WEDDING SEASON SALE! 30% Off Training Systems!

Your content will be up shortly. Please allow up to 5 seconds

Sigma SD Quattro | Highest Resolution & Most Unique Mirrorless System yet?

By Anthony Thurston on February 23rd 2016

In addition to the lenses that we mentioned in our other post today, Sigma also announced a couple of new mirrorless cameras this morning. The SD Quattro and SD Quattro H are a pair of uniquely designed Foveon-based mirrorless cameras that make use of Sigmas proprietary SA lens mount.


Sigma’s New SD Quattro Mirrorless Cameras

The Sigma SD Quattro line of cameras is made up of two models: a 36MP*(*equivalent, according to Sigma) APS-C Foveon-based camera and a 56MP*APS-H  Foveon-based option.

Honestly, we really have not received much information on these cameras yet other than what I have noted. We can see the unique design, and according to what little we have been able to find, these are built with dust and moisture-resistant construction (weather-sealed).

Bildschirmfoto-2016-02-23-um-08.20.26-700x329 Bildschirmfoto-2016-02-23-um-08.20.30-700x362 Bildschirmfoto-2016-02-23-um-08.20.34-700x378


I must be honest; I am really curious about the design of these. Part of me really likes it, and part of me is sort of wondering who on Earth approved this. I mean, take the viewfinder placement for example, on the right side of the camera…if you shoot with your right eye, then this will be great for you, but for left eye shooters like myself, this is a very odd location.

Then there is the mount choice. People keep getting on Canon and Nikon wanting F/EF mount mirrorless cameras. Well, this is likely what we would get if they did that…a weird ‘tube’ protruding from the camera. It looks odd here for the SA mount, and it would look odd on a Nikon/Canon body.

No pricing or availability has been announced for these yet, but I would guess that they will end up somewhere near the higher end of the other Quattro cameras.

What are your thoughts on these new mirrorless cameras from Sigma? Leave a comment below and let us know what you think!

[Via Sigma Website]

This site contains affiliate links to products. We may receive a commission for purchases made through these links, however, this does not impact accuracy or integrity of our content.

Anthony Thurston is a photographer based in the Salem, Oregon area specializing in Boudoir. He recently started a new project, Fiercely Boudoir to help support the growing boudoir community. Find him over on Instagram. You may also connect with him via Email.

Q&A Discussions

Please or register to post a comment.

  1. Andrew Leinonen

    I totally understand their reasoning to use the proprietary SA mount so that they can lock in their buyers with their proprietary lenses…

    …but it does feel like it wastes the huge potential of mirrorless. If they had made it an E-Mount or M4/3 mount – or even a shorter, proprietary “SM” mount – they could sell all the Sigma 19/30/60mm lenses already designed for short-flange mirrorless, or adapt their larger DSLR glass with an autofocus adapter as per their just-released E-mount to SA adapter.

    Of course, most of us would probably just use dumb adapters and stop down our very sharpest old glass to get a taste of that amazing Foveon magic. So obviously that’s why Sigma has done what they’ve done. But I don’t like the decision, nonetheless…

    | |
    • adam sanford

      This could just be a tech project to play around with an interchangeable lens setup and dial-in how it all works before they embark on an EF / FX mount product of their own.

      | |
    • Dave Haynie

      I think, given Sigma’s market, they were wise to not change the lens mount, at least if they had to go to a proprietary mount. Given the APS-H-ish sensor, they definitely wanted to deliver a full frame lens or so, which would have precluded Micro Four-thirds, probably the only popular lens mount they could just go out and license. And had they changed lens mount, they would have alienated the Sigma Faithful… they’re out there :-)

      | |
  2. Jean-Francois Perreault

    I agree with the weird viewfinder location. If you need to use the d-pad (selecting focus point?) then your thumb might tickle your nose a little if you’re a left eye shooter.

    As for the mount, yeah that’s pretty much what a native EF-EF-S mount would look like. It’s not that bad if one would only put EF lenses on it. An adapter would pretty much give the same result anyway. But if that’s any indication of what it would look like, I think I’d rather use an adapter and get smaller native lenses when I want something lighter.

