> If that is the case, then it is nothing like the MegaPOV patch.
> Highlights are a surface reflection effect, and have nothing to do with
> the eye viewing them. What you describe sounds like post-process
> highlights, which sounds perfectly useless.

You might have misunderstood what I'm attempting.

The eye, as we all know, is lightsensitive. Now, what happens
when you look at a mirror which reflects the sun? The glaring
light won't stay just "on top" of the mirror, which is what happens
with POV-Rays specular highlight, it can only be generated on
the surface itself.
This glaring effect is what I want.
In MegaPOV, it took the rgb values and used their intensity
in order to determine how much "glow" might be added. Problem
with this is, a specular highlight might have rgb <1,1,1>, and a
white paper might have rgb <1,1,1> too. They'd be treated the
same.
What I want is to limit this "glow" to just the specular highlights,
and avoid the overall softening of the image as was (is?) the case
with MegaPOVs post-process glowing. (Haven't used the latest
version, so that might have changed).

So, for now, its just post-process highlights, but I want to
increase their effect across the borders of the objects themselves.

To return to the eye-example: there are certain lightsensitive cells
in your eye (I'm not too firm on the english words for that), which,
when reacting strongly, might get their neighbours to act as well.
Thus, a blinding pointlight might be seen as a large white "wall" of
light.

> Well, that does not depend on any highlight method, or even the
> existance of highlights. The blurring method is the most obvious way of
> doing it, and is actually a fairly accurate approximation of what really
> happens in the eye. You wanted one thing, so you attacked the code and
> did something completely different?

As mentioned above, I wanted to limit the effect to the specular
highlights, so that the effect would only occur where light is really
reflected, not just shining on a white object.

As for the objects: I'm using two triangles per pixel, because of
the possibility that the camera might be angled or such. Although...
The boxes per pixel might be skewed with a matrix... Hm. Well,
it's just been a quick approach. Haven't got the time to fumble
with it, and it doesn't seem to actually have THAT much use... I
do crazy stuff, but when its completely unnecessary...

Thanks for your suggestions though!

-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

Post a reply to this message