A couple of months ago I finally succumbed to the temptation of buying a Sigma sd Quattro H. Ever since the camera was announced, some years ago now, it intrigued me.  As a sometimes delighted, sometimes frustrated owner of the Merrill and Quattro dp fixed-lens series, this new interchangeable lens Foveon sensor mirrorless camera seemed like something I could put to good use.  Of course, it being a Sigma, things are not as simple as they could be. The camera is indeed mirrorless, and fairly compact, but it is designed to take Sigma’s DSLR lenses. This is not totally bad news, as the recent generations of Sigma lenses have been gaining a strong reputation for Zeiss-like levels of optical performance and build quality at a quarter of the price. Unfortunately, they are not a quarter of the weight, or the bulk, and a quarter of Zeiss prices is still a lot of money.

But anyway, here I am, with a sd Quattro H body (let’s shorten that to sd-H from now on), a 35mm f1.4 Art lens, and a 24-70 f2.8 Art lens. The latter is really huge.  And now I need to be convinced all this was a good idea.

The Sigma sd Quattro H with 24-70 f/2.8 lens next to the Olympus E-M1 with equivalent 12-40 f/2.8

I have used the Quattro dp0 quite extensively, mainly as a “panoramic” camera with the 21:9 ratio. That, together with previous Merrill dp2 & dp3 experience meant that I was not blind to the potential issues. In suitable conditions these cameras can be jaw-droppingly effective, but the range of conditions that can be reliable considered suitable is narrow, to put it mildly.

Although the usual claim by enthusiasts of these Foveon sensor cameras is of remarkable resolution (which they have, but let’s not go overboard), for me the killer feature is (and again, in the right conditions), a film-like delicacy of colour and colour transition. This can justify me packing the dp0 Quattro as special-use secondary camera, but the question is, are the results clear enough to justify the sd-H and 24-70 lens, four times heavier and bulkier?

Before following up on that, let’s just have a quick recap of what the sd-H offers. There’s a full, in-depth review at DPReview, so I’m not going to spend much time on technical stuff here. The body is very well built, and feels like it cost more than it did. It is comfortable to hold, despite its unconventional shape.  The controls are well laid out and easy to reach, although I would prefer the QS Quick Menu button to be in a similar position to that on the dp body. The menu is a paragon of good design - it’s a pity so few people will see it. The back of the camera has a typical Sigma quirk, with two screens side by side. The second, smaller one is used to display shooting parameters. And unlike the dp series, there is an electronic viewfinder, which is quite large and comfortable, but suffers from the difficulty of getting a high rate video stream from the Foveon sensor. Still, it is serviceable. Basically from an ergonomic point of view things are pretty good.  Oh, and there is an option to produce linear DNG output instead of X3F Raw files, which means you can open them directly in Lightroom, etc. Although I’m not sure I’d recommend that. Oh, and the autofocus can only be described as “****** hopeless”.

I have used the sd-H properly now on 4 outings. One to Venice, which didn’t go well, two quick trips to the local Valle Verzasca, and just recently a long weekend in Tuscany. It’s still all a bit inconclusive. I got some nice results in Verzasca, but I was very constrained by the lack of Depth of Field preview. Also the lack of an orientable screen or finder can be very restricting. Basically it’s not a lot fun using the sd-H on a tripod, but generally that is where it works best.

this kind of detailed, softly lit scene is where the Sigma cameras do excel

For my trip to Tuscany I went well prepared. The area is one I know extremely well (I even published a book about it) and there’s absolutely no stress to get the shot, because either I’ve already got it, or I’ll get it next time. I also made a DNG colour profile for the camera, and took along my MacBook Air to be able to do some on the spot verification.

So, I did some handheld shots, and some tripod shots, initially all in DNG, and imported them into Lightroom. I was pretty disappointed. For example, the clichéd shots of Tuscan poppies were just smudges, with reds either overblown to flat areas with no detail or clipped to white. Just like digital cameras 20 years ago.  A shot where the ISO crept up to 800 looked like some Chernobyl aftermath. Some shots were inexplicably soft (the 24-70 lens is stabilised, but it’s no Olympus), which I’d also noticed in Venice. And generally the resolution and sharpness was not impressive at all.

