DP2m: first images, first problems

I started my last thread a long time ago, in the meantime I bought a Fuji X-E1, tested with my short telephoto Voigtlander lens, re-sold it (liked the camera and the color, but not at all the rendering of textures), my DP1s is now frozen with the lens extended (an evident design mistake that Sigma never recognized, this is not serious at all), my DP2 emits strange sounds but it is still working, I decided then to try the DP2m

(•) First images
My first images are accessible following the link below,

**USE UP/DOWN KEY ARROWS OR MOUSE WHEEL TO NAVIGATE**

the gallery (just a collection of images, not a project) is optimized for 1400px wide or wider screens, it works also for smaller screens but you need to rely on the downsizing capabilities of your browser

you can enjoy (I hope) the images and stop here avoiding the bla bla about technical problems

http://www.photografeet.org/2015dp2mfirst.html

(•) First (usual) problems
(1) My main concern regards the purple/green areas in the clouds that I obtain in highly contrasted skies. Following the recommendation found in this forum I tried (in LR5): desaturating the purple/magenta channels; moving the hue of the purple channel towards the blue; using color noise reduction; using local moire/fringe correction. Nothing really works for me, there are some improvements in some cases but the purple/green areas often do not completely disappear, you can find some of them also in the images I posted. Maybe this is linked to the (incorrect) exposure of the photo, even though, honestly, I’m not sure of that. But the exposure is a sort of nightmare with this DP2m. In highly contrasted situations the meter is off up to 2 stops, the live view histogram and the screen preview are totally useless (in some cases the histogram does not change even correcting the exposure of 4 stops, it does not move at all, the same for the preview (my firmware is v1.05, the DP2 is imprecise but it is relatively consistent)) and highlights are quite easy to clip.

(2) My second concern is about micro contrast. In some case it is really too high for my taste, I like better the one of the original DP2. To reduce it I decrease the contrast and the presence (in LR5), but still I’m not completely happy with the results.

(3) The WB and the colors are improved w.r.t. the ones of the DP2. I use AWB that I correct to my taste in SPP, but some times it is quite difficult to find a good WB. I know that somebody hate AWB, but when I try all the other WB options in SPP they are really far from the one I like (in general they have a strong yellowish cast). I never understood if there is a difference to change the WB in SPP or to use directly a specific WB when one takes the image.

Every comment/advice/suggestion is very appreciated. Many thanks in advance.

mclaudio wrote:

you can enjoy (I hope) the images and stop here avoiding the bla bla about technical problems

http://www.photografeet.org/2015dp2mfirst.html

(1) My main concern regards the purple/green areas in the clouds that I obtain in highly contrasted skies.

Hello Claudio, I'll let others respond to that. Lin Evans might be able to help.

(2) My second concern is about micro contrast. In some case it is really too high for my taste, I like better the one of the original DP2. To reduce it I decrease the contrast and the presence (in LR5), but still I’m not completely happy with the results.

How about Gaussian smoothing with a fairly low radius? Assuming that LR has it. Have you tried low resolution (never medium) mode? I use it most of the time unless I really, really need high resolution.

(3) The WB and the colors are improved w.r.t. the ones of the DP2. I use AWB that I correct to my taste in SPP, but some times it is quite difficult to find a good WB. I know that somebody hate AWB, but when I try all the other WB options in SPP they are really far from the one I like (in general they have a strong yellowish cast). I never understood if there is a difference to change the WB in SPP or to use directly a specific WB when one takes the image.

There is no difference for the review image in SPP. Either way, WB is set by the values passed as meta-data in the X3F file to SPP. All the WB values are passed so, if you change the WB in SPP, it reloads the X3F using the newly-selected WB's from-the-camera values.

The image in the camera LCD may use different (faster) parameters. You can check that by comparing the X3F embedded JPEG to a JPEG conversion saved from SPP. Someone here will know how to extract the embedded JPEG. There are several methods; I use FastStone Viewer but I think it can be done out of SPP - anybody?

We seem to differ considerably as to WB (I have the SD1 Merrill). I never use AWB, which is easily fooled by many types of scene. I shoot with the in-camera WB selected to my best guess for the scene. Then, in SPP, I color-pick a neutral object in the area of interest in the captured image. I always shoot raw.

mclaudio wrote:

I started my last thread a long time ago, in the meantime I bought a Fuji X-E1, tested with my short telephoto Voigtlander lens, re-sold it (liked the camera and the color, but not at all the rendering of textures), my DP1s is now frozen with the lens extended (an evident design mistake that Sigma never recognized, this is not serious at all), my DP2 emits strange sounds but it is still working, I decided then to try the DP2m

