The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Sigma Quattro series - Enduring issues

Quentin_Bargate

Well-known member
The last time I owned a camera with a serious sensor flaw was the Kodak 14n. After a while, Kodak replaced it with the near identical SLRn which used a modified sensor and offered a sensor upgrade for owners of the 14n. I paid for the upgrade. The upgraded camera was known as the 14nx.

Unfortunately for Kodak, the whole 14n debacle ended their involvement in pro digital cameras.

I still have my 14nx around somewhere. It might soon be joined by the Quattros.
 

pingu666

New member
yeah i think the physical design is much better on the merrils

id like a few extra buttons, tilt screen, hsm focusing with the window etc, evf, optional battery grip, another scroll wheel, bulb mode, remote trigger, onboard flash

and a model with a M mount or E mount :D
 

The Ute

Well-known member
While out for a walk Sigma took the wrong fork in the road.
They need to get back onto the correct path ASAP.
Give us what we truly want. Not some DoDo bird know as the Quattro.

;)
 

Malina DZ

Member
Bearing in mind that the Quattro sensor is still based on 3 layer architecture, I wonder why Green and Red layers don't capture luminance information? Is it cancelled out somehow to avoid file size bloating?
"While the top layer captures both luminance and color information, the middle and bottom layers with their larger surface area capture color information only." (Source: Technology | dp Quattro | Cameras | SIGMA GLOBAL VISION)

Well, Sigma really does cancel that luminance detail out to slightly reduce high ISO noise (1 stop) and processing time. From IR interview:
DE: I see. Because each layer has all the colors, you can correlate. So you can look at the top layer and say "Okay, we know we've got some red here," then you look at the red layer to see how much is there, and you can sort of take that out.
SR: Precisely -- it creates the correlation automatically, and therefore, you can remove some redundant information that you didn't actually need. Part of the advantage that you get from that is that you are then able to increase the signal.
Sigma Q&A Part II: Does Foveon’s Quattro sensor really out-resolve conventional 36-megapixel chips?

That's where the Quattro failure lies in (in my eyes) - omitting red and green color channels luminance detail. If Quattro sensor architecture is indeed as presented, there is a chance for Sigma to adjust image processing for collecting full RGB information, even though high ISO noise advantage and processing speed will be sacrificed. If implemented, I would consider adding a Quattro-based camera to my arsenal.
 

xpatUSA

Member
Bearing in mind that the Quattro sensor is still based on 3 layer architecture, I wonder why Green and Red layers don't capture luminance information? Is it cancelled out somehow to avoid file size bloating?
"While the top layer captures both luminance and color information, the middle and bottom layers with their larger surface area capture color information only." (Source: Technology | dp Quattro | Cameras | SIGMA GLOBAL VISION)

Well, Sigma really does cancel that luminance detail out to slightly reduce high ISO noise (1 stop) and processing time. From IR interview:
DE: I see. Because each layer has all the colors, you can correlate. So you can look at the top layer and say "Okay, we know we've got some red here," then you look at the red layer to see how much is there, and you can sort of take that out.
SR: Precisely -- it creates the correlation automatically, and therefore, you can remove some redundant information that you didn't actually need. Part of the advantage that you get from that is that you are then able to increase the signal.
Sigma Q&A Part II: Does Foveon’s Quattro sensor really out-resolve conventional 36-megapixel chips?

That's where the Quattro failure lies in (in my eyes) - omitting red and green color channels luminance detail. If Quattro sensor architecture is indeed as presented, there is a chance for Sigma to adjust image processing for collecting full RGB information, even though high ISO noise advantage and processing speed will be sacrificed. If implemented, I would consider adding a Quattro-based camera to my arsenal.
Color theory is not an easy subject to grasp.

And the Sigma/Foveon statements are incorrectly dumbed-down for non-engineers. The best statement of what the sensor captures is their spectral response graph which is virtually unchanged since the SD9, including that of the Quattro.

Ted
 
Last edited:

Malina DZ

Member

xpatUSA

Member

tagscuderia

New member
Hi Quentin, regards the highlight clipping, does this occur in 5MP RAW mode? I've seen comparisons showing different rendering of colour and tone in this mode so it would appear that the data is processed differently.
Seem to recall that others have dug into the RAW files and found information in the "Green/Red" layers that goes unused if the "Blue" layer is blown - presumably because it's a quarter of the resolution.
If the 5MP mode is processed as a true Foveon image then perhaps the blown highlight issue is resolved? Personally a mix between 5MP and bracketing would allow me to consider a DP0.
Cheers,
Tom.
 