Foveon colour at ISO 800. Ouch.

Oh, and the classic Foveon green flare made a unwelcome appearence.

the classic Foveon green flare

But then… when I got home, I opened the same images on my desktop computer, which has a fully calibrated Eizo monitor attached to it, and there a rather different picture emerged (literally!). The overblown reds turned out to actually hold detail. The softness in some cases turned out not to be so soft. Some of the poppy field shots turned out ok. And the photos which I took in X3F format are technically not so catastrophic. So the lesson there is that perhaps my 2011 MacBook Air is not the best tool for evaluating image quality.

Foveon colour at ISO 200. A bit better.

Hardly a portfolio shot, but technically this one worked ok.

Since there was quite a lot of mountain bike touring involved on this trip, I didn’t limit myself to the Sigma. I also took my Olympus E-M5ii, with my old and travel worn 14-42EZ pancake zoom. This, unlike the sd-H, could happily travel in my backpack. Oh, the shots show a somewhat alarming softness on the right bottom corner, but if you don’t look too closely, the combination actually works pretty well.  Of course, putting a “proper” lens on the Olympus narrows the gap quite a lot, making me question the sd-H even more.

When quickly reviewing photos to illustrate this post, I noticed some shots taken in previous years in Tuscany using the dp0. That camera has a smaller sensor than the sd-H (APS-C rather than APS-H), but a fixed precisely matched (and ultra wide) lens. And even as thumbnails, the shots just pop off the screen. I expected the sd-H, with Art lenses, to have the same effect, but so far, it doesn’t. I’m not quite ready to put it on eBay, but as it stands at the moment it could not justify its place in my camera bag on a real trip.  I guess we still need to work on our relationship.

This is beginning to get there.

 

The recent very welcome and generous firmware update (2.0.1) for Sigma’s DP Quattro cameras has added several new dimensions to working with the output. The update brings two major new features:

• DNG Output
• SFD (Super Fine Definition mode)

Of the two, DNG is probably the most significant: it means that it is no longer necessary to use Sigma Photo Pro to process raw images.  They can now be opened directly in various DNG compatible applications. I’ve tried Adobe Lightroom, Photoshop, Photo Mechanic, Alien Skin Exposure, Iridient Developer. They all work fine. This is a huge boost to workflow, and opens up Sigma Quattro cameras to all those put off by having to use Sigma Photo Pro (which, by the way, is actually nowhere as awful as the internet echo chamber would have you believe).

On top of this, you can now use software such as XRite ColorChecker to create customer camera profiles to use in compatible applications.

SFD is a bit more esoteric. SFD mode combines 7 standard frames automatically shot at different exposures to obtain a single file with extended shadow and highlight detail, and lower noise. Basically an auto-HDR, but thankfully without the kitsch. Unlike, say, Olympus in-camera HDR, here all individual frames are saved as RAW and can be selected, discarded, or individually edited in Sigma Photo Pro, if you wish to go too that level of detail (no DNG here).  I’d read some pretty negative opinions of SFD from Sigma SD Quattro owners, so I was curious but not all that excited.

So, yesterday evening I went out to grab a few shows to try all this stuff out. Please note this was just a “kick the tyres” sort of thing, in a context of taking photos that I might actually be interested in, not “testing”, taking a million shots of the pot plant, brick wall, or cat nearest to the couch.

DNG output

The addition of DNG output makes a huge potential difference to workflow enhancement. The question is, what, if anything, are we losing? There are a lot of drawbacks to using Sigma Foveon cameras, but there are also two big plusses, resolution, and colour character. Compromising on either of those arguably makes using the cameras pointless. So, using a frame which is not far off a torture test, with deep shadow and bright highlights, not to mention abundant, potentially troublesome green tones, let’s dive in.

I’ve taken two identical shots, one recorded to X3F file, and the other to DNG. I have processed the X3F file in Sigma Photo Pro (SPP) using “Auto” settings - I don’t usually use Auto, but here it seemed to make sense, as I assume something similar must happen internally in-camera to create RGB data for the DNG file. I’ve opened the DNG file in Lightroom, on default settings, but changed the camera profile to Standard to match SPP. Here are the full frames, compared in Lightroom Library module.

