1840-1940 by the late Brian Coe. It's out of print and somewhat hard to find but worth the effort.

[Thank you dwig. --Mike]

What a nice birthday present for somebody I know.

FYI,
M42, Pentax K, Nikon F mount lenses can fit on Sigma SA mount cameras with adapters, and apparently it's possible to convert the Sigma cameras to mount Canon EF lenses since they are the same electrically, and have the same flange depth.

This guy sells kits to turn various sigma mound cameras onto Nikon or Leica R mount cameras.

http://www.sigmacumlaude.com/products.html

I like that they've gone with a full-frame compatible mount. I wonder if that suggest the direction Sigma will go with the sensor eventually? And, for the record, Sigma will convert many of their lenses to other mounts for a fee.

Perhaps this is a poor comparison but Sigma's dedication to this sensor reminds me of Mazda and their long obsession with the rotary engine. Despite sounding like a good idea and looking good on paper (including some legitimate successes) they've never quite been able to figure out how to make it work as well as it theoretically should.

Mike says: It owns the rights to an inherently superior sensor architecture

Foveon (or any passive absorption multilayer sensor) is never (ever) going to be "better" than Bayer sensor in almost every measurable way. They will always have worse midtone (shot) noise, shadow (read) noise performance and color fidelity than a Bayer sensor (for the same pixel amplifier design). The only parameter they can win on is color moire (but they'll still have luminance moire).

The layers are separate but they don't just detect R, G or B: each detects a mix of the primaries. Some blue light is absorbed in the top, middle and bottom layer. Some green light is absorbed in the top, middle and bottom layer. Some red light is absorbed in the top, middle and bottom layers.

Without dyes to form a definite color cutoff (like Bayer) you are relying on absorption in the silicon to do the filtering which is gradual and probabilistic.

The Sigma claim that the three layers are blue, green and red. They even do this in their patents. But it's just not true.

The probability of a photon being absorbed in the top layer is highest for the blue, next most light for greenish and least likely for the red but all three "color ranges" do get some absorption in that layer. It's a desaturated cyan.

In the middle layer some of the blue has been filtered (but not all) so that middle layer is a desaturated yellowish green (some blue, less red and mostly green).

The bottom layer is orangish red (mostly red, a little green and a very little blue). And IR if they omit the hot filter.

To get to real RGB primaries you have to process these "raw" color signals though a color matrix to remove the crosstalk between the raw layers giving even noisier RGB signal as you are multiplying and subtracting three noisy signals to get to the R, G or B primary colors. Each of the primary colors has a photon shot noise (and at higher ISO, read noise) contributions from each of the layers.

This is the major reason for Foeveon sensors poor high ISO performance. The signal to noise ratio goes down quicker than you'd expect compared to other sensors. It also contributes to poor color fidelity because the separation of color is much poorer than dye based primaries with sharper cut off and a closer match to human color receptors.

"White light" is about 60% greenish, 30% reddish and 10% bluish in terms of photo counts so even though the top layer is most sensitive to blue there is a lot of green (when taking real images) absorbed in that layer. That's the reason they can subdivide the top layer into four subpixels or consolidate the middle and bottom layer into 4x larger pixels (depending on your POV).

The "Quattro" design works by reducing the resolution of the bottom two layers so as to improve their noise performance and so overall noise performance whilst trading off color resolution at the red end of the spectrum. This is an admission that the top layer isn't really "blue" otherwise you wouldn't get most the luminance information at high resolution (which is mostly in the green channel that's why Bayer CFA have two green pixels). The top layer clearly has a lot of green (luminance) data otherwise this reduced resolution in the lower layers wouldn't work.

Now Foveon has all high photo noise issues (from adding/subtracting all three channels from each other) along with a variation of the demosaicing problem (they have to do color interpolation to get RGB for each pixel). This seems to be the worst of both worlds.

The color fidelity of "differential absorption" multilayer sensors, those,like the Foveon, that don't use dyes, is worse than CFA sensors. This is a problem with all sensors, single or multilayer, that don't use dyes to tailor the color response of each layer to precisely match the human eye's color response.

Color fidelity gets worse with low photon count ("high ISO") performance as noise from each layer is mixed into each color. This doesn't happen as much with CFA sensors though noise does mess up color interpolation. This chrominance noise and the increased luminance noise are the core limitation of the Foveon sensor at "high ISOs". I'm impressed that they're doing as well as they are but they've pretty much hit the limit of what they can do with this type of sensor.

They probably do interpolation (a weak form of demosaicing) when they generate RGB from the raw data. That's one of the claimed advantages over the Bayer has been

There also the problem of the pixel count in marketing. This is a 19Mpx resolution sensor in greenish cyan (and perhaps most of the luminance channel) and a 5Mpx sensor in the greenish yellow and organish red layers. Yet they still claim it's a 39Mpx (eq) sensor. Bizzare. (Just add the count of the PDs together. .) At best (blues and greens e.g. landscape photography involving leaves?) it's a 19Mpx sensor but for other colors the resolution will be worse but perhaps no worse than Bayer.