Last edited:

Quentin_Bargate

Well-known member
Interesting question, Tom. I have never tried as I cannot see any situation in which I would use 5mp raw, I am afraid. And if you are right about the red and green layers not being used if blue is blown, that might possibly explain why this issue affects the Quattro and not true 3 layer Foveon sensors.
 

xpatUSA

Member
Hi Quentin, regards the highlight clipping, does this occur in 5MP RAW mode? I've seen comparisons showing different rendering of colour and tone in this mode so it would appear that the data is processed differently.
Seem to recall that others have dug into the RAW files and found information in the "Green/Red" layers that goes unused if the "Blue" layer is blown - presumably because it's a quarter of the resolution.
If the 5MP mode is processed as a true Foveon image then perhaps the blown highlight issue is resolved? Personally a mix between 5MP and bracketing would allow me to consider a DP0.
Cheers,
Tom.
Tom,

I've looked at a HI res. Quattro file in RawDigger and, as expected, the "red" and "green" layers are indeed 5MP. As far as I know, in LO res. on the Quattro, the "blue" (top) layer gets binned on-chip 2x2 and the conversion to RGB thereafter is done in the same way as for all previous Foveon sensors - with all that that implies :)

If anybody cares to post a Quattro 5MP raw file somewhere, I'll be glad to look at it and report back here . .

Ted
 
Last edited:

tagscuderia

New member
Logically this highlight clipping only applies to the Quattro. A combination of the luminance layer's small well-size and 1/4 res "Green/Red" layers; a true Foveon sensor has the ability to fall back on the lower layers to render the pixel. N.B, I'm guessing based on what Ted and others have found.
Unfortunately, if I'm correct then it's an oversight by Sigma's engineering team and a critical (fatal?) flaw in the Quattro architecture!
@Quentin, if you could provide a 5MP RAW of a scene that exhibits the issue at 20MP for Ted, we could at least establish how Sigma is treating data in this mode. I would but I'm waiting on the DP0.
Cheers,
Tom.
 

tagscuderia

New member
Unfortunately not it would seem.
The only things holding me back from the DP0 (besides availability) is this highlight issue and weather sealing. I don't need 20MP files and 5MP Foveon files can still be printed at decent sizes so I had everything crossed for the Small RAW mode. If I take the plunge I'll be sure to post my findings here.
 

Quentin_Bargate

Well-known member
Unfortunately not it would seem.
The only things holding me back from the DP0 (besides availability) is this highlight issue and weather sealing. I don't need 20MP files and 5MP Foveon files can still be printed at decent sizes so I had everything crossed for the Small RAW mode. If I take the plunge I'll be sure to post my findings here.
I never use the smaller files, but if I get the chance, I will post a couple of 5mp raw files this weekend.
 

xpatUSA

Member
Unfortunately not it would seem.
The only things holding me back from the DP0 (besides availability) is this highlight issue and weather sealing. I don't need 20MP files and 5MP Foveon files can still be printed at decent sizes so I had everything crossed for the Small RAW mode. If I take the plunge I'll be sure to post my findings here.
Hello again Tom,

Thanks to Quentin I was able to look at a couple of LO res. Quattro shots with RawDigger (RD).

The raw numbers for saturation are 14,336. Just for info. Sigmas raw numbers rarely match the ADC outputs, there's always something done in-camera before writing to X3F. The channels are like Merrills in relative size, namely they are the same pixel size which indicates to me that the blue sensels are binned on-chip using Foveon's VPS technology. In the lesser-exposed image there are spikes at 14,336 in the R & G channel histograms. These are the more sensitive "over-exposure" sensor pixels doing their job and can be ignored as far as image quality is concerned.

A blue channel export from RD is the same file size as a green channel export.

Based on the grass/flowers scene which is slightly blown in raw, the channels saturate far more evenly than those of the Merrill. All in all, I would be quite tempted by a DSLR Quattro but would probably only use it in LO res because I don't print.

I put some files for you up here: Index of mySite/phot/temp

Two channel exports
Two raw histograms
One RGB export

Enjoy, they won't be there for ever ;)

Ted
 
Last edited:

Stoneage

Member
This image illustrates the problem well. It is a medium contrast image of a slightly messy kitchen work surface (mine - sorry!). At first sight, it is richly detailed and with excellent colour. But look again at the lemons and the tulip tips. The yellow of the lemon skin is completely bleached with jagged edges, as are the tulip tips, in a very unnatural looking way
Quentin, would it be possible to make the X3F of the kitchen photo available?
I would like to try an open source converter that some guys in the dpreview forum developed. I have a similar situation where the converter did a better job on the weird highlight clipping than SPP.

Stoneage
 
Top