(•) First images
My first images are accessible following the link below,

**USE UP/DOWN KEY ARROWS OR MOUSE WHEEL TO NAVIGATE**

the gallery (just a collection of images, not a project) is optimized for 1400px wide or wider screens, it works also for smaller screens but you need to rely on the downsizing capabilities of your browser

you can enjoy (I hope) the images and stop here avoiding the bla bla about technical problems

http://www.photografeet.org/2015dp2mfirst.html

(•) First (usual) problems
(1) My main concern regards the purple/green areas in the clouds that I obtain in highly contrasted skies. Following the recommendation found in this forum I tried (in LR5): desaturating the purple/magenta channels; moving the hue of the purple channel towards the blue; using color noise reduction; using local moire/fringe correction. Nothing really works for me, there are some improvements in some cases but the purple/green areas often do not completely disappear, you can find some of them also in the images I posted. Maybe this is linked to the (incorrect) exposure of the photo, even though, honestly, I’m not sure of that. But the exposure is a sort of nightmare with this DP2m. In highly contrasted situations the meter is off up to 2 stops, the live view histogram and the screen preview are totally useless (in some cases the histogram does not change even correcting the exposure of 4 stops, it does not move at all, the same for the preview (my firmware is v1.05, the DP2 is imprecise but it is relatively consistent)) and highlights are quite easy to clip.

(2) My second concern is about micro contrast. In some case it is really too high for my taste, I like better the one of the original DP2. To reduce it I decrease the contrast and the presence (in LR5), but still I’m not completely happy with the results.

(3) The WB and the colors are improved w.r.t. the ones of the DP2. I use AWB that I correct to my taste in SPP, but some times it is quite difficult to find a good WB. I know that somebody hate AWB, but when I try all the other WB options in SPP they are really far from the one I like (in general they have a strong yellowish cast). I never understood if there is a difference to change the WB in SPP or to use directly a specific WB when one takes the image.

Every comment/advice/suggestion is very appreciated. Many thanks in advance.

My experience when starting with the Merrill was that it took a long while to change habits from the pre-Merrill cameras and I still there is a way to go to get the best out of the Merrill files.

With regards to the blotching - I use ISO 100 and with careful exposure find that I can now have the chroma in SPP down/off (slider moved fully left) and have very little purple green noise. If I need to I will use the colour noise reduction (slider moved to about 15) in ACR or Lightroom to fix.

Micro contrast can be too high with some subjects - my approach after all else fails in terms of good exposure, reduction of contrast, levels adjustment etc is to increase the luminance noise reduction in ACR or Lightroom which smoothes the image.

WB is generally best, I find, if I make a Custom WB using an Expodisc.

If these are your first efforts with the Merrill I am impressed.  They are beautiful.

Jan

mclaudio wrote:

I'm not referring to the purple noise in the shadows, I'm talking about quite huge purples areas in the skies. The chroma noise in LR5 does not affect these purple areas.

Hmm I'm not sure I have seen much of these purple areas in my own work. Are you using ISO 200 rather than 100? Are you using Adobe RGB colour space?

I know, but I'm not a photographer, I'm a walker with a camera, I'm outside for walking and enjoy the landscape not for taking pictures, and the Expodisc starts to become too elaborate for me.

Fair point except I'd add 'a walker with a camera' who takes very fine photographs!!

mclaudio wrote:

What really gets me sick are the quite huge purple areas in the skies, these do not seem to me the same blotches one finds in the shadows, they are present also in the highlights. What is strange is that in the Develop module of LR5 (using ProPhoto RGB) these purple areas are much less visible than in the Library module that, if I correctly understood creates sRGB previews. I don't know if the problem is also linked to mappings/translations between color spaces.

Claudio, is there anywhere you could put the worst X3F on the web so I can download it for analyis in RawDigger and for converting in RawTherapee?

I would post the conclusions, if any, here for your information of course.

Also there is Kalpanika, have you tried it yet?

Here's the X3F opened in RawDigger's RGB render view:

There are your purples and greens in all their glory. RawDigger uses a fairly simple conversion from raw to RGB and includes none of the fancy stuff that SPP applies even before you see it's review image. And, to me, that says there is a LOT of fancy stuff needed by the DP2M

I also opened the X3F in RawTherapee:

I used RT's SD1M 'overcast' input profile. Still, the obvious discoloration looks depressingly familiar. RawTherapee uses conversion code derived from Coffin's DCraw and clearly does not have SPP's fancy bits either!

I have to say right away that none of my SD1 Merrill captures come even close to being that bad - when opened in either RawDigger or RawTherapee. To me, that points the finger firmly at the DP2M and maybe it's shorter lens distance from the sensor and different lens construction than those for the Sigma mount. Glad I sold mine

My conversion in SPP 5.5.3 looks much like yours:

That locks you into SPP for conversion, IMHO, or perhaps giving Kalpanika a try (I lost my copy in my latest computer death). Sometime, I'll reload it and see what it comes up with in your X3F or maybe someone here can give it a shot and post their findings.

pedrolito wrote:

hi,

I was just curious and tried out the xf3-file.

