On 2/20/21 11:50 AM, Norbert Kern wrote:
...

Another amazing image. I agree with others on the color comp and that 
vase. I guess as children you were inside while I was outside working on 
the estate grounds :-). Took me a second to recognize the Vermeer - it's 
mirrored...

As I suspect, you suspect, given your code comments, what you are seeing 
is the normal inversion I was talking some about while fixing the 
quilted pattern for povr last fall. Something which happens often enough 
when you play with bump sizes above about 0.5. You have only triangles 
there - one surface - and when the normal inverts the surfaces turns 
inside out(1).

Attached an image. All without the light blocking box. In the top left 
your original scene. In the upper right the bump size set at 0.5. In the 
lower left a v3.7 render showing the existing v3.7 / v3.8 quilted normal 
inverting with a bump size of 5. In the lower right the povr branch's 
fixed quilted normal - which does extra work to prevent the 
primary/major normal direction from inverting - still with a bump size 
of 5.

If you had instead used small boxes with some thickness > the min 
intersection depth the shadow test code would likely pick up the 
"second" box surface correctly and see the intersection at the first 
surface as being in shadow.

(1) - And this touches a POV-Ray secrete with respect to shadows... As 
often as not the surface appears to be in shadow because the normal 
points away from the light source and so it appears unlit when in fact 
POV-Ray considers it lit. The actual shadow detection code usually fails 
to correctly pick up the at surface intersection on the shadow test. ;-)

Aside: My guess is that you needed the extreme normals for radiosity to 
reasonably light the wall texture? If so, an alternative is to lower the 
finish brilliance to something toward 0, so the shading model gives less 
than the usual importance to the incidence of the light rays hitting the 
wall. This is the purpose of brilliance(2).

(2) - I have in my head there is a brilliance setting in the global 
radiosity block with which Christoph was playing in v3.8. I cannot 
remember the state of it at the moment. It was, I think, a boolean - if 
so, probably as to whether to use the textures finish brilliance setting 
if other than 1.0. But..? Not sure what if any implications for 
radiosity and finish brilliance with v3.7.

Bill P.

Post a reply to this message

Attachments:
Download 'nrmlbmpszinvrt.jpg' (54 KB)

Preview of image 'nrmlbmpszinvrt.jpg'