William F Pokorny <ano### [at] anonymousorg> wrote:
> On 6/14/23 09:27, Chris R wrote:
> > I would expect to see some variation in the color in that region due to light
> > traveling different distances based on the camera angle, but not a complete drop
> > out.
> >
> > Any idea why this might be happening?
>
> My guess is the top of the liquid shape is open - or for some reason
> effectively open as traced.
>
> The interior fade function - unlike media - works correctly only when
> the interior shape is "clean." Meaning there is a clearly defined pair
> of surfaces of the interior shape defining the distance. The second
> surface also provides the starting color(a) ahead of the fade to color
> calculation.
>
> (a) - Some exceptions apply.
>
> Aside: This very likely also why there is no fading to color in front of
> the green axis cylinder.
>
> Both the fade and colored-transparency images posted indicate there is
> no second surface. Why that might be, I don't know.
>
> An interesting experiment would be to replace the liquid shape with a
> sphere. My guess is that will work OK except perhaps where rays are
> nearly tangent to the sphere's surface.
>
> Bill P.
>
> Aside: It looks to me like there might be some other noise in the posted
> images too - or maybe it's just "floaters" compliments of my old eyes.
> :-)

Thanks for looking into this Bill.  The object is actually an isosurface whose
function is a combination of a torus, cylinder, and truncated cone.  The upper
limit of the object is simply "y - height".  Maybe the cone function isn't clean
enough to define the top for fading purposes?

As for the noise, I forgot to mention the liquid contains a CO2 stream rising
from the bottom, so that's probably what you are seeing.  I should have removed
that from the model before posting this to make it clearer.  I tested without
the bubbles and I get the same results.

In any case, I have switched to doing this with media instead of fading, as it
is easier to add in the general foaminess of the beverage that way, and the
media versions work well.  I just found it curious.

-- Chris R.

Post a reply to this message