Pages: 1 [2]   Go Down

Author Topic: Doug Peterson's article gives good insight into development of the IQ250  (Read 21035 times)

jerome_m

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 670
Re: Getting back to colour rendition…
« Reply #20 on: February 09, 2014, 08:27:30 am »

A high color sensitivity offers opportunities to very precicely tweak certain color differences, either making them more pronounced, or less obvious.

I don't think it is that simple. If the camera sees differences where the human eye does not, removing these differences will be very difficult in post. For example: if a camera exagerates skin defects on a human face (e.g. veins or the blue-green under the eyes), removing these defects will be lots of work.
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Getting back to colour rendition…
« Reply #21 on: February 09, 2014, 08:57:57 am »

I don't think it is that simple. If the camera sees differences where the human eye does not, removing these differences will be very difficult in post. For example: if a camera exagerates skin defects on a human face (e.g. veins or the blue-green under the eyes), removing these defects will be lots of work.

Hi Jerome,

Skin offers specific challenges, not necessarily related to high color discrimination but rather to profiling. Human skin reflects a lot of IR, but does so diffusely, more deeply from within the skin tissue. The surface reflects/absorbs different colors (including some (semi-)specular reflection of ambient colors), depending on pigmentation and vein structure. Eliminating the IR contribution, which registers in all three CFA channels and desaturates and shifts color, is important for accurate skin color, and the profiling + WB will then allow to make it more pleasing to our eyes.

The profiling can always assign similar (but different) colors to the same output color, therefore discrimination can be lowered afterwards. When no discrimination exists to begin with, then no manipulation will be possible (other than an overall shift in like colors).

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Getting back to colour rendition…
« Reply #22 on: February 09, 2014, 09:00:08 am »

True, but this is not what we were talking about. This is about a given patch of pigment being rendered of a slightly different colour than the one the human eye would perceive. Metamerism is about two given patches of different pigments which appear of the same colour to the human eye but are rendered of a different colour by the camera.

I think the proper term here would be metameric failure. Metamerism is when two different SPDs appear to be the same color. It can be desirable, since color photography would not be possible without metamerism.

Wikipedia: "In colorimetry, metamerism is the matching of apparent color of objects with different spectral power distributions. Colors that match this way are called metamers."

Bill
Logged

jerome_m

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 670
Re: Getting back to colour rendition…
« Reply #23 on: February 09, 2014, 09:31:12 am »

Eliminating the IR contribution, which registers in all three CFA channels and desaturates and shifts color, is important for accurate skin color

I have read about that theory before, but I don't think IR is the real problem here. First, metameric failure can happen entirely within the visible bands. Second, most digital cameras are actually very well filtered for IR. Astronomers know the problem, since many cameras already fail to register the H-alpha ray at 656.28 nm. For example: some solar telescopes are designed to show that particular band. They present a visible but quite red image to the human eye. I know from experience that most digital cameras fail to register an image. I had to remove the IR filter on a camera to solve that problem.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Getting back to colour rendition…
« Reply #24 on: February 09, 2014, 10:34:38 am »

Hi,

The description of SMI given by DxO is enclosed.

Best regards
Erik



The Color Sensitivity Index and Sensitivity Metamerism Index are metrics for the discrimination sensitivity, not color accuracy. A high color sensitivity offers opportunities to very precicely tweak certain color differences, either making them more pronounced, or less obvious.

Logged
Erik Kaffehr
 

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Getting back to colour rendition…
« Reply #25 on: February 09, 2014, 11:07:10 am »

Hi,

The description of SMI given by DxO is enclosed.

Hi Erik,

Accurate is mentioned as the possibility to discriminate between metameres. When the sensor cannot discriminate between pure Yellow, and a mix of Red and Green, it cannot be white balanced accurately. It does not mean that very subtle color nuances can be accurately discriminated, that aspect is described in the other metrics mentioned on that DxO page.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
color accuracy: Kodachrome and Velvia did not succeed due to accuracy
« Reply #26 on: February 09, 2014, 11:45:30 am »

I should add that speed isn't the only reason for reduced color accuracy in most DSLRs and consumer cameras, by intentionally manipulating the color response, it's possible to make images look better under poor lighting conditions, which tend to have "spikes" in the color spectrum.
I propose that there can be a big difference between "objectively accurate color" and "color that works well for the main intended usage". Kodachrome and Velvia were both very popular is some areas of professional and artistic photography, but more for their desirable distortions of color than for superior color accuracy. (AFAIK, Kodachrome has a slight skew to red that was flattering in portraits  -- at least of caucasians.)
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: color accuracy: Kodachrome and Velvia did not succeed due to accuracy
« Reply #27 on: February 09, 2014, 12:51:38 pm »

Hi Bart,

Thanks for info.

