On 3/25/2017 4:19 PM, clipka wrote:
> Am 25.03.2017 um 18:36 schrieb Mike Horvath:
>> On 3/25/2017 11:38 AM, clipka wrote:
>>> In this HSL solid, S
>>> is quite clearly measured radially from the vertical axis.
>>>
>>
>> Nope. It is from the center point too.
>
> Just read up on HSL, and you are right: In your spherical HSL diagram,
> the distance from the center of the spehere does indeed correspond to
> the parameter called "Saturation" in the HSL model.
>

https://commons.wikimedia.org/wiki/File:HSLSphere.svg

Is this image easier to understand? It is missing the "donut hole" in 
the middle, however. So all the grays collapse to a point, unfortunately.


> *BUT* that "Saturation" parameter differs significantly from what I'd
> consider the physical definition of saturation, where only monochromatic
> light would be defined as fully saturated, while white light (at any
> brightness, i.e. any shade of grey if you will) would be considered
> fully unsaturated, which is what the "Chroma" parameter in the CIE LCh
> model corresponds to.
>


Do these figures show what you are talking about?

https://en.wikipedia.org/wiki/File:Hsl-hsv_saturation-lightness_slices.svg

It seems "chroma" needs to be converted to "saturation first". Not sure 
the best way to do this, as I haven't thought of it before.

>
> If you want to get anywhere close to your HSL sphere with the CIE LCh
> model, you have to represent the "Chroma" parameter as distance from the
> vertical axis, otherwise you'll get a torus-ish shape instead of a
> sphere-ish one. I also /think/ you'd get closer to your ideal by
> sticking to the fully canonical rendering, with the "Lightness"
> parameter represented as distance from the bottom plane.
>

I'm not concerned so much with the outer shape. The spherical 
parameterization is what's important to me. But "chroma" needs to be 
converted to "saturation" first, as in the above image.


> At any rate, publishing a non-canonical representation of the CIE LCh
> model will certainly do more to confuse people about the colour model
> than to educate them.
>

I labeled all the images as "original research", and nothing links to 
them except a user page.


Mike

Post a reply to this message