    Unless performance is out of this world, I don’t think it’ll sell very well. Well, at least I wouldn’t buy it.

    | |
  3. adam sanford

    1) This isn’t *that* new — Merrill + Quattro are their Foveon ‘platforms of the future’, replete with a sensor you can’t use above pedestrian ISO levels and has the ergonomic sensibilities of a flaming turd. (These used to have fixed lenses and they now have an SA mount. Sure. Lovely. Kudos.)
    2) They do not have nearly the hard pixel count they are reporting — I am not Stan Rogers or Dave Haynie, but I believe this is a Bayer v. Foveon thing in which they get *more detail per pixel* than we might with traditional sensors. To some extent, you can believe the hype (depending on what you shoot and if you take it to print).
    3) I’ve polled the Canon faithful and about two thirds of them want a future FF mirrorless offering to have the full EF mount and chunky grip so that they can natively use all EF lenses on day one. In other words (though the SA mount and APS-H are clearly different), in very high level / broad strokes, Canon folks want *something* like what is pictured here, but with Canon ergonomics, menus, etc.

    | |
    • Andrew Leinonen

      They don’t have the hard pixel count they’re suggesting, but right now the APS-C sized Quattros are strongly competitive with the A7r II and 5DS in terms of base ISO image quality and resolution. So they’re doing something right.

      You need to go medium format before you see a step up in quality from the Quattros. Assuming the narrow shooting envelope (low ISO, relatively low DR, fixed lens choices) suits the kind of work you do.

      | |
    • adam sanford

      Agree. The one thing you cannot knock with these rigs is the base ISO IQ. It’s *really* good.
      But for all it’s limitations, it feels like MF before they moved to CMOS. Low framerate, horrific noise even at moderate ISO levels, etc. So if you can live with those sort of tradeoffs, mazel tov, but otherwise…. Ouch.

      | |
    • Dave Haynie

      Sigma’s a little crazy with their claims of sensor size. Basically, their original sensors delivered full RGB pixels. So they had something in the 5Mpixel range, but they started counting every sensor site as a pixel, so they claimed 15Mpixel. Which was a complete lie… it’s a 5Mpixel sensor without any Bayer interpolation. And the big problem they had was the way the sensor works. They’ve put different photodiodes at different layers in the silicon, and based on how silicon absorbs light, they do some not-super-simple math to deduce color. The problem is, the silicon is blocking all light, just some frequenies more than others. So they got an interesting result, but terrible low-light performance. And the low-light dropoff was weird, as the lower layers started showing noise long before the top layer.

      To solve this, all the recent Foveon sensors have a full resolution pixel layer on top, which of course takes in all color of light, then 1/4 resolution (pixels doubled in both directions) in the lower layers. So what’s essentially luma is at full resolution, but the additional information used to deliver color is the same on two layers for pixel buckets of 4.

      In this specific case, the APS-H version (which is apparently about halfway between a Sony APS-C and a Canon APS-H in size) is about 25.7Mpixels, and the sublayers are each 6.425 Mpixels. I have absolutely no idea how they come up with 51Mpixels.

      The effective resolution would be better than a pure Bayer sensor camera at 26Mpixel, but it’s arguable by just how much. When you De-Bayer a raw image, you’re interpolating the missing colors based on surrounding samples. The over all effect is a little bit of softening — I’ve heard it’s more like 82% of a perfect RGB sensor these days. Mostly, it’s distortion, like purple lines and moire patterns, when the interpolation can’t deal with edge conditions. So there’s no interpolation here, but a different class of color errors, given the lower resolution layers. If they did an absolutely perfect RGB simulation, given the other number (which was for a Leica M9, but I’m not about to claim Leica’s color interpolation algorithms are significantly better than anyone else’s), that would mean their 26MPixel camera is a good match to a 31Mpixel Bayer camera, and should avoid moires better. Only problem.. the previous version of this type of Foveon sensor was maybe good to ISO 800, ISO1600 if you’re a real Sigma fan. And you’re giving up fast shooting, video, etc. not just low light.

      | |
    • Andrew Leinonen

      Dave, the closest analogue I have for Foveon sensor performance (at base ISO only) would be the full-RGB sensor-shift pixels produced by the Pentax K3 II and the Olympus E-M5 II.

      I am strongly skeptical of the 82% Bayer effective interpolation number.

      Looking at Imaging Resource’s testing of these pixel-shift modes, it’s clear that despite large sensor size deficits and higher pixel densities (putting higher acuity demands on the lenses), the 24MP K3 II in 4-shot mode produces demonstrably better image quality than a 36MP AA-less Bayer sensor in the D810.

      So the value of full-RGB pixels provides at least a 50% premium in terms of effective real-world image resolution vs. demosaiced images.

      As you mention, the Quattro sensor is a bit of a special case because it does require some interpolation due to the non-uniform colour layers. But I’d be comfortable putting their current ~20MP sensor on fighting terms with that same D810, so as fuzzy or theoretical as their math is, it’s worth something in the real world.

      | |