Pages: 1 ... 5 6 [7]   Go Down

Author Topic: RawTherapee  (Read 183084 times)

MichaelEzra

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1146
    • https://www.michaelezra.com
Re: RawTherapee
« Reply #120 on: April 25, 2013, 07:16:00 am »

A bit more engaging illustration with an intentionally slightly overexposed input image:
Logged

Hening Bettermann

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 945
    • landshape.net
Re: RawTherapee
« Reply #121 on: January 05, 2015, 04:27:13 pm »

Hi!

I am exploring RawTherapee and have a question. I want ProPhoto as the output profile, but with a linear gamma. RT's own RT_Large_g10 seemed to be just that - but it has the Perceptive rendering intent baked in, and I want Rel Col. So I tried to choose ProPhoto, then Output Gamma: linear_G1.0; or Free Gamma: 1.0. In either case, the Output Profile selection is dimmed. What does this mean? Will the dimmed profile be used? And the dimming only means that it can not be changed any more after the gamma is chosen? 

And in case of the Free Gamma: What is the 'Slope' as opposite to gamma?

Maybe the RawPedia entry 'Color Management Addon' contains the answer to my questions - but it is in French, and I don't understand it.

I posted this question in the RT Forum, but there seems to be little traffic there.

Happy new year!

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1850
    • Frank Disilvestro
Re: RawTherapee
« Reply #122 on: January 05, 2015, 04:50:16 pm »


And in case of the Free Gamma: What is the 'Slope' as opposite to gamma?


The Tone Response Curve (TRC) used in the ICC profiles consist of a linear portion just at the beggining or toe and then a curve expressed as a power function. This is to address an issue at low values where the slope of the curve increases significantly towards infinity. The basic idea is to limit the slope.

For instance, the TRC used for sRGB has a slope (toe) of 12.92 and a power of 2.4. (gamma 2.2 is an approximation)

Another example: Adobe limits (this might change) in their ACE the slope of the curve for Adobe RGB with the function (this is not part of the original Adobe RGB specification):

C= MAX(C' ^ 2.19921875, C'/32) for C' in the range of [0.1]

So the power or gamma is close to 2.2 and the slope is 32

Detailed info here: https://www.adobe.com/digitalimag/pdfs/AdobeRGB1998.pdf

Some converters such as DCRAW (I think RT is using it backstage) allows the user to specify both the slope (toe) and the power

Iliah

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 770
Re: RawTherapee
« Reply #123 on: January 05, 2015, 05:05:53 pm »

> RT's own RT_Large_g10 seemed to be just that - but it has the Perceptive rendering intent baked in

Is it a matrix profile, or a lut-based? If it is a matrix profile, only colorimetric intents are in play. Baking in perceptual intent into a matrix profile has no effect at all currently.
Logged

Hening Bettermann

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 945
    • landshape.net
Re: RawTherapee
« Reply #124 on: January 05, 2015, 06:05:40 pm »

Hi!

Thank you for your fast replies!

Francisco, thank you for your explanation. I understand it 'qualitatively'. Just not sure what is the unit of the slope? Or: which number would indicate a straight line with an angle of 45 degrees? (which would be the continuation of the gamma power, if gamma = 1, if I understand that correctly).

Iliah, I don't know. But if I open the processed TIF in PhotoLine, the intent is shown as Perceptive.

In the meantime, it looks like I found the answer to my first question in the out.pp3 file: It says: WorkingProfile=ProPhoto, OutputProfile=ProPhoto, Gammafree=linear_g1.0

So ProPhoto is used - just not sure if it is because I chose it as the output profile, or because the working profile is used if gamma is chosen separately. But for practical use right now, this makes no difference to me, so I can go on with my workflow.

Thanks again!

Iliah

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 770
Re: RawTherapee
« Reply #125 on: January 05, 2015, 06:34:11 pm »

Dear Hening,

I looked at it ( https://code.google.com/p/rawtherapee/source/browse/rtdata/iccprofiles/output/RT_Large_g10.icc?r=e64205d1366b721669874752981fc5634ef2783c ), it is a matrix ICC v.2 profile, it does indeed has perceptual intent set as preferred, which is not supported for matrix profiles, especially of version 2. Any sane color management will fall back to relative colorimetric in this case. Generally it is considered to be an error to tag a matrix profile with perceptual intent.
Logged

Hening Bettermann

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 945
    • landshape.net
Re: RawTherapee
« Reply #126 on: January 05, 2015, 07:39:12 pm »

Hi again Iliah,
thank you for this clarification! Strange that the RT crew would do this then.

Iliah

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 770
Re: RawTherapee
« Reply #127 on: January 05, 2015, 07:56:37 pm »

Dear Hening,.

Intent in matrix profiles of v.2 does not matter, they may have used a wrong template or something, all other data in profile is OK and the profile should not cause any problems.
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1850
    • Frank Disilvestro
Re: RawTherapee
« Reply #128 on: January 05, 2015, 09:43:18 pm »

Just not sure what is the unit of the slope? Or: which number would indicate a straight line with an angle of 45 degrees? (which would be the continuation of the gamma power, if gamma = 1, if I understand that correctly).


The slope has no unit. it is the ratio between y and x in a cartesian system. A straight line with an angle of 45 degrees will have a slope of 1

The attached graph shows the components of the sRBG TRC, with a linear part (blue, slope 12.92) and the curve with power 2.4 for the first 560 values of a 16 bit representation. The line and the curve intersect at y=0.0405
The second graph shows the resulting curve, so for values of y < 0.04045 the linear part is used and for greater values the curve with power=2.4 is used

Note: y values between 0 and 1

Regards

Hening Bettermann

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 945
    • landshape.net
Re: RawTherapee
« Reply #129 on: January 06, 2015, 06:41:17 am »

Thanks to the both of you for your explanations!

Hening Bettermann

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 945
    • landshape.net
Re: RawTherapee
« Reply #130 on: January 06, 2015, 02:53:11 pm »

It turns out that the Perceptual rendering intent is baked into ProPhoto!

Iliah

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 770
Re: RawTherapee
« Reply #131 on: January 06, 2015, 03:03:25 pm »

Dear Hening,
It is a common mistake.
You may want to read Myth #21 from
http://www.colorwiki.com/wiki/Color_Management_Myths_21-25
Logged

Hening Bettermann

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 945
    • landshape.net
Re: RawTherapee
« Reply #132 on: January 06, 2015, 03:18:38 pm »

Thank you Iliah! The much better so for me, since I want the rel col intent. :-) The mis-labelling is however annoying.

dennbel

  • Newbie
  • *
  • Offline Offline
  • Posts: 29
Re: RawTherapee
« Reply #133 on: January 17, 2015, 07:42:37 am »

Windows version 4.2.73 now out. Just keeps getting better!
Logged

Isaac

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3123
Re: RawTherapee
« Reply #134 on: January 19, 2015, 07:17:30 pm »

Windows version 4.2.73 now out.

Well, Windows 64-bit version 4.2.73 now out.

I hadn't thought I'd make any use of the film simulation profiles, but I do use them to shake-things-up -- particularly ideas about B&W processing.
Logged

dennbel

  • Newbie
  • *
  • Offline Offline
  • Posts: 29
Re: RawTherapee
« Reply #135 on: January 20, 2015, 05:51:43 am »

Oh yes 64 bit version. Now the version 4.2.74 W64 bit version is up.
Logged
Pages: 1 ... 5 6 [7]   Go Up