Color management question

Hello,
in RT I am using a GrangerRainbow image to analyse my ICC-profiles created with Argyll. The image has no embedded ICC-profile. However, if I switch in the RT colour management tab between “no profile” (see bottom) and “use embedded profile” (top) I see significant changes in the preview image. How come? I would have thought that without an embedded profile these two settings should produce the same result? The image as displayed with “embedded profile” ist the correct one.

Hermann-Josef

embedded defaults to srgb if there is no profile embedded in the image

1 Like

Thanks for this information.

Still I am puzzled. The “embedded” version is identical to what I see e.g. in PhotoShop, if I open the image there. It is also very similar to what I see in RT if I apply one of the profiles created for my scanner. The version with “none” looks very different and strange, e.g. the narrow cone in the blue and the sharp division between green and yellow in the lower part.

Hermann-Josef

My guess: Photoshop takes the RGB values to be in its large working color space, then for display on the monitor maps them into sRGB. RT, Ingo tells us, presumes the RGB values are sRGB and displays them unaltered.

Thanks for your remark.

The version “embedded”, which according to Ingo, is the sRGB representation looks identical to the image I see in imageJ, where no colour management is involved. So my question is, what produces the strange changes in the RT “none” version? This is certainly not the data as contained in the file, since this is what imageJ displays.

Hermann-Josef

Can you zip up and attach the files in question? (Zip them so the forum doesn’t strip color profile info).

Actually, a build of ImageJ might have color management. See http://www.russellcottrell.com/photo/colortransformer2.htm Rather than introducing a complication, PatDavid’s request is the way to pursue this.

If no icc data exist,

  • with “embedded if possible” RT guesses that the encoding gamma is sRGB (or 2.2) and applies the sRGB gamma correction. Most (all) picture viewers do the same …
  • with “no profile” no gamma correction is applied (“no profile” 's main use is with RAWs which are linear i.e. gamma=1 i.e. no gamma).

Looks like your picture is gamma encoded.

Thanks a lot to all the replies.

I have downloaded the GrangerRainbow from here: http://colorremedies.com/realworldcolor/downloads.html. @RTCharles, yes I know about the ColorTransformer. But this is only a plugin to convert images. If one loads an image into imageJ any embedded profile is ignored and the data are displayed as they are.

As I have said, I would like to know, why the image is so drastically changed, if I choose “no profile” in RT’s Colour management tab.

Hermann-Josef

As I have said, it’s wrong guess of gamma encoding used in your files.
Although primaries are absent the RGB values are gamma encoded (probably gamma=2.2).
By “Use embedded” RT guesses gamma sRGB (very close to gamma=2.2) encoding and applies the related gamma correction
By use of “noprofile” option, RT guesses linear encoding (gamma=1) and applies no gamma correction. Pictures which need gamma correction, look over-bright and with low contrast when no gamma correction is applied on them…

The image with “no profile” does not look over-bright. The blue range is shrinked to a narrow cone and a sharp edge appears between green and yellow. This is what puzzles me. This cannot be due to a missing gamma-correction.

What I see in my raw images from the scanner is that images with no gamma-correction appear dark, and not bright.

Hermann-Josef

@Jossie which version of RT are you using? There was a bug fixed in the last few weeks related to this.

@Morgan_Hardwood I am using version 4.2.1291 on a Windows10 64bit system.

Hermann-Josef

4 posts were split to a new topic: Showing color space info in IrfanView

@Jossie try http://rawtherapee.com/releases_head/windows/RawTherapee_WinVista_64_Gtk3_Release_4.2.1375.zip

@Morgan_Hardwood
I have tried the new version and it shows the same behaviour.

@toze3

Be careful with IrfanView and its colour management enabled. If you make any changes to the image (e.g. crop) and save this, IrfanView incorporates the monitor ICC-profile into the output file. So your colours will be damaged!

Hermann-Josef

I split the hijack comments into a new topic, @Jossie if you like you could re-post your reply to toze3 in the new topic.

May I come back to this topic because it is not really solved from my point of view.

I use RT to process images from my scanner. I use the socalled “raw-format” for the scanner data, which only means that they are not gamma-corrected, i.e. linear data. Thus they should appear dark on the screen if an ICC-Profile is embedded, which was derived from gamma-corrected data. So far so good.

Now, if I switch to “no profile” in the colour management tab for such an image the image on the screen becomes bright and very saturated. This I do not understand. What is happening here? In Rawpedia it is stated that for “no profile” setting

including no gamma correction, which means they will look bright.

First of all, why is my linear image getting brighter if I turn off the embedded profile?
Second: I think the statement in RawPedia is not correct in general. An image without gamma-correction will appear dark in the screen not bright.

Hermann-Josef

@Jossie Herman-Josef …
the correct terminology for non-linearly exported data is “encoding” i.e. a sRGB image is “gamma encoded”… “gamma correction” is the inverse proccess which converts gamma encoded data to linear.

Gamma encoding increases the values … so without gamma correction they are displayed brighter.

If you want to display correctly your linear data use a gamma=1 icc profile (like RTsRGBg10.icc) as input profile.

In RT, by using “no profile” not only the gamma is not corrected, but all color management is absent. You see oversaturated colors because of color management absense in combination with the use of ProphotoRGB space as working space … use sRGB for working space and the oversaturation becomes undersaturation … (BTW … both are wrong)

1 Like

@ilias_giarimis

Thank you very much for your reply. Probably I did not ask in the correct way. Let me try again:

The data in my image are linear, i.e. no gamma-encoding and have an embedded profile which was derived from gamma-encoded scans. Such an image looks dark if displayed, both in RT and other software.

My question was, why the image turns into bright, if I select “no profile” in RT. Here are screenshots from RT on an sRGB-monitor (profile set to “neutral”).

The high saturation only is visible on my EIZO with Adobe RGB gamut. So please ignore that for a moment. The main issue for me is to understand, why the image turns bright if I turn off the profile (and colour management altogether) in RT. I would conclude that some gamma-encoding is taking place, since the image without profile looks “rather normal”. But why??

Hermann-Josef