Sigma sd Quattro H real world samples gallery

Well that 35mm has horrendous CA towards the edges. Is Adobe unable to do corrections on foveon files?

Apr 10, 2017permalink

Mescalamba

Not really, just Foveon sensor makes it really visible. On standard Bayer you wont notice it.

Dxomark has same lens reviewed and they are actually really good. Just Foveon is absolutely punishing for any regular lens.

Apr 10, 2017permalink

@Mescalamba, that's the worst excuse I've ever heard. Do you realize that CA may be 10 pixels wide?

Apr 10, 2017permalink

WastingTime

This might be due to the layered design of the Foveon Sensor. I own that lens and does not show that amount of CA not even on a Canon 5DS R.
However I would assume that Lightroom should be able to handle it.

Apr 10, 2017permalink

Mescalamba

You dont get the essential thing.

Bayer sensors have real resolution much lower than mpix count. Foven has resolution exactly as stated. When they say something about comparable Bayer resolution, they not kidding.

So reason why you see that CA is cause Foveon having truly insanely high resolution plus there is no CA correction, which in most manufacturers case is either done in camera (half-baked into RAW) or in some SW later.

If you want perfect lens, you need to look elsewhere than Sigma. They make very good lens for very good money. I dont think they ever stated they make best lens. :D

Also many lens from today main manufacturers are kinda crap without SW.

Apr 11, 2017*permalink

CaPi

What lenses would you recommend for this camera body then?

Apr 11, 2017permalink

My point was more that Adobe should be able to handle it correctly. Foveon makes CA worse not because of its superior resolution but because of crosstalk on the lower layers. Light hitting the edge of the sensor at an angle could be as much as 4 pixels off by the time it reaches Foveons very deep red layer. Then any correction has to contend with the lower resolution in the lower layers. If Foveon was using quantum film which is much thinner this problem would be less. Still think there must be something wrong with this lens though.

Apr 12, 2017permalink

fedway

A few of the images have a magical quality about them (for example processing Image 2). Some I just can't make to work. More than most cameras, I think this is one example of a tool that the user really has to know its limitations and its strengths, and only use accordingly.

Apr 10, 2017*permalink

kotepum

I've noticed that as well. Most of the landspace samples look like they've been taken with a cell phone camera.

Apr 11, 2017permalink

Flashback

=MF

Apr 10, 2017permalink

(unknown member)

The color depth Foveons produce is unparallel. Sigma has proved it time and again. No camera beats it. Multilayered sensors to the masses NOW!

Apr 10, 2017permalink

Michael Berg

I'm not sure what you mean by color depth. The camera writes 12 bits per color to DNG and 14 bits per color to X3F, which is about on par with most other high end cameras.

Foveon sensors provide per-fotosite color information resulting in better sharpness and definition in the "a" and "b" channel (in the L,a,b space). But the sensor design in the Quattro is a bit of a compromise. While it is a semi true-color sensor with multiple layers, the Quattro sensor layers are not all full resolution as they are in the Sigma DP Merrill sensor:

- Blue layer: full spatial resolution, broad panchromatic sensitivity
- Red layer: 1/2 spatial resolution = 1/4 the data of the blue layer
- Green layer: 1/2 spatial resolution = 1/4 the data of the blue layer

Which implies 14 bits + 7 bits = 21 bits per image pixel.

The camera does produce lovely images but doesn't outclass cameras like the Fuji XT-2 by any significant margin as far as color processing goes.

Apr 10, 2017permalink

It more about colour space than colour depth which makes the colours of a foveon different (not more correct but then its not like bayer is perfect either). Different can be a good and bad thing. Myself I've found it extraordinary for landscapes but a bit off for people.

Apr 10, 2017permalink

(unknown member)

Yeah, perhaps color space is more appropriate to say. You feel like the image is more real, the colors more penetrating. I don't know, it's something very different that reminds me of film. Everything looks more real. It beats any other camera either for portrairts or anything else. I worked as a minilab operator for a few years and was trained by Kodak and Noritsu folks, and, well, in my opinion, nothing beats it.

