 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
As a rule, I don't use named colors in computer applications, because I am
convinced that the bureaucrats who assign colors to names are colorblind.
However, it seemed to me early on that something else was going on with the
named colors in colors.inc. Most of them seemed too pale, and many were
entirely the wrong hue.
About 2 years ago, I wrote a scene to plot the stock colors in a 3-D HSL
structure, just to get an idea of how they were distributed. After seeing all
the colors at once, labeled, it occurred to me: What if the colors were
assigned using gamma 2.2 or sRGB? After all, I had gotten the sense that
assumed_gamma was a fairly new feature at the time I discovered POV-Ray, and it
would seem reasonable that the colors were not assigned with linear rendering in
mind.
I have used assumed_gamma 1 in nearly all of my scenes since the day after I
downloaded POV-Ray. Literally. (I checked.) It's just something I don't even
think about. I re-rendered the scene with gamma 2.2, and the color assignments
made a lot more sense.
The other day, I wrote a scene to display the two versions of the stock colors
side-by-side. (Most of the grays are omitted.) The scene uses assumed_gamma 1.
The left half of each test object uses the color straight from colors.inc. On
the right half, the sRGB transfer function is applied.
Having solved this mystery, colors.inc is still not off the hook. Coral, to me,
is not a synonym for orange. The Pink looks more like puce. SlateBlue and
MediumSlateBlue look far brighter than any slate I've witnessed, and I don't
know what "steel blue" even means. (Steel looks a neutral to warm gray to me.)
Worst of all, the color named "Flesh" is /totally/ unrealistic. DarkWood is
much closer to the flesh color over most of my body, with my palms and soles
between MediumWood and Feldspar depending on the computer monitor, the position
of my hand, and the temperature of the room.
P.S. The timing with A.D.B.'s post is pure coincidence.
Post a reply to this message
Attachments:
Download 'linear_v_srgb.png' (595 KB)
Preview of image 'linear_v_srgb.png'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Cousin Ricky" <rickysttATyahooDOTcom> wrote:
> The other day, I wrote a scene to display the two versions of the stock colors
> side-by-side. (Most of the grays are omitted.) The scene uses assumed_gamma 1.
> The left half of each test object uses the color straight from colors.inc. On
> the right half, the sRGB transfer function is applied.
In this second render, the diffuse and ambient values were boosted to the
saturation threshold.
The scene file is posted in povray.binaries.scene-files.
Post a reply to this message
Attachments:
Download 'linear_v_srgb-max.png' (620 KB)
Preview of image 'linear_v_srgb-max.png'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> As a rule, I don't use named colors in computer applications, because I am
> convinced that the bureaucrats who assign colors to names are colorblind.
> However, it seemed to me early on that something else was going on with the
> named colors in colors.inc. Most of them seemed too pale, and many were
> entirely the wrong hue.
>
> About 2 years ago, I wrote a scene to plot the stock colors in a 3-D HSL
> structure, just to get an idea of how they were distributed. After seeing all
> the colors at once, labeled, it occurred to me: What if the colors were
> assigned using gamma 2.2 or sRGB? After all, I had gotten the sense that
> assumed_gamma was a fairly new feature at the time I discovered POV-Ray, and it
> would seem reasonable that the colors were not assigned with linear rendering in
> mind.
>
> I have used assumed_gamma 1 in nearly all of my scenes since the day after I
> downloaded POV-Ray. Literally. (I checked.) It's just something I don't even
> think about. I re-rendered the scene with gamma 2.2, and the color assignments
> made a lot more sense.
>
> The other day, I wrote a scene to display the two versions of the stock colors
> side-by-side. (Most of the grays are omitted.) The scene uses assumed_gamma 1.
> The left half of each test object uses the color straight from colors.inc. On
> the right half, the sRGB transfer function is applied.
>
> Having solved this mystery, colors.inc is still not off the hook. Coral, to me,
> is not a synonym for orange. The Pink looks more like puce. SlateBlue and
> MediumSlateBlue look far brighter than any slate I've witnessed, and I don't
> know what "steel blue" even means. (Steel looks a neutral to warm gray to me.)
> Worst of all, the color named "Flesh" is /totally/ unrealistic. DarkWood is
> much closer to the flesh color over most of my body, with my palms and soles
> between MediumWood and Feldspar depending on the computer monitor, the position
> of my hand, and the temperature of the room.
>
> P.S. The timing with A.D.B.'s post is pure coincidence.
>
Some of my personal gripes:
Maroon is much to red-purple. It's suposed to be a medium dark brown.
The left part of "Very Dark Brown" is closer in shade, but a bit to
dark, to the actual maroon nuts i get from the maroon tree from my backyard.
Salmon is supposed to be a palish shade of pink, but, some parts are darker.
New Midnight Blue is much to light and Midnight Blue looks under
saturated. Using Navy Blue, right part, with a lowered diffuse would be
beter.
For "Flesh", there are so many shades, from almost white to very dark
browns, passing by the olivish, redish and yellowish shades and more.
It's impossible to have one shade and realy be able to call it flesh
without adding region, season, etnicity, darkness/lightness, ...