I did some comparisons with, comparing Sony Alpha 99 to Nikon D 610 and I think they were quite similar, but I need to look into this more.

Best regards
Erik

I propose that there can be a big difference between "objectively accurate color" and "color that works well for the main intended usage". Kodachrome and Velvia were both very popular is some areas of professional and artistic photography, but more for their desirable distortions of color than for superior color accuracy. (AFAIK, Kodachrome has a slight skew to red that was flattering in portraits  -- at least of caucasians.)
Logged
Erik Kaffehr
 

markd61

  • Newbie
  • *
  • Offline Offline
  • Posts: 36
Re: color accuracy: Kodachrome and Velvia did not succeed due to accuracy
« Reply #28 on: February 09, 2014, 02:17:02 pm »

I propose that there can be a big difference between "objectively accurate color" and "color that works well for the main intended usage". Kodachrome and Velvia were both very popular is some areas of professional and artistic photography, but more for their desirable distortions of color than for superior color accuracy. (AFAIK, Kodachrome has a slight skew to red that was flattering in portraits  -- at least of caucasians.)

I agree. When I started working in custom photo labs in the 70's Vericolor instruction sheets advised the photographer to photograph a gray card at the beginning of a roll to provide  reference to the lab for printing. Those instructions were intended for the early printers that had fixed filtration and timing and not the newer printers that were entering the market that allowed a skilled printer to override the printer to correct the images. On more than one occasion I had a photographer that insisted that I print an entire roll at the setting that produced the perfect gray card frame. Of course, the prints were awful as the photographer did not understand that their exposure variations, subject matter, lighting conditions and the photographer's intent all affected the results of subsequent frames.
To be sure,  we were not being perverse but we printed them to illustrate the error and produced a set of correctly printed (and pleasing) prints to compare.
Every day we tweak individual images to optimize color to produce the final result we desire, not to be "accurate"

I would submit that the all the testing really is a means to standardize upon a set of tools that helps us understand what we need to bring any sensor into a range that we can then adjust for our creative purpose and not to somehow attain some Holy Grail of color fidelity.
Logged

Telecaster

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3686
Re: color accuracy: Kodachrome and Velvia did not succeed due to accuracy
« Reply #29 on: February 09, 2014, 02:41:24 pm »

Every day we tweak individual images to optimize color to produce the final result we desire, not to be "accurate."

I would submit that all the testing really is a means to standardize upon a set of tools that helps us understand what we need to bring any sensor into a range that we can then adjust for our creative purpose and not to somehow attain some Holy Grail of color fidelity.

Well put IMO. The Holy Grail of color fidelity is unattainable in any case, short of an accurate emulation of the entire human eye/brain system, color being an interpretive & subjective thing.

-Dave-
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Getting back to colour rendition…
« Reply #30 on: February 10, 2014, 01:05:24 am »

Hi,

I checked the colour sensitivity curves at DxO and found that the D800E and the Sony A7r are very close. I also compared with the IQ180, the latest Phase One camera DxO has tested.

The colour sensitivity per pixel lags behind on the IQ 180. Having more pixels helps, so normalised for print size the IQ 180 has some advantage at minimum ISO.

If the colour sensitivity is a proper measure for colour accuracy, I would say Sony based CMOS is better than the IQ 180 at the pixel level.

I am not sure how relevant the DxO colour measurements are for real world photography. But the data I have seen doesn't really show that Nikon gave up colour rendition for ISO. The Sony Alpha SLT 99 I have is a special case, as they have a 'translucent' mirror in the optical path, so they loose half an EV on Nikon in ISO-rating.

Best regards
Erik


Hi Erik,

Accurate is mentioned as the possibility to discriminate between metameres. When the sensor cannot discriminate between pure Yellow, and a mix of Red and Green, it cannot be white balanced accurately. It does not mean that very subtle color nuances can be accurately discriminated, that aspect is described in the other metrics mentioned on that DxO page.

Cheers,
Bart
« Last Edit: February 10, 2014, 01:08:28 am by ErikKaffehr »
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: color accuracy: Kodachrome and Velvia did not succeed due to accuracy
« Reply #31 on: February 10, 2014, 01:12:05 am »

Hi,

I would agree on that. What I feel I need is a colour rendition that is credible, so I get a good starting point. Pink or magenta skies are not very credible to me, for instance.