Apr 10, 2017*permalink

Sir Nick of High Point

I must point out that bit depth is only a function of the A/D converter and has nothing to do with the actual amount of information captured. For example, many A/D converters in home audio gear are advertised as 16 bit, but due to converter noise the true depth is closer to 10 bits. Then there is the question of sensor output. Even if you have a true 14 bit converter, the sensor may only be capable of 8 bits, and the file is just bloated with redundant data. The true end-to-end performance of the system must be considered, not just a single value from a spec sheet.

Apr 10, 2017permalink

AbrasiveReducer

I like the color in the Bloody Mary. But yeah, it's still a low ISO camera. I can't help thinking the potential is there, if someone with deep pockets wanted to do the R&D.

Apr 10, 2017permalink

Daft Punk

To my eyes some of the sample photos have a lovely quality to them.

I am impressed.

Apr 9, 2017permalink

Alexdi

Resolution is top-drawer. There's something to be said for that 50MP claim.

I'm not as impressed with the color as I once was with the original versions of Foveon, though it's hard to pin down why. There's a learning curve to these raw files. Dynamic range is a throwback; just getting the shadows into a usable state requires pretty extreme corrections. There's false coloring if you really push them.

I can see a niche for landscape and studio photographers. Personally, though, I've been spoiled by Fuji and I'm willing to sacrifice per-pixel sharpness for ease of post-processing. Way back when, I was astonished by the detail of the 11MP 1Ds. It's hard to say existing 24MP bodies aren't good enough on that score.

Apr 9, 2017permalink

HowaboutRAW

You're correct about the weird cyan and magenta blotches, colouring, in the shadows of low ISO DNGs.

It would be nice to see a couple of dozen X3F files shot with this body too.

Apr 9, 2017permalink

HowaboutRAW

Glad that DPR posted out of camera DNGs from the Quattro H, but could we also get some X3F sample files?

Apr 9, 2017permalink

J. Michael

I Agree! Some Raw (X3F) Files would be "very" nice, to compare to the DNG files. I don't own the Quattro H, but it's on my radar, and would like to see how the files processed in SPP would look.

Apr 10, 2017permalink

J. Michael

I spoke too soon, the last images shown have both the DNG and Sigma Raw files
available for download........mea culpa.

DPR, Thanks for posting the images.

Apr 10, 2017permalink

barrym1966

Not a bad little holiday Camera

Apr 9, 2017permalink

Lan

Sadly the elephant's still standing in the room. Whilst the base ISO samples look lovely on the whole, anything over and above that really isn't very good at all.

That ISO1600 sample looks like something a compact camera might have produced 15 years ago, and I'm not talking about a good compact camera either...

If you're always shooting at base ISO, this could be a good choice. If you're shooting anything above that, I'd buy something else; anything else!

Apr 9, 2017permalink

HowaboutRAW

No, compared a compact from 15 years ago, this ISO 1600 (DNG) is excellent, now there are banding problems with this file.

I'd like to see some real world X3F files too. PhotographyBlog posted some, but only a studio ISO 1600 example from the SD Quattro H.

Nope, I've not looked at the jpeg coming out of the ISO 1600, and I won't.

Apr 9, 2017permalink

It's a pixel-peeper paradise. On the other hand, if you are interested in photography, this camera is a liability, not an asset. It's only advantage, high per pixel resolution, vanishes as soon as you resize the image to the normal viewing size.

Apr 9, 2017permalink

dr.noise

You could implement that reasoning to practically anything in the world. Like "This Ferrari's only advantage, the speed, vanishes as soon as you put it in a traffic jam". What is normal viewing size, btw?

Apr 9, 2017permalink

Go look at the 100mp phase one on the comparison page. Select you favourite camera with a reasonable pixel count. Select print comparison to resize the images to 8mp. Notice how much clearer the 100mp is even downsized (look at the foliage or sponges or brushes). That's the beauty of oversampling and why it is done for every form of recording (audio, imaging, communication).

Apr 10, 2017permalink

@Zdman, here is a better exercise for you. Download the raw files, process them in the latest version of LR, export at the viewing size, and let a third party to evaluate. He has a 50% chance of distinguishing 100MP from 24MP original file. In fact, if you use a bit higher sharpening resizing lower resolution file, he will likely decide that file had the higher resolution. Perceptual acuity has a lot more to do with sharpening and contrast than original resolution.

