Re: what should monochrome option do exactly?

[Daniel J S. <dan...@ie...>]

On 12/14/2011 03:45 PM, Tait wrote:
>> set term foo monochrome (1,0.8,0.5)
>> set term foo monochrome red
>>     [above is equivalent to "set term foo monochrome (1,0,0)"]
>> set term foo monochrome sepia
>>
>> Thoughts?
>
> My first thoughts are, "is that useful?" "is it necessary?"

Don't know.  Could be to a small percentage of people, but if it is an 
easily implemented thing then perhaps it is worth making the feature 
available.  I searched for some examples of sepia tone.  Here are a few:

http://akvis.com/en/coloriage-tutorial/examples/sepia-photo.php
http://www.pxleyes.com/photography-picture/4bb664349c057/Civita.html
http://www.eltonography.com/albums/STANDARD/TCONN.html

My point is that I would call the above monochrome, but not greyscale.


> But if we do go that route, I see no reason to limit the possibilities
> to a linear arithmetic weighted average. If one wants a nonlinear
> dependence on one or more of the color channels, the more general
> solution would be something like "set term foo monochrom using
> ((2.0*$1**2+$2+$3)/(2*255**2+2*255))" or replace 255 with 1.0 if we're
> using a 0-1.0 scale instead of 0-255. I am assuming $1, $2, $3 are red,
> green, blue respectively.

Actually, what you are describing is something different.  I was 
describing the monochrome scale; you are describing the input mapping. 
Let's see if I can draw a little I/O diagram:

                ------------                   ----------
RGB values     | I(R,G,B) |                   |a_r*mono|--> R
    or       -->|   or     |--> mono values -->|a_g*mono|--> G
Mono values    |no mapping|                   |a_b*mono|--> B
                ------------                   ----------

The formula you have given would be I(R,G,B) in the above diagram. 
(a_r,a_g,a_b) are the weighting parameters that I was describing.  And 
for greyscale, a_r=1, a_g=1, a_b=1.  I think, and I may be wrong, that 
currently one could apply the formula you are describing at the point of 
reading data from an image file.  Manipulating a_r, a_g, a_b would be 
possible as well, but with slightly more complex manipulations of input 
data.

Dan