X3F on the left, DNG on the right

Ok, you’re not going to be able to tell much from these, but the idea is to get an overall feel. The colour in the DNG file seems a little muted compared to the X3F. We’ll look at that a bit later.

The top left corner of the shot is a bit troublesome - again, we’ll look at this when discussing the SFD mode, but let’s see how X3F and DNG compare.

Top left crop, DNG version

Top left crop, X3F version

Well, there are some minor differences, but nothing I’d personally lose sleep over. What gets interesting is looking at the potential for colour calibration.  Using the XRite Colorchecker Passport, a made a quick Lightroom profile for the SD0.

Left, custom profile, right, Sigma standard

The difference there is quite clear. I’m not yet entirely sure I prefer the custom profile version, but I think it is probably more accurate. Personally I actually like the somewhat muted saturation that the Sigma standard profile gives, but well, it’s certainly opening up a whole new dimension. It will be interesting to compare a Sigma custom profiled shot of an identical scene shot with another camera.

Lightroom actually has all the built-in SPP colour modes as profiles under calibration. I assume these come in with the DNG. If Adobe had actually put some effort into actively supporting Sigma Quattro cameras, I think we’d have heard about it.

As I said above, I’m not all that unenthusiastic about SPP, but even so, there are a lot of good arguments for using DNG, and so far, I don’t see any significant drawbacks. The out of camera DNGs are huge - about 110Mb to the 60-ish Mb of a corresponding X3F, but apparently running them through Adobe DNG converter shrinks them with no adverse side effects. I haven’t tried it yet.

SFD Mode

A couple of shots here demonstrate the usefulness of SFD mode. Unfortunately I was in a bit of a hurry, and the location I was shooting from was a bit precarious, so I wasn’t very careful about details and the non-SFD version is shot a f/4.5, so the background is not sharp.  Note that this sort of shot, where there is moving water, is not supposed to work in SFD mode.  I found it to work fairly well. The significant thing here is the highlights, in the patches of bright water. The SFD mode has yielded a lot more highlight detail.

Full scene, SFD mode

Now, two close ups of the two modes.

Close up, standard mode (auto exposure)

Close up, SFD mode

Note, the white balance is different in SFD mode. It was actually much cooler, but I made a rough adjustment in Photoshop. The differences in sharpness are irrelevant, as mentioned above, they are down to to wide an aperture on the non-SFD shot. However, what is significant is the extra detail in the bright water patches in the SFD shot. In the non-SFD shot they are completely blown.  Also, note, from this example, the SFD rendering of blurred water is comparable to the non-SFD. However, at the edge of the full frame, there was some minor leaf movement, and there artefacts do show up. Sometimes they will be acceptable, sometimes not. Indeed, I took a different SFD shot with moving water, and in that case some of the patterns were a bit weird.  But it looks like it could be useful, in some situations.

Note, I didn’t encounter any of the processing time nightmares broadcast on the forums. On my computer, Sigma Photo Pro 6.5.3 took under 2 minutes to process an SFD set (.XFI file). That seems reasonable to me.

A second example was aimed more at shadow detail. The full frame is below (in this case I was more interested in shadow detail).

Looking in detail at the debris in shadow in the lower left:

Close up, standard mode (auto exposure)

Close up, SFD mode

I would say that is pretty conclusive.

However, the highlights (sky patches, top left) are blown in both cases. The sky was completely overcast, so that is fine, however the loss of detail in the foliage isn’t so impressive. Here the SFD version does a better job, but neither is brilliant. An extreme case, maybe, but also just one of the Foveon sensor’s weak points.

Close up, top left, standard mode (auto exposure)

Close up, top left, SFD mode

I’m not sure how much I might end up using SFD mode. I would imagine it is more generally useful for static subjects in controlled, indoor conditions, but from this very brief trial, it does seem more useful for outdoors photography than I expected. Anyway, I’m certainly not going to complain.

Thanks very much to Sigma for this update - they don’t really get a lot of publicity, but I doubt there is a company today more dedicated to making great photography equipment at reasonable prices. And they just keep improving.