OK, so I'm *finally* starting to understand and appreciate the use of
assumed_gamma 1.0, after all these years of using/"abusing" 2.2. Who says old
dogs can't learn new tricks?! ;-)

Now that POV-Ray v3.7 is here, with its *recommended* use of assumed_gamma 1.0,
one of the problems that previous assumed_gamma 2.2 users (like me) are going to
have will be picking colors that look 'proper' there. (Given that we change our
*bad* behavior and go with the new gamma!) That is, colors that match what we
are used to 'seeing' in gamma 2.2 space.

Typical graphics apps have color-pickers that work in the 'standard' gamma 2.2
space--Photoshop, Painter, etc. But colors visually chosen there don't map *as
seen* into a gamma 1.0 world; they render with the typical 'washed out' look.

The color/shade shift between assumed_gamma 2.2 and 1.0 is the fundamental
reason why I've always resisted using the latter in POV-Ray, and not so much
because of the different *lighting* interaction. (I've only lately been
realizing that the odd appearance actually arises from TWO 'separate'
sources--the color shift and the pure lighting shift--and that I never separated
the two into distinct categories. I'll take it on face value that the *light*
interaction is accurate and correct in v3.7; it's solely the color shift that
concerns me.)

To solve this problem, it seems to me that it would be useful to have a little
stand-alone GUI visual color-picker app that showed its on-screen colors 'bent'
(or rather, UNbent!) into gamma 1.0 linear space, and gave us the
unprocessed/unaltered rgb equivalents. Simply put, it would AFAIU allow us to
easily pick "what we see is what we get" colors for v3.7. In other words, if a
particular gamma 2.2 color we like to use looks 'washed out' in this app, then
we simply visually choose a more pleasing color there. It's a given that the
overall 'range' of colors/shades in the picker will look 'different' from what
we're used to seeing; but that's of no consequence.

I'm aware that v3.7 has new keywords --srgb <.....> for example-- to 'convert'
colors that are chosen in typical 2.2 gamma space to 3.7's assumed_gamma 1.0
environment--and that's a wonderful addition; but I'm thinking that a
color-picker set up for gamma 1.0 might make POV-Ray code writing easier--or
perhaps more intuitive for old-time users-- by not having to deal with the new
srgb(f,t) keywords at all.  (I do realize that my old legacy scenes using
assumed_gamma 2.2 *will* need the srgb substitutions, if I want to easily update
those scenes to v3.7)

In my age-old assumed_gamma 2.2 world, I've used a free little single-purpose
Windows app called colscr32 ("color scroller") to quickly and visually choose a
color for use in POV-Ray. (Much easier than trying to figure out abstract rgb
*numbers* that might work.) It gives me the POV-Ray rgb-equivalent values from
0.0-to-1.0, like <.4,.8,.6>. That's basically the kind of simple app I have in
mind. (It's unfortunate that colscr32 isn't useful in a gamma 1.0 environment.)

I've lately discovered that I can get a gamma 1.0 color-picker in my (old)
version of Photoshop: changing the gamma of its image display to 1.0 (under
FILE/COLOR SETTINGS/RGB SETUP) also changes the shades in the color-picker.
Nice! (I don't know why I never noticed that before--or maybe I thought it was a
bug!!) From what I see, an identical(?) color and shade *can* be chosen
there--relative to gamma 2.2 space--just in a different location in the color
field. So this set-up could be used successfully; but a little stand-alone app
would be more convenient, less resource-intensive--and built 'from the ground
up' to work this way.

I've posted a demonstration image--made for MYSELF mostly, but perhaps others
might find it enlightening as well. (Created with v3.6.2--I haven't yet
downloaded the new v3.7--but the *basic* assumptions are probably the same.) It
uses a 'bumpy' sphere model on purpose; in real life, we hardly ever see a
'plain sphere with a perfectly smooth and diffuse surface.' IMO, the middle
image looks quite good--and will probably look even better in 3.7.

But I have a fundamental question:
Is my Photoshop gamma-change example even a 'technically legitimate/accurate'
way of doing this? It seems that an *almost*-identical color can be chosen re:
gamma-2.2 space, but perhaps not 100-percent accurately. At least, that's what
my eyes tell me, as well as the color-picker numbers: the color match is
extremely close, but not spot-on. (It may be that certain subtle shades of color
are simply missing from this model--OR that I'm just not being careful enough!)

So another question arises: Which of the following two choices would produce the
more accurate/correct color rendition in v3.7 with assumed_gamma 1.0:

1) picking the color from Photoshop's (or colscr32's) 'typical' gamma 2.2 color
space and using srgb <.....> to convert it?
 OR
2) picking the color from a gamma 1.0 space (not necessarily Photoshop's) and
using rgb <......> with no conversion?

In other words, is a gamma-1.0 color-picker even necessary??

I'll admit that there are many POV-Ray images from the past that look absolutely
great and stunning, using assumed_gamma 1.0. In pre-v3.7 days, how did these
artists choose colors that would look good? (That is, 'visually pleasing' colors
that would match what a color-picker in 2.2 gamma space would show.)  Did others
simply use the rather cumbersome power-law trick? I.e., instead of rgb
<.4,.8,.6>, using <pow(.4,2.2),pow(.8,2.2),pow(.6,2.2)>? I'm curious as to what
technique(s) have been so successfully used before now.

In any case, doing a net search for such a specialized gamma 1.0 color-picker
app, I see that it's an old and recurring request among users of other CGI
programs; but nothing seems to be available. Any ideas for the Windows platform?
Preferably FREE, 'lightweight' and single-purpose (not GIMP or some other large
app, in other words.)

Ken

Post a reply to this message

Attachments:
Download 'color_picker_example_ken.jpg' (444 KB)

Preview of image 'color_picker_example_ken.jpg'