I think the honest answer is in the dimensions of the JPEGs these cameras generate:

Sensor: 29mp Sigma Foveon X3 APS sensor, 23.5 x 15.5 (1.5x crop) 5424 x 3616 JPEG or 14-bit raw

Sensor: 44.8mp Sigma Foveon X3 1.3x sensor, 26.6 x 17.9 (1.3x crop)
Images: 6200 x 4152 JPEG or 14-bit raw

So the APC-C generates 20Mpixels images from a 20MP top layer and 5Mpx underlayers. The APS-H sensor is 24Mpixels from a 24Mpx top layer and 6Mpx underlayers. Not quite what Sigma would like you to think.

A rule of thumb for Bayer sensors is they have a resolution of about 0.7 pixel pitch when the image is demoisaiced. So a 36Mpx Bayer sensor (D810 or Pentax K1) will both equal this in resolution, you'll have good high ISO performance and color fidelity. Plus you'll be able to process the images in your regular workflow.

The trade off for no color moire (but not no luminance moire) is worse color performance and worse high ISO performance. For a landscape (or other daylight photographers) this might be an good trade off but in general the Bayer CFA trade offs are not as bad: like democracy it's the "least worst" solution.

Being "open" doesn't change these disadvantages. I'm pretty sure that all the major camera companies evaluated the Foveon sensor and decided not to use it for all of these reasons.

Eeek, that was longer than I expected!

Some comments about the camera rather than the sensor.

This is the second "mirrorless DSLR" (i.e. a mirrorless camera that uses a DSLR lens mount) after the Pentax K-01. You can expect to see both Canon and Nikon move to "mirrorless DSLR" designs in their consumer cameras (at least) at some point in the future as the saving of mechanical parts, assembly time and cost is significant.

These cameras have phase detection autofocus (PDAF) but it's only 9-point for PDAF along with CDAF (as you'd expect with sensor based PDAF). The CDAF (I think) is an improvement over the previous SD.

They look a bit like a boxy small medium format camera. I suspect this is the market that Sigma have decided to target: daylight or controlled studio lighting.

The usual outstanding problems are can Sigma write software and design good UIs for their cameras. They haven't managed this in the past with cameras being very slow in operation.

I'm surprised I haven't seen the Sigma cameras palced in SciFi movies. They do look rather retrofuturistic from a future that never quite arrived. Perhaps that sums them up.

Unintended consequences.

This announcement tipped me over the edge into ordering… a dp0 Quattro with their LVF-01 loupe accessory. I looked at Sigma's native lens offerings and decided (rightly or wrongly) that I wasn't going to be able to do better with the new cameras than I would with the 21mm-e on the dp0Q, and there's an instant rebate on the current dpx Quattro line, so…

For me it will replace my 500C/M and 4x5s, and also the SWC and XPan I never had. Plop it on a tripod, grab a cup of coffee and relax - the kind of deliberate shooting I enjoy.

My photographic hair shirt arrives next week.

While I've never used a Sigma camera, I've always been intrigued by the reports of amazing image quality. The problems with Sigma Photo Pro have been discussed at length here (when was the last time any of us used raw processing software from any manufacturer other than Sigma (because nothing else works)?

There's another problem, though... These cameras use the worst possible lens mount! The SA-mount is so rare that even B+H lists the vast majority of lenses as special order only... Given that B+H stocks full medium format lens lines for systems that sell in the low thousands of bodies annually, AND ultratelephoto and tilt/shift lenses that sell hundreds of copies (or less) worldwide on an annual basis, my best estimate of the number of SA-mount cameras out there so far is a few thousand at most.

The counterargument to this is "if the cameras sell, the lenses will be more popular". The problem with this is that the SA-mount is not a simple set of lenses to stock (when Fuji first introduced the X-mount, there were three lenses, although the fourth arrived quickly). Now there are 20+, but there a million or so bodies in the world to use them on, so a big dealer can move all of them, and a smaller dealer will pick a nice selection.

The SA-mount features a LOT of overlapping lenses (17-50 and 17-70, 8-16,10-20 and 12-24, and two separate 150-600 lenses plus a 50-500). Sigma theoretically makes everything they make for Canon EF in a SA version (both the flange distance and the electronic interface are identical, so it's a simple matter of bolting on a different bayonet plate). I suspect there are some of the exotica (300-800 f5.6, 200-500 f2.8) that may have NEVER SOLD A COPY in the SA-mount version. How is a dealer to choose from 30+ lenses that work only with bodies that have never sold at all? Most dealers will probably ONLY stock the 17-70 Contemporary as a "kit lens" for the new bodies, and say "anything else is special order, although there's quite a bit to choose from". This is sure to make the cameras less appealing!

Good for Sigma, who are an endearing company and who deserve to thrive. I have the DP2s and DP2M. The former is reminiscent of the Rollei 35 as a compact travel camera: its IQ is wonderful. The Merrill is, as others have said, medium format in a compact. The lens on the DP2M is the best I've used, beating even Leica and Mamiya IMHO. The Quattro sensor seems to me to be a step backward and the samples I have seen look more 'Bayer-like'. The essence of Foveon quality is not quite there anymore, so I'd need some convincing to get either a DPQ model or one of the new SDQs. The ideal to me would be the new SD body with the Merrill sensor; or better, the DPM body with an EVF. I'd recommend anyone to have a try with one of the original DP compacts, now available 2/hand very cheaply. Or indeed try a Merrill, the steal of the century even at full price. Don't get too worried about SPP. It's not that bad, and you can use it just to make a TIFF for later processing in PS etc.

The comments to this entry are closed.

Portals




Stats


Blog powered by Typepad
Member since 06/2007