Initial impressions Sigma dp2m

[Samouraï]

sigma raw converter produces inaccurate color and that is why people like it because it looks different from the Adobe dominated raw converter market.

the colors that you see in digital files are a product of software interpolation, that is why a RAW files opened in default adobe profile LR will look completely different from one opened in RPP... so its the Sigma RAW converter that you should thank for "fevon color", because all digital cameras capture three grayscale images for each channel the color is added during interpolation in software.



fevon cameras have great resolution and micro contrast due to lack of bayer filter, but then the lack of dynamic range makes them at best a nice substitute to positive color film - that is if you know how to correct colors and process properly.

From what I've read, the Sigma DP Merrill cameras walk away with around 11-12 stops of dynamic range (RAW). Besting the Fuji X100. And it's a stop or two better than my 5D Mk II at the same ISO (I just realized that my DSLR is getting awful long in the tooth).

http://www.techradar.com/us/reviews/cameras-and-camcorders/cameras/compact-cameras/sigma-dp1-merrill-1111027/review/5

Apparently the SD1 Merrill fares even better, splitting the 5D III and the D800 at 200-400 iso with 12.5 stops of dr. http://www.techradar.com/us/reviews/cameras-and-camcorders/cameras/digital-slrs-hybrids/sigma-sd1-merrill-1088902/review/5#articleContent

As for colour: well, that's a little more subjective. But I'm pleased with the accuracy.

 

[Margu]

From what I've read, the Sigma DP Merrill cameras walk away with around 11-12 stops of dynamic range (RAW). Besting the Fuji X100. And it's a stop or two better than my 5D Mk II at the same ISO (I just realized that my DSLR is getting awful long in the tooth).

http://www.techradar.com/us/reviews...ct-cameras/sigma-dp1-merrill-1111027/review/5

Apparently the SD1 Merrill fares even better, splitting the 5D III and the D800 at 200-400 iso with 12.5 stops of dr. http://www.techradar.com/us/reviews...a-sd1-merrill-1088902/review/5#articleContent

As for colour: well, that's a little more subjective. But I'm pleased with the accuracy.

dynamic range = ISO range. what is sigma DP's ISO range, 100 to 1600 = 5 stops, but of course anything above 400 is useless with fevon.

 

[redisburning]

that is not dynamic range at all. dynamic range is a measure of signal to noise strength, period. it is the difference between the maximum recordable value and the noise floor, nothing else.

as far as his comments regarding a lot of it being in the raw processor, well on FM there has been considerable demonstration to that point. if you downsize step-wise and use the right medium-pixel sized edge contrast enhancer, you can get awfully close. you do need considerable resolution, of course, to do so.

 

[dfatty]

that is not dynamic range at all. dynamic range is a measure of signal to noise strength, period. it is the difference between the maximum recordable value and the noise floor, nothing else.

so the genius has a fundamental misunderstanding of dynamic range, to add to his fundamental misunderstanding of how foveon sensors record info for 3 different colors at each pixel site. but i'm sure he reads a lot of dpreview, lol.

 

[burancap]

Looks like this wonderful thread has had a flyby from the current reigning RFF resident Ag Cat. Way to go... Minttu needs to go back on sale to restore order.

 

[Monochrom]

what a beautiful and enlightened chat about dynamic range....

i only can say...i love the dynamic range of my sigma dp2 M...it´s very dynamic...don´t you agree?

😱

 

[noimmunity]

俺ほどマザコンいないんじゃないかな、さみしくなるよなぁ。

inaccurate color

 

[dfatty]

what a beautiful and enlightened chat about dynamic range....

i only can say...i love the dynamic range of my sigma dp2 M...it´s very dynamic...don´t you agree?

😱

that's some fantastic dynamic range, fantastic! lol.

 

[VertovSvilova]

that is not dynamic range at all. dynamic range is a measure of signal to noise strength, period. it is the difference between the maximum recordable value and the noise floor, nothing else.

I agree, and I understand that to be the common definition. DXO-mark defines dynamic range as the difference between saturation of the photosites and the point at which the signal to noise ratio hits 1:1 (where the signal and noise are equal.)

And color sensitivity is "the number of reliably distinguishable colors up to noise. Roughly speaking, two colors are considered as distinguishable if their difference is larger than the noise. In this respect, color sensitivity is the generalization of color to the notion of tonal range." http://www.dxomark.com/en/About/In-depth-measurements/Measurements/Color-sensitivity


I personally don't usually have much issue with the DP2 in respect to color accuracy. Just like other digital cameras, it stumbles on certain colors and of course often in relation to WB. But overall it does pretty well, imho.