In my case, it may be somewhere between flesh, new tan and feldspar,
with a small amount of pink added.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Cousin Ricky" <rickysttATyahooDOTcom> wrote:
> I don't know what "steel blue" even means.
Steel can be protected by creating a protective layer of black iron oxide on the
surface in a process called "blueing". It can range from a shiny black to an
amazing "milky" blue in some of the old Smith & Wesson revolvers.
When steel gets heated, it can discolor, and cause a rainbow effect - which is
used to gauge tempering to a certain hardness before quenching.
Perhaps it's hard to see, or perhaps it's the power of suggestion, but the
1-foot blade of my knife (stainless) although "silver" has a very very faint
blue to it - it could be an effect of the rippling of the metal, the polishing,
oils on the surface giving some sort of iridescence, or some actual blue
component to the reflected light.
It's an interesting project that you've started to bite off there.
I seem to recall that "way" back, some person or persons compiled a sort of
color name library or libraries so that you could look up just about anything by
name and there would be an RGB value for it.
As others have pointed out, some colors are derived from the things they impart
their color to. Oranges, salmon, walnut, carmine, etc. I suppose that there
might be an adequate number of images on could search to average a result,
though I can only imagine that there would be issues with lighting and angle and
radiosity and exposure and....
I guess the easiest things to experiment with would be pure chemicals that are
the exact shade of the color. Carmine, mauve, fuchsia, orpiment, ....
Art stores usually have oil paints based on these exact pigments, which is
amazing given that they're based on lead, chromium, mercury, cadmium, etc.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 30.11.2014 00:35, schrieb Cousin Ricky:
> As a rule, I don't use named colors in computer applications, because I am
> convinced that the bureaucrats who assign colors to names are colorblind.
...or are living in a parallel universe called Haute Couture.
-Ive
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Bald Eagle" <cre### [at] netscape net> wrote:
> Art stores usually have oil paints based on these exact pigments, which is
> amazing given that they're based on lead, chromium, mercury, cadmium, etc.
What a toxic array!
Do they still sell lead-based pigments in art stores? I was surprised when I
read an art book in the mid 1970s that recommended white lead primer, but that
book was written in 1955. Modern whites are based on titanium, which is safe
enough to use in toothpaste.
Chromium produces a moss green pigment. This form is trivalent, so your art
studio need not be declared a toxic waste site.
Vermillion is historically a mercury pigment, but I personally have never seen a
"vermillion" with mercury in it. Perhaps the genuine article had been
discontinued by the mid 1970s. (Wikipedia says that mostly only the Chinese
make the stuff nowadays. Not surprising.)
Cadmium pigments range from deep red to vermillion to yellow. I don't know what
chemical process selects the hue. Lately, I've noticed cadmium "hue" paints,
which suggests that "cadmium" paints are using a less toxic substitute nowadays.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 30.11.2014 00:35, schrieb Cousin Ricky:
> As a rule, I don't use named colors in computer applications, because I am
> convinced that the bureaucrats who assign colors to names are colorblind.
Just overly creative I'd say - someone invents "the colours of this
year's season", and they need to have /some/ name.
> About 2 years ago, I wrote a scene to plot the stock colors in a 3-D HSL
> structure, just to get an idea of how they were distributed. After seeing all
> the colors at once, labeled, it occurred to me: What if the colors were
> assigned using gamma 2.2 or sRGB? After all, I had gotten the sense that
> assumed_gamma was a fairly new feature at the time I discovered POV-Ray, and it
> would seem reasonable that the colors were not assigned with linear rendering in
> mind.
That is exactly the case.
Well, to be more precise, those colors were assigned using just whatever
freakin' gamma the author's system happened to exhibit; whether that was
2.2 or 1.8 or whatever is up for speculation.
Good work on this stuff!
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> I guess the easiest things to experiment with would be pure chemicals that are
> the exact shade of the color. Carmine, mauve, fuchsia, orpiment, ....
> Art stores usually have oil paints based on these exact pigments, which is
> amazing given that they're based on lead, chromium, mercury, cadmium, etc.
>
Not to mention some eye makup based on antimonium (blue) and arsenic
(white). Heavily used in classical times. The health hasard been
balanced with the fact that they may prevent one from going totaly and
irrevocably blind.
There is also uranium red. Broadly used in glazing from the 20's to the
50's.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> The other day, I wrote a scene to display the two versions of the stock colors
> side-by-side. (Most of the grays are omitted.) The scene uses assumed_gamma 1.
> The left half of each test object uses the color straight from colors.inc. On
> the right half, the sRGB transfer function is applied.
Good job - I think this highlights the problems with having the
incorrect gamma very nicely. If you do happen to use the "wrong" gamma,
as some people do for artistic purposes, you can often manually "fix"
areas of light and shade by adjusting the scene lighting. However, your
picture shows this very nicely, having the wrong gamma actually gives
you the wrong hue for most colours, and this can't be fixed by messing
about with the brightness of the finish or the lights.
Luckily this has all been fixed in POV now, this would have (and did)
cause several huge threads a couple of years ago :-)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