Apr 10, 2017*permalink

@dr.noise, "This Ferrari's only advantage, the speed, vanishes as soon as you put it in a traffic jam"

Though I'm tired of irrelevant car comparison, I'll play along this time.

If this is the only Ferrai's advantage (your words, not mine), then it would be just a one trick pony, absolutely useless anywhere around me, where you can't drive above 80mph.

Isn't it quite obvious, if whatever theoretical advantage you think you may have, but you cannot realize it in reality, it's useless?

Apr 10, 2017permalink

dr.noise

@Contra Mundum, Unlike Ferrari, nobody forces you to downsample your photos, so you really can realize the resolution advantage. This advantage is very real and not theoretical.

Apr 10, 2017permalink

MyReality

@Contra Mundum - Careful, you are using too much logic on this forum. Some posters here due not understand logic or objectivity.
If it was practical, I could produce a test that would bring down the chance of choosing the Foveon print or any other sensor type print accidently to 10% or less. Use the same size print and have a group of people match the print to the cameras. I would guess that the accuracy rate would be very small.

Apr 10, 2017permalink

dr.noise

@MyReality
> Use the same size print and have a group of people match the print to the cameras.

That was done before with A0 size prints and... you will be surprised.

Apr 15, 2017permalink

@Contra. Sorry late back here. Sure you can sharpen any image to anyone's taste but for pure unaliased sharpness with no halo's you're going to need to oversample, its physics. So to get back to your original point resizing does not make all the advantages vanish, you can debate the usefulness of those advantages at length as you have but you can't overlook that there will be some.

Apr 19, 2017permalink

jackspra

Great image quality.

Apr 9, 2017permalink

zodiacfml

Spending time analyzing the DNG files in Lightroom, I discovered that the SD-H has the image quality I expect from a previous experience with the 4.7MP Foveon generation which has little noise and great color. The Merrill gen, though sharp, do not have the IQ for great color. The Quattro, while fixes the Merrill issue somehow, is filled with grain/sand on all images.
This Quattro-H minimizes that sand to something I don't mind much anymore. I hope they release the DP series using this APS-H sensor soon.

Apr 9, 2017permalink

SETI

Very good IQ, especially from Sigma converter. That «awfull grain» is not a problem at all =) I want to own that camera!

Apr 9, 2017permalink

If you think so than try to retouch it in photoshop with clone stamp or healing brash OR PATCH TOOL(the worth!)... Just be aware... Thats all im saying.
I wanted it too, but after i saw first x3f - download them and process it in all kinds of ways - sandy gain is present allover the place...( Very sad.
Anyway im not convincing people NOT TO i just sharing my tears...)))) i real believed Sigma would made it Good...

Apr 9, 2017permalink

SETI

I see... and no way to remove it in converter?

Apr 9, 2017permalink

Well, i spend few hours with different samples that i dig thru inet and made quite big collection of x3f from sdq and sdqH made with different lenses and after SPP post sandy look gone and took all the details with it... ;)
Maybe someone can post some tutorial how to get rid of it, but there is not alot of function on spp anyway sooooo, i dont know... maybe after dancing with shaman drum you can achieve something good..?

Apr 9, 2017*permalink

SETI

Sad news =) Details are amazing!

Apr 9, 2017permalink

desmond2046

I just wish Sigma can develop a *normal* looking camera.

Apr 9, 2017permalink

3DSimmon

How boring that woumd be

Apr 9, 2017permalink

I thought this too, but the abnormal look apes a medium format pro camera, and looks the part for portraitists, for instance.

Apr 9, 2017permalink

kotepum

Nothing wrong with that one... other than the SA lens mount :)

Apr 11, 2017permalink

On my computer the samples gallery "100% details" delivers an enlargement of maybe 125%. This, plus the resizing that has to happen, causes one to think less of the images.
To get true pixel-level view I have to download the JPG and then look at that at 100%.
Perhaps this is working the same way for you?