Best regards
Erik


I would submit that the all the testing really is a means to standardize upon a set of tools that helps us understand what we need to bring any sensor into a range that we can then adjust for our creative purpose and not to somehow attain some Holy Grail of color fidelity.
Logged
Erik Kaffehr
 

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: color accuracy: Kodachrome and Velvia did not succeed due to accuracy
« Reply #32 on: February 10, 2014, 01:52:52 am »

Well put IMO. The Holy Grail of color fidelity is unattainable in any case, short of an accurate emulation of the entire human eye/brain system, color being an interpretive & subjective thing.

-Dave-
Something that cannot be fully attained can still be a worthwhile goal.

I want my camera gear to enable "neutral" images. If I want to do creative things with color, I want to do this myself using a color filter, gel my flash or push a slider in Lightroom.

-h
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: color accuracy
« Reply #33 on: February 10, 2014, 06:21:19 am »

But under which light?
My problem with my Canon camera and default Lightroom profiles is that colors appear to be anything but neutral under any light. They appear to have been made with the goal of "wow". I had to make my own profiles in order to come closer to what appears "neutral" to me.

-h
Logged

G*

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 69
Re: Doug Peterson's article gives good insight into development of the IQ250
« Reply #34 on: February 11, 2014, 09:19:28 am »

Niels "The Image (Quality) Professor" Knudsen surely is the most prominent hero in Doug’s text. His color-profiles are highly praised, also those for CMOS cameras that are not PhaseOne-built, i.e. mostly SoCaNikon DSLRs. And as a real hero he’s doing his thing alone, in the basement, with genius-like magic (= no numbers once the basics are set). At least that’s how the story is being told.

My point of view is slightly different. I am really not pleased with CaptureOne’s profiles for the D800E. They have a built-in color-sink towards a brownish red with hardly any discrimination between yellowish-green and reddish-orange. That makes for "healthy", uniform skin tones, but CaptureOne would also be able to perform this stunt on demand with its skin tone color correction tool. With the standard profiles there’s no choice left. So I would like to ask PhaseOne if it would maybe make some sense to double-check their basement-magic with numbers and a second or even third pair of eyes.

Please note: This is no critique towards the person of Niels Knudsen. But maybe towards PhaseOne’s concept of quality control and epic self-praise.
Logged

Doug Peterson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4210
    • http://www.doug-peterson.com
Re: Doug Peterson's article gives good insight into development of the IQ250
« Reply #35 on: February 13, 2014, 01:00:57 pm »

Niels "The Image (Quality) Professor" Knudsen surely is the most prominent hero in Doug’s text. His color-profiles are highly praised, also those for CMOS cameras that are not PhaseOne-built, i.e. mostly SoCaNikon DSLRs. And as a real hero he’s doing his thing alone, in the basement, with genius-like magic (= no numbers once the basics are set). At least that’s how the story is being told.

My point of view is slightly different. I am really not pleased with CaptureOne’s profiles for the D800E. They have a built-in color-sink towards a brownish red with hardly any discrimination between yellowish-green and reddish-orange. That makes for "healthy", uniform skin tones, but CaptureOne would also be able to perform this stunt on demand with its skin tone color correction tool. With the standard profiles there’s no choice left. So I would like to ask PhaseOne if it would maybe make some sense to double-check their basement-magic with numbers and a second or even third pair of eyes.

Please note: This is no critique towards the person of Niels Knudsen. But maybe towards PhaseOne’s concept of quality control and epic self-praise.

You're welcome to create your own ICC profile using any ICC-compliant software.

All I can say is that, in practice, most D800 users I work with are happier with the color in C1 than in LR.

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Doug Peterson's article gives good insight into development of the IQ250
« Reply #36 on: February 13, 2014, 01:20:30 pm »

Hi,

Have you actually tried this?

For one, there have been a vivid discussion here on LuLa, and Esben HR, who says that he is the one in charge of the color management engine in C1 says that it is quite complex like here: http://www.luminous-landscape.com/forum/index.php?topic=82891.msg672308#msg672308

I have also tried build a profile using QPCard software, and it works great with Lightroom but not at all with C1. Well, QPCard says it's ICC-profiles won't work with C1 for the very same reason as Esben says above.

Personally I have worked some with C1 but decided to ditch it, but I have a long preference for LR since I have being using it since early beta days.


Best regards
Erik



You're welcome to create your own ICC profile using any ICC-compliant software.


Logged
Erik Kaffehr
 
Pages: 1 [2]   Go Up