Just for fun here are some Macbeth color checker shots done with the DP2 and the Nikon D600. Both cameras were on auto WB (the MacBeth was in ambient daylight.) The first image is an OOC jpeg from the DP2. The second is an OOC jpeg from the Nikon. The third is the DP2 raw developed in Sigma PP. The fourth is the Nikon raw developed in ACR. The last one is the DP2 raw developed in Iridient.

Everything was set at default. There was nothing additional done in the raw developing other than the Sigma PP X3f default, and ACR and Iridient defaults. The only thing I did was turn them all into smaller jpegs via PS. The Nikon OOC jpeg is flatter as I have lower contrast, neutral color, etc., set up in the camera for jpegs.

The DP2 didn't get everything right and neither did the Nikon (just check the files against your own physical MacBeth checker.) But the DP2 did pretty well overall. And in real world use it's been okay for me so far. Nothing has been horribly off or inaccurate. And I don't think that I'm seeing the world only in some sort of unique Foveon inaccurate color universe. My images seem pretty normal and pretty much 'real world looking.' To me, the DP2 seems no different than other camera sensors and their color algorithms in respect to both its accuracy and inaccuracy.

 

[VertovSvilova]

btw, here's a 'one-click' gray balance correction in PS on the DP2 raw file developed in Sigma PP. The other colors will fall where they fall. The Nikon did get the gray balance better, but there's nothing about the DP2 that can't easily be fixed (same with Nikon or any other color digital device.)

I feel that 'accurate color' is somewhat of an oxymoron in the sense that color is perceptual. Sure, we can scientifically say what color is (wavelength reflected off of a subject, etc..), but color is interpreted by differing eyes and differing brains and certainly by differing digital devices. In the 'old' film days we used to stick a Kodak box in the first frame as a color reference. In the digital age, we can correct color on the fly and while tethered to a computer, etc.. Product photographers can correct the product color accurately with the product right there in hand as their reference while they photograph it. Other colors in the scene may not be so 'accurate' but it will be correct for the actual product, etc..

Anyway, I personally don't see any big issues with the Foveon sensors in respect to color rendition. They are just as 'accurate' as any device and can easily be corrected to the 'accuracy' that one needs relative to the subject matter. Color is a trickster for sure. And there are so many variables in respect to how and where those interpreted colors will be viewed, and in relation to other colors in the image and even outside of the image. 'Accuracy' can end up as somewhat of a compromise.

 

[Keith]

The Sigma threads are known for their harmony and lack of bull**** ... please lets just ignore what the poster who seems determined to change that has said and move on.

If this fails to happen I'm selling all my Sigma stuff! 😀

 

[Margu]

btw, here's a 'one-click' gray balance correction in PS on the DP2 raw file developed in Sigma PP. The other colors will fall where they fall. The Nikon did get the gray balance better, but there's nothing about the DP2 that can't easily be fixed (same with Nikon or any other color digital device.)

I feel that 'accurate color' is somewhat of an oxymoron in the sense that color is perceptual. Sure, we can scientifically say what color is (wavelength reflected off of a subject, etc..), but color is interpreted by differing eyes and differing brains and certainly by differing digital devices. In the 'old' film days we used to stick a Kodak box in the first frame as a color reference. In the digital age, we can correct color on the fly and while tethered to a computer, etc.. Product photographers can correct the product color accurately with the product right there in hand as their reference while they photograph it. Other colors in the scene may not be so 'accurate' but it will be correct for the actual product, etc..

Anyway, I personally don't see any big issues with the Foveon sensors in respect to color rendition. They are just as 'accurate' as any device and can easily be corrected to the 'accuracy' that one needs relative to the subject matter. Color is a trickster for sure. And there are so many variables in respect to how and where those interpreted colors will be viewed, and in relation to other colors in the image and even outside of the image. 'Accuracy' can end up as somewhat of a compromise.

its great that you went and did your tests in order to find out for yourself, but unfortunately that simple test is only telling you about only one basic aspect of color property, the hue.

the other color properties, saturation and luminosity are the real important factors in digital photography. for example fevon reds are muted, and by that extension all colors that are a mix of red, which means all the important warm colors, including skin tones will have the same saturation and lightness as the cool colors making the color contrast of photos look bland and flat.

here is an example of a very good photo, but look at it more carefully and you realize that the color contrast is flat, the reddish buildings have the same lightness and saturation as the blue sky and the sky itself looks completely off and unnatural because the clouds that fall in the red channel have the same luminosity as the blue sky which falls in the blue channel! that is completely unacceptable in any color correction 101 manual.

http://www.flickr.com/photos/ycheng_12345/11571469944/in/pool-sigmadp2m#

people might argue that at least there is b&w which looks good because of the detail and microcontrast, but b&w is all about tonal range which is based on color contrast, and lack of color contrast combined with a limited dynamic range means the fevon b&w will look good but it will never look natural and even remotely like b&w film.