Apr 9, 2017*permalink

Looks like it is a Chrome browser problem.
I tried Microsoft Edge browser, and it works just fine with "100% details". The problem is definitely there in my Chrome browser. I don't have other browsers so I can't report on any of them, unfortunately.

Apr 9, 2017permalink

Turns out it is a Windows problem (thanks to Bond, who suggested this):
I hit CTRL+, saw 150%, then hit CTRL- a couple times and saw 100%. Problem still there.
I hit CTRL+ and the little box had a place to "Reset to defaults" which I was able to click (you have to do this really fast because that little box is only there a second or so). Problem GONE!
Windows has too many little hooks squirrelled away!

Apr 9, 2017permalink

ilza

If all you want is to reset the browser's magnification to 1:1, then on computers you can try Ctrl-0. AFAIK, it works on all browsers.

Apr 9, 2017permalink

samfan

GO SIGMA!! Thanks Dpr!

Apr 9, 2017permalink

Toselli

I can't download any DNG (I gen an xml file error), am I doing something wrong?

Apr 9, 2017permalink

For my experience with foveon (SD9,10,15 and dp2m) i can say with confidence - THIS ARE THE WORSE what foveon can deliver! As my friend told me - "SANDY" IMAGE QUALITY. How come that frustrating gain can appear as iso100 especially on out-of-focus bokeh area. You simply cant get a clear image like on SD15 witch was absolutely stunning back in the days when 15 mp was alot! The quattro technology totally destroys Foveon beauty! Very disappointed!!!(((
I was hopping Sigma will do something about that ugly gain in new SD models - but no...((((
I think Merrill look overall was the best! As for consider of DP2M model... Details wise and sd15(14) - for colors!
My god - i was waiting for a new SD model for years! And after thay came out - sdq and sdqH... DISAPPOINTMENT!!! VERY SAD!!! went for Sony though...
Otherwise - NEW LENSES - BIG SURPRISE AND I CAN ONLY BOW ON SIGMA FOR THEM!! THANKS YOU SIGMA FOR REINVENT YOUR BEAUTIFUL LENSES!!!!
P.S. Make more mirrorless-miracle lenses PLEASE!

Apr 9, 2017permalink

The noise isn't really that bad. You have to be very fussy to worry about it, especially when the detail is there. Plus it's quite organic looking, like Fuji noise. And unlike the chroma-nasty noise of nikon and everyone else.

When I was accidentally looking at the images upscaled by my browser to 150%, I would agree that the noise is objectionable, as is the loss of detail. But just dump on the upscaled jpegs and go with the native resolution pics.

Apr 9, 2017permalink

Im talking about GAIN here. Its BAD! (noise and gain are totaly different things) What do you men by "organic looking"? Film gain? - nothing in common! been working with film (scaning on digital imagine Fuji Frontier) for years... i repeat - nothing in common!

Apr 9, 2017permalink

TN Args

'Sandy' image quality is only the result of naive processing of Quattro files. There is no need to have that result at all. A surprising number of 'old Sigma' users don't seem to have the basic processing skills to get the best from Quattro cameras, which, done properly, is the most detailed, clear and sharp, colour-accurate images ever from a Sigma camera. The poisonous angry ranting from 'old Sigma' users smacks of a fundamentalist religious movement (because Quattro is not the One True Way) and not of camera users delighting in the best they can get from a camera.

Apr 9, 2017permalink

PLEASE PLEASE!!!! Can you please show me the image that was edited or shoud i say "done properly" with quattro sensor, that dont have that augly sandy effect !!!! I really want to eat my words sort of speak. I really want to be wrong about quattro. But its the reality to me at this moment.
If you can post some clear samples on base iso with quattro PLEASE!!!!
Or maybe some sort of link wiht samples or tutorial.
Prove me wrong!!! PLEASE!!! Thanks!

Apr 9, 2017permalink

Don't you mean 'grain'. Obviously gain on the sensor also increases noise, so maybe you really do mean 'gain'. To be pedantic, this is noise not grain. Grain is only genuinely possible on film. But the point about this 'sandy' look is that it has an organic, grainy look like some film, which is not at all a bad thing. I've always preferred the grainy bokeh of film, for instance, which actually helps to somehow offset bokeh problems. Mind you, this grain is far too fine to really give the advantage except at high ISO. It's barely visible in the original 25MP images.