I converted using SPP (values from top downwards: -0.5/0/0/0/0/-0.8/0.4, daylight) then edited in RT 4.2.447 and corrected the wb there also.

What I found was some red fringes but no purple blotches or clouds. None. Period.

You should try again. The photos are great, your cam is fine.

Cheers

P.

I too developed the X3F and couldn't find any purple discolouration. This is curious.

Saved X3F Settings:

Exposure: -0.2

Contrast: -0.5

Shadow: 0.0

Highlight: +0.2

Saturation: -0.2

Sharpness: 0.0

X3 Fill Light: -0.1

I've converted your image in SPP 6.3 in neutral colour mode with no adjustments and there is no  significant blotching visible on hi-res screen.

The underscore  in the file name suggests  that you're using AdobeRGB colour space in camera settings, instead of sRGB, and the exif confirms that.

half size jpg from spp

full size jpg from spp

what can be seen as uniform colour area in the sky/mist is a result of jpg interpolation of pixels with no signifcant colour difference into a one blop . That's how jpg compression works  to reduce the file size .

Marek L wrote:

I've converted your image in SPP 6.3 in neutral colour mode with no adjustments and there is no significant blotching visible on hi-res screen.

full size jpg from spp

what can be seen as uniform colour area in the sky/mist is a result of jpg interpolation of pixels with no signifcant colour difference into a one blop . That's how jpg compression works to reduce the file size .

Can I ask what JPEG quality setting you used for the above image?

Reason I ask is that EXIF for yours says YCbCR 4:2:0 meaning that it has sub-sampling but Claudio's JPEG says YCbCR 4:4:4 meaning it has no sub-sampling and, therefore, no interpolation. That puts his SPP slider setting at 9+ and yours probably less?

No disrespect intended to your raw conversion . . .

Sorry, please forget the question, your original size is in 4:4:4 (not sub-sampled) same as Claudio's. Apparently, DPR uses lower quality when downsizing the original to the post image. Sure fooled me

mclaudio wrote:

(•) First images

WOW - it seems you have used at least one other camera before I am impressed. Nice images.

(•) First (usual) problems
(1) My main concern regards the purple/green areas in the clouds that I obtain in highly contrasted skies. Following the recommendation found in this forum I tried (in LR5): desaturating the purple/magenta channels; moving the hue of the purple channel towards the blue; using color noise reduction; using local moire/fringe correction. Nothing really works for me, there are some improvements in some cases but the purple/green areas often do not completely disappear, you can find some of them also in the images I posted. Maybe this is linked to the (incorrect) exposure of the photo, even though, honestly, I’m not sure of that. But the exposure is a sort of nightmare with this DP2m. In highly contrasted situations the meter is off up to 2 stops, the live view histogram and the screen preview are totally useless (in some cases the histogram does not change even correcting the exposure of 4 stops, it does not move at all, the same for the preview (my firmware is v1.05, the DP2 is imprecise but it is relatively consistent)) and highlights are quite easy to clip.

The purple/green is somewhat mysterious. Sometimes there is none, and then, all of a sudden you have lots of it. I know why it is there, but I have a hard time predicting when it is a risk for it.

(2) My second concern is about micro contrast. In some case it is really too high for my taste, I like better the one of the original DP2. To reduce it I decrease the contrast and the presence (in LR5), but still I’m not completely happy with the results.

For some mysterious reason, Sigma has chosen to (by default) enhance local contrast greatly for Merrill sensors. In later versions of SPP you are supposed to be able to "fix" that by using portrait mode. OK - that gritty, dirty look when the local contrast is exaggerated has its charm, but gets tiresome in the long run IMHO. BTW, it is not really micro contrast, which is at pixel level, but rather like using "clarity" in Lightroom, adding higher contrast to textures.

(3) The WB and the colors are improved w.r.t. the ones of the DP2. I use AWB that I correct to my taste in SPP, but some times it is quite difficult to find a good WB. I know that somebody hate AWB, but when I try all the other WB options in SPP they are really far from the one I like (in general they have a strong yellowish cast). I never understood if there is a difference to change the WB in SPP or to use directly a specific WB when one takes the image.

Colorwise, the Foveon sensors are strange beasts. White balance is not, like for RGB cameras, a matter of balance between the RGB channels. There are different conversion matrices for each white balance. Many have been the problems seen when trying to fix colors. Now, the later cameras (e.g. Merrill and Quattro) are better than ever regarding color.

So, if you have a strange lighting, you'd better measure it using AWB or using custom WB. Can be a bit tricky otherwise.