 

[VertovSvilova]

Sorry, but I can't make an assessment from a random photo on a photo hosting web site; I can only assess my own images and knowing how they were produced.

Anyway, in your initial post, you had an issue with the "accuracy" of colors from the Foveon chip. You suggested that the color world of the Sigma designed raw processor (Sigma PP) was different than that of the Adobe raw interpolation from conventional Bayer array sensors. The MacBeth color checker is used for color accuracy. The chart's color patches have spectral reflectances and can be measured using Munsell notation numbers which is based on the three color dimensions: hue, value and chroma. The Macbeth is indeed a useful tool for determining color 'accuracy.' There has to be a measurable constant otherwise what you see as 'violet' will not be what I see as 'violet' etc., etc.. This is pretty basic. Nonetheless, the Foveon chip replicates these constants (the measured color patches) as well as does any other sensor design.

However, the 'disadvantage' to the Foveon sensor is that there is less color separation. A Bayer array must rely on the 'natural' color absorption of the chip's design itself which is then interpolated by the processing algorithms for its color mapping. The color coming from the Foveon is less saturated than what one would see from the Bayer. So with the Foveon raw processing, the color needs to be amplified which can then lead to increased chroma noise. So therefore, the signal to noise is higher. But that 'amplification' can be controlled in post processing technique and by dealing with Sigma's PP algorithms oneself. I have no issues in dealing with Sigma's raw processor in respect to its color saturation. I find that it's a bit overboard with handling this color saturation issue, but that's easily dealt with. One can also use Iridient (Mac only) which sometimes handles this better, depending on the subject matter (I use both depending on subject matter, lighting, color, etc..)

Personally, I've yet to fight colors with the DP2. If it was indeed problematic I would have dumped the camera right away. I'm not a fan of any specific sensor type, I only use the tools and materials that work for me (including film.) I picked up the DP2 not so much because of the Foveon chip, but because I needed a smaller form factor that could produce files similar to the D800E for a particular project I'm doing. So far, it's been performing quite well and with no issues whatsoever in respect to color rendering. I might add that I also print (Océ Lightjet prints on RA-4 paper) and that adds another layer of dealing with color interpretation and "accuracy."

btw, I notice that you keep spelling Foveon as "fevon."

 

[Margu]

color separation is color contrast and if that is linear such as the case with fevon you'll get an image that looks unusual or even pleasing at first but its novelty is going to wear out over time because its ultimately unnatrual, that is not how our eyes sees the colors.

the image which i posted is very typical of fevon, especially the sky bit. in a picture like that if you don't have separation/contrast between the sky and the objects on the ground or the sky and clouds then you have a flat image. the feeling of depth is lost.

once you get over the close up images all shot for detail and seriously get into proper photographer where dof, contrast, skin tones and color contrast is important then you're not going to be happy with a fevon.

so the fevon colors are inaccurate in all three factors, hue, saturation and lightness. you can fix the hue (hue is assigned during raw conversion - white balance) but fixing the hue makes little difference sense if the other two factors of saturation and lightness are incorrect.

the picture of sky and clouds has to have enough separation where its obvious that the white clouds are floating in the sky and are not "part" of the sky. with almost all fevon images that i have seen where sky and clouds are involved, there is no separation between the clouds and sky.

 

[VertovSvilova]

once you get over the close up images all shot for detail and seriously get into proper photographer where dof, contrast, skin tones and color contrast is important then you're not going to be happy with a fevon.

so the fevon colors are inaccurate in all three factors, hue, saturation and lightness. you can fix the hue (hue is assigned during raw conversion - white balance) but fixing the hue makes little difference sense if the other two factors of saturation and lightness are incorrect.

the picture of sky and clouds has to have enough separation where its obvious that the white clouds are floating in the sky and are not "part" of the sky. with almost all fevon images that i have seen where sky and clouds are involved, there is no separation between the clouds and sky.

As I mentioned, I'm using the DP2 for a specific project that excludes the subject matter that you are defining. Therefore, it's a proper tool for my particular use. If it were my only tool, then perhaps I would eventually notice those issues you find problematic with the Foveon sensor. But in the meantime, I find it to be working for me in this context. Just like I use certain film emulsions for certain projects and subject matter, I will also choose certain digital tools.

In the end, the DP2 is ergonomic, fully functionally, very small to carry, and it produces the type of files I need for this project. But I also do find the DP2 limited for many uses. However, it was picked specifically for this current body of work which also required a size factor. My printer's test prints have so far been very good for the size and type of print required (RA-4.)

I'll end by saying that it's a good tool for my purposes. And so far I've been quite satisfied to work with both its positive attributes and with its deficiencies. Most all tools are about compromise but an adept author can always work around that.