Meanwhile, a bit of judicious noise cleanup, and you still get stunning detail that beats the previous models. So really all we are talking about here is an extra step in your pp. Nothing to make a big deal about.

Apr 9, 2017permalink

Yes do really mean GRAIN not gain. My bad. But on my taste Merrill was ALOT better and had more organic film like grainy look... The only problem was low MP count witch i prefer... Maybe if thay made Merrill camera with this specs it would be much much better results! And again ive work around with x3f files via SPP 6.0.1 or later (the latest with SDQH support) And tried to get rid of this "organic sandy look" and... nothing! In the end details were GONE.
My perfection on a look would be sd15 but small sensor and low mp made me to sell it.
Any way its just a meter of taste... So no hard fillings. ;)
And The Name is Bond, if you have something to show from quattro sensor with clear look and some tips for editing, I WOULD LOVE TO TRY. Im tellin you i spend hours trying to achieve something usable - no luck.

Apr 9, 2017permalink

TN Args

@GotHotPhoto just turn sharpness slider down in Quattro photos, and do not turn noise reduction slider down too far. The default sharpness setting is too high both in-camera and in SPP, and this creates the so-called sanity look. It is created by software. Also be sure to use the latest version of SPP.

Sadly, images on the internet tend to be processed with default settings, especially on review sites, where it makes sense as a matter of policy but makes Quattro files look worse than they should.

Apr 9, 2017permalink

TN Args, Thank you for a tip but as i already told here with that kind of technique - it takes all the details with it. The screenshots below shows the results with your suggestions (left-no correct, right-with sharpness -2 and noise reduction in the midle - VIA SPP 6.4.1)
https://www.dpreview.com/galleries/8598337897/photos/3618559/01
https://www.dpreview.com/galleries/8598337897/photos/3618560/02
Please check and replay!

Apr 10, 2017permalink

(unknown member)

Very cool stuff. 51mpx Bayer quality packed into 25mpx!

Apr 9, 2017permalink

nicolaiecostel

Looks like what I used to get on my D80 back in the day.

Apr 9, 2017permalink

Strange. When downloaded and viewed at 100% on my computer, the images are smaller and as detailed as one expects from the foveon; which is to breath-takingly.

But here at 100% they appear uprezzed (to simulate 50MP?) with corresponding appearance of a loss of detail and more visible noise. I wouldn't have realised the issue had I not been familiar with the Foveon test images at dpreview, which are stunningly detailed.

Apr 9, 2017permalink

aha!! I've worked it out. I've got chrome set at 150% zoomed because my 1080 screen makes everything too small on my 13" screen.

Apr 9, 2017permalink

I have a similar problem, also with Chrome. But I have my internet page zoom set to 100% and it still over-magnifies the "100% details" view. What exact Chrome setting did you find and fix?

The way I know for sure I have a problem is to download a JPG and then use a viewer such as Faststone to view that image at 100%.

Apr 9, 2017*permalink

Settings->advanced

You'll find it there. However I would be surprised if this was additive. If (on windows) ctrl + or - indicates 100% , as it does in an annoyingly briefly visible indicator, then it should override any chrome-wide setting and actually be at 100%.

I've just tried it and the page setting does override the chrome-wide setting.

It may be that you have configured your computer to higher than 100%. Even in Windows, some apps ignore rescaling and appear at their native rendering, and maybe the viewer you've used does this.

On win10 the system setting for sizing is :

right-click the desktop->pop-up menu
Display Settings -> Opens a settings window
"change the size of text...." etc, which is a single adjustment

Apr 9, 2017permalink

I had everything set to 100% everywhere.
I hit CTRL+, saw 150%, then hit CTRL- a couple times and saw 100%. Problem still there.
I hit CTRL+ and the little box had a place to "Reset to defaults" which I was able to click (you have to do this really fast because that little box is only there a second or so). Problem GONE!
Thanks very much! Windows has too many little hooks squirrelled away!

Apr 9, 2017permalink