RE: Gamut mapping newbie question
RE: Gamut mapping newbie question
- Subject: RE: Gamut mapping newbie question
- From: Max Derhak <email@hidden>
- Date: Sun, 03 Apr 2016 02:49:48 +0000
- Thread-topic: Gamut mapping newbie question
Claas,
You bring up a good point. The operations involved can and often do change depending upon the rendering intent. My comments were largely focused on relative rendering (using v4) as it has more predictable results in relationship to gamut mapping. I tend to think of gamut mapping as substituting a "wrong color" for one that you cannot reproduce, and rendering as choosing alternative colors to meet some reproduction objective (like adding contrast or limiting posteriztion of blends for colors going from outside to inside the gamut. This may be an over-simplification that may cause others to disagree with this view of things, but it works for me.
With the perceptual intent the issues of rendering and gamut mapping are blurred as the goal is to get "pleasing" output. This means that things often get moved about for various color reproduction objectives (both inside and outside the gamut). Thus your experience is very real. Even if two profiles have the same Relative gamut, their effective perceptual gamuts can be quite different (with respect to the connecting PCS) with rendering of colors that produce very different results. There is nothing to guarantee an identity relationship when using the perceptual intent.
> Isn’t the mapping „baked“ into the LUTs and applied by the CMM without any analysis of the actual gamuts?
Yes, for the static/dumb CMM the mapping/rendering is baked into the LUTs, and the CMM simply maps output from the Source profile as an input to the Destination profile. This makes using the perceptual intent not as predictable because different profile generation vendors will do different things to get what they feel is "pleasing" output. Using relative intent with black point compensation is more predictable since the relative intent (by definition for v4 profiles) should not perform any non-relative adjustment of colors inside the gamut.
Best regards!
Max Derhak (PhD)
Principal Scientist
-----Original Message-----
From: colorsync-users-bounces+max.derhak=email@hidden [mailto:colorsync-users-bounces+max.derhak=email@hidden] On Behalf Of Claas Bickeböller
Sent: Saturday, April 02, 2016 8:06 AM
To: 'colorsync-users?lists. apple. com' List
Subject: Re: Gamut mapping newbie question
Max,
your answer somehow doesn’t fit to my experience. Maybe you can explain my confusion.
> Gamut mapping is not needed when the resulting PCS color is inside the gamut of the destination profile.
I agree that mapping is not needed if all colours of the source are inside the destination.
But depending on the reproduction aim (rendering intent) a gamut compression is needed (or desired) should some colours be outside. That is e.g. to preserve the overall image „appearance“.
But anyhow even if all colours are inside the destination gamut what I see in real life (especially when using the perceptual intent) is gamut compression.
Take an RGB image (e.g. the image with the girl in front of the magenta background of the roman16 shows it quite well even on my Macbook Pro display) and convert it to CoatedFOGRA39 (the Adobe CS profile).
Then convert it using the perceptual RI to ISO Coated v2 (ECI).
The result you see doesn’t fit to what you wrote here, not?
> Thus, gamut mapping is not performed for PCS colors within the intersection of the source and destination gamuts.
The source and destination device gamut is identical (as both are based on FOGRA39) but anyhow you’ll see a quite huge color shift in my example.
Isn’t the mapping „baked“ into the LUTs and applied by the CMM without any analysis of the actual gamuts?
Thanks
Claas
> Additionally, destination gamut colors outside of the source gamut are not accessed.
>
> For cases utilizing a Smart/Dynamic CMM:
> It is possible for the CMM to perform rendering/re-rendering along with gamut mapping by taking into account both the source and destination gamuts to achieve various reproduction aims. In this case mapping between non-intercecting regions of the gamuts is possible. However, the extent to which this happens is a proprietary aspect of the smart/dynamic CMM implemeneter.
>
> Max Derhak (PhD)
> Principal Scientist
>
>
> -----Original Message-----
> From: colorsync-users-bounces+max.derhak=email@hidden [mailto:colorsync-users-bounces+max.derhak=email@hidden] On Behalf Of Roger Breton
> Sent: Friday, April 01, 2016 10:41 AM
> To: ''colorsync-users?lists. apple. com' List'
> Subject: Gamut mapping newbie question
>
> I am told that, gamut mapping only occurs when the Source and Destination gamuts are of different sizes and shape?
>
>
>
> Best / Roger
>
> _______________________________________________
> Do not post admin requests to the list. They will be ignored.
> Colorsync-users mailing list (email@hidden)
> Help/Unsubscribe/Update your Subscription:
>
> This email sent to email@hidden
>
> _______________________________________________
> Do not post admin requests to the list. They will be ignored.
> Colorsync-users mailing list (email@hidden)
> Help/Unsubscribe/Update your Subscription:
>
> This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden