"Cousin Ricky" <rickysttATyahooDOTcom> wrote:
> Mike Horvath <mik### [at] gmailcom> wrote:
> > On 11/22/2015 4:22 PM, Alain wrote:
> >  > You may try to use rgb instead of srgb.
> >  > If the original render used assumed_gamma 1 and uncorrected rgb values,
> >  > then, using srgb would give incorrect results.
> >
> >
> > Well, the colors are supposed to be exact values using Lightsys. I want
> > to get them physically correct.
>
> Alain is correct.  CIE.inc was written years before the srgb keyword was
> introduced to the POV-Ray 3.7 betas, and was designed for proper use without it,
> using assumed_gamma 1.0.  You already use assumed_gamma 1.0 in this draft, as
> well as in v1.0 of this module (although the included illustration is washed out
> for some reason unknown to me).  The srgb keyword was intended for color values
> from outside sources.  By using it on CIE.inc-generated values, you're just
> messing up what was already physically correct.
>

The original image looks a bit washed out to me, too (although I don't have a
true Munsell color wheel to look at.) I have an idea about why that might be so
(and about why the new 3.7 render looks more saturated, but *possibly* still
incorrect) Just a guess: I assume the original render was done in v3.6xx. *If*
it was rendered with an assumed-Gamma of 1.0 instead of 2.2, then the color
values you chose to use may have been 'gamma-lightened' incorrectly there;
whereas the newer render in 3.7, although set up correctly in the SDL with
#version 3.7 and assumed-gamma 1.0, may itself be doing some gamma-fudging
because of the original color values that were used. Hmm, sounds complicated!
;-)

Over and above all this, my own eyes (and CRT monitor) tell me that there's
something not-quite-right about the *greens* in both images; they look a bit
'olive'-colored to me. That's just a subjective assessment, of course, but I
wonder if anyone else sees it that way?

Something else I noticed: There seem to be shadows falling on some of the color
blocks, at least in the *original* image; perhaps that effect is producing a
somewhat false impression of the colors. Maybe a shadowless light would help(?)

Post a reply to this message

Am 23.11.2015 um 11:41 schrieb Mike Horvath:
> On 11/23/2015 4:42 AM, scott wrote:
>> If MapGamut returns linear RGB values (it looks like it does, I couldn't
>> see anything that indicated it does a rgb->srgb conversion) then you
>> must use the rgb keyword in POV as Alain suggested.
>>
>> The srgb keyword is used to supply colours in srgb space, and POV will
>> then "undo" the srgb gamma correction before using the values
>> internally, which is not what you want in this case. Using rgb will
>> force POV to use the values you supply directly.
>>
> 
> This lightened them up a little, but they are still too desaturated.

You probably need to replace the "ambient" with "emission".

The "ambient" keyword was originally intended as a very simple model for
diffuse illumination from other objects: Even with only one light
source, the shadowed sides of objects aren't completely black, as they
are illuminated by light scattered from other objects. The "ambient"
mechanism models this by simply adding a bit of constant brightness to
an object. (Typical "ambient" values for this purpose would be around 0.1.)

On the other hand, people found that "ambient" could also be used to
model glowing objects, i.e. ones that emit light of their own. (For this
purpose, "ambient" would typically be set to 1.0.)

Enter radiosity. This mechanism is intended to model the same thing that
"ambient" was originally designed for -- light scattered from other
objects -- except that it gives far superior results. So on the one hand
"ambient" is obsolete in scenes with radiosity, and up until 3.6 it
would typically be disabled by either defining the materials to not have
any ambient in the first place, or by setting the global "ambient_light"
to zero to allow re-use of materials designed for non-radiosity scenes.
However, on the other hand, the latter would prevent the use of
"ambient" to model glowing objects.

To solve this dilemma, POV-Ray 3.7 introduced the "emission" keyword,
which now takes over the role of "ambient" for glowing objects, while
the use of "ambient" is now limited to its original role as "poor man's
radiosity", by automatically disabling it whenever radiosity is used.

It seems to me that your original scene didn't use radiosity, so the
"ambient" parameter served both roles there; so now that you have added
radiosity, rather than blindly replacing "ambient 1" with "emission 1"
you will probably have to use a lower value; I think "emission 0.5"
gives good results.

Note however that radiosity messes with the apparent colour of objects,
as it adds some coloured illumination from nearby objects; this may be
undesired in an image intended to represent a colour model, so my
personal recommendation would be to disable radiosity in the scene
again. Alternatively, you may want to set "gray_threshold" to 1.0 to
make radiosity illumination completely colourless.

Post a reply to this message

Am 23.11.2015 um 18:24 schrieb Kenneth:

> Over and above all this, my own eyes (and CRT monitor) tell me that there's
> something not-quite-right about the *greens* in both images; they look a bit
> 'olive'-colored to me. That's just a subjective assessment, of course, but I
> wonder if anyone else sees it that way?

Maybe it is due to the colour space being clipped to sRGB in the output
image?

Post a reply to this message

On 11/23/2015 5:34 AM, Cousin Ricky wrote:
> On 2009 November 10, you posted scene file munsell_color_solid_cie.pov to
> p.b.s-f.  That scene renders what I expect are the correct color intensities,
> notwithstanding the unrealistically high ambient value.
>
>
>

The older scene also looks washed-out to me when rendered in 3.7. I 
don't have 3.6 installed at the moment.


Mike

Post a reply to this message

On 11/23/2015 3:00 PM, Mike Horvath wrote:
> On 11/23/2015 5:34 AM, Cousin Ricky wrote:
>> On 2009 November 10, you posted scene file munsell_color_solid_cie.pov to
>> p.b.s-f.  That scene renders what I expect are the correct color
>> intensities,
>> notwithstanding the unrealistically high ambient value.
>>
>>
>>
>
> The older scene also looks washed-out to me when rendered in 3.7. I
> don't have 3.6 installed at the moment.
>
>
> Mike


Okay, I just re-rendered the scene specifying #version 3.6 instead of 
#version 3.7. There is a huge difference! I cannot explain it however. I 
don't know why there is such a big difference.


Mike

Post a reply to this message

Am 23.11.2015 um 21:08 schrieb Mike Horvath:

> Okay, I just re-rendered the scene specifying #version 3.6 instead of
> #version 3.7. There is a huge difference! I cannot explain it however. I
> don't know why there is such a big difference.

Specifying "#version 3.6" will put POV-Ray in backward compatibility
mode; among other thins, this will re-activating the "ambient" mechanism
even when radiosity is used, which might explain the difference.

Post a reply to this message

On 11/23/2015 4:10 PM, clipka wrote:
> Am 23.11.2015 um 21:08 schrieb Mike Horvath:
>
>> Okay, I just re-rendered the scene specifying #version 3.6 instead of
>> #version 3.7. There is a huge difference! I cannot explain it however. I
>> don't know why there is such a big difference.
>
> Specifying "#version 3.6" will put POV-Ray in backward compatibility
> mode; among other thins, this will re-activating the "ambient" mechanism
> even when radiosity is used, which might explain the difference.
>

So, radiosity automatically turns ambient off in 3.7?

Post a reply to this message

"Kenneth" <kdw### [at] gmailcom> wrote:
> The original image looks a bit washed out to me, too (although I don't have a
> true Munsell color wheel to look at.) I have an idea about why that might be so
> (and about why the new 3.7 render looks more saturated, but *possibly* still
> incorrect) Just a guess: I assume the original render was done in v3.6xx. *If*
> it was rendered with an assumed-Gamma of 1.0 instead of 2.2, then the color
> values you chose to use may have been 'gamma-lightened' incorrectly there;
> whereas the newer render in 3.7, although set up correctly in the SDL with
> #version 3.7 and assumed-gamma 1.0, may itself be doing some gamma-fudging
> because of the original color values that were used. Hmm, sounds complicated!
> ;-)

What confuses me is that when I render munsell_1929.pov directly (after making
the correction from cam_view to Muns_cam_view), the result is much more
saturated than the pre-rendered illustration.  I don't see how the pre-rendered
illustration could have come from that scene as published.

File munsell_1929.pov is also missing a #version directive; when I use POV-Ray
3.6.1 or use POV-Ray 3.7 with Version=3.6 on the command line, the colors are
like those of munsell_color_solid_cie.pov that I mentioned above (which, in
retrospect, I probably ran as 3.6).  Without the directive, POV-Ray 3.7 shuts
off the ambient, and the colors become very dark.  Perhaps boosting the diffuse
to 1.0 would lead to desirable results in 3.7.  As for clipka's suggestions, I
would keep the radiosity and use gray_threshold 1.  I haven't done any test
renders yet, though.

To Mike:  If you wish to keep the module compatible to 3.6, you obviously cannot
use emission.  If you want the objects to glow and be compatible with both 3.6
and 3.7, while keeping radiosity, you must put #version 3.6; at the head of the
file, and stick with the high ambient.  If you don't care about 3.6
compatibility, then use emission instead of ambient, and definitely put #version
3.7 at the head of the file.  However, with a higher diffuse value (which I
recommend), you will have less need for emission.

> Over and above all this, my own eyes (and CRT monitor) tell me that there's
> something not-quite-right about the *greens* in both images; they look a bit
> 'olive'-colored to me. That's just a subjective assessment, of course, but I
> wonder if anyone else sees it that way?

When I try to render the data directly, the greens from G5 up look very cyan to
me.  Munsell is supposed to visually space out the hues evenly, but the swath
from G5 to BP5 transitions very slowly to my eyes.

What weirds me out is that Munsell uses the very purplish illuminant C instead
of the normal D65 or D50.  Whether it is used as the color system white point or
the reference white point I have been unable to find out.  (That information is
probably available for purchase.)  Worse, I have been unable to understand what
"reference white point" even means, which certainly hampers my understanding of
color.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Am 23.11.2015 um 21:08 schrieb Mike Horvath:
>
> > Okay, I just re-rendered the scene specifying #version 3.6 instead of
> > #version 3.7. There is a huge difference! I cannot explain it however. I
> > don't know why there is such a big difference.
>
> Specifying "#version 3.6" will put POV-Ray in backward compatibility
> mode; among other thins, this will re-activating the "ambient" mechanism
> even when radiosity is used, which might explain the difference.

I tried it, and that is exactly the case.

Post a reply to this message

Mike Horvath <mik### [at] gmailcom> wrote:
> On 11/23/2015 4:10 PM, clipka wrote:
> > Am 23.11.2015 um 21:08 schrieb Mike Horvath:
> >
> >> Okay, I just re-rendered the scene specifying #version 3.6 instead of
> >> #version 3.7. There is a huge difference! I cannot explain it however. I
> >> don't know why there is such a big difference.
> >
> > Specifying "#version 3.6" will put POV-Ray in backward compatibility
> > mode; among other thins, this will re-activating the "ambient" mechanism
> > even when radiosity is used, which might explain the difference.
> >
>
> So, radiosity automatically turns ambient off in 3.7?

Yes.  This is one of the biggest differences between the two versions.

Post a reply to this message