Re: Colorimetric Accuracy in the Field
Re: Colorimetric Accuracy in the Field
- Subject: Re: Colorimetric Accuracy in the Field
- From: Robin Myers <email@hidden>
- Date: Wed, 05 Jun 2013 12:41:30 -0700
If you look at an xyY chromaticity diagram, where is the neutral axis? It is a vertical axis extending orthogonally from the xy chromaticity coordinates of the illuminant on the xy plane. So a shorthand way to describe this neutral axis is simply to use the term "illuminant", it is much shorter than using the term "the axis defined by the illuminant's xy chromaticity coordinates projecting from the xy plane".
The same thought can be extended to an XYZ space, but remember that the neutral axis in XYZ space is a diagonal. The closest I can think would be "an axis defined from the tristimulus values of the illuminant to the black point represented by all zero tristimulus values". It is much easier to use the term "illuminant" and assume the reader had sufficient understanding of the issues to understand the brevity.
Since the RGB values of a chart image defines a sampling in RGB space, and when using a color chart the neutral patches are used by color engineers to represent the RGB space's neutral axis, then the chart neutral patches can be envisioned as a sampling of the neutral axis.
Robin Myers
On Jun 5, 2013, at 11:24 AM, Ben Goren wrote:
> On Jun 5, 2013, at 10:59 AM, Robin Myers <email@hidden> wrote:
>
>> In the case of the pigments used in the ColorChecker family, the neutral axis is skewed in XYZ space slightly from that of the illuminant, D50 (the Prophoto RGB white point).
>
> Perhaps you can help me out, here -- I'm trying to make sense of this.
>
> I honestly don't see the relation between the neutral axis of a pigmented chart and an illuminant -- any illuminant.
>
> The neutral axis of a pigmented chart would best be described as those patches with a flat spectral response; those patches which are equally reflective regardless of the wavelength / frequency of the photons bouncing off of it. And a chart doesn't need to actually have any such spectrally flat patches. Indeed, in the real world, no chart every has any truly perfect spectrally flat patches, though many charts have patches that are reasonably close.
>
> Illuminants don't have a neutral axis, but they do have a color to them, expressible as an absolute value in XYZ space.
>
> A *PROFILE* for a particular illuminant is going to have a neutral axis that runs through the XYZ coordinates for that illuminant's color...but every spectrally-flat patch on the chart is, by definition, always going to lie on that axis. (And patches with a hint of color, such as the N9.5 patch on a ColorChecker, will be very close to but not actually on that axis.)
>
> For a profile with a different illuminant, the neutral axis will be different, but the truly-neutral patches will again lie on this new-and-different axis in this new-and-different profile for this new-and-different illuminant, and the almost-neutral patches will will be close to but not actually on the neutral axis.
>
> You can do various transformations to, for example, produce a print that, when viewed in one light appears as if it was being viewed in an entirely different light, but I don't think that's what you're describing.
>
> Perhaps you can help me understand either where my misunderstanding lies or maybe clarify the point you're trying to make?
>
> Thanks,
>
> b&
_______________________________________________
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