POV-Ray : Newsgroups : povray.pov4.discussion.general : Playing with ideas for additional facets methods. : Playing with ideas for additional facets methods. Server Time
2 Mar 2024 21:15:49 EST (-0500)
  Playing with ideas for additional facets methods.  
From: William F Pokorny
Date: 14 Oct 2023 18:26:04
Message: <652b157c@news.povray.org>
In one of the recent threads, Sam mentioned the facets pattern and how 
it replaces normals rather than perturbs them in one of its two modes.

It got me thinking how we might get facets like patterns without the 
need for picking different facet modes depending on the shape's form and 
surface orientation(s).

Attached is an attempt which is perhaps not bad?

It is currently set up by creating regions of constant values a little 
like cells or solid crackle, but at the moment based upon povr's 
f_hash() and a number of other newer inbuilt functions though which of 
those latter might make it into any final code is still somewhat 
undetermined.

After the value regions are set up, the 'accuracy' feature of the 
normal{} block for normal perturbations based upon scalar value patterns 
is used, but with mostly larger than typical settings due the dimensions 
of the value regions. Basically, the pattern works by taking samples 
from different value regions around the base the raw normal vector on 
the surface(*).

(*) The samples would usually come from a scalar value field and not 
discrete values.

Yes, this test code is still normal perturbation and not normal 
replacement, but I think the direction shows promise(a). Image attached 
with a constant, f_hash value regions set up - only changing the 
accuracy settings.

(a) - Thinking the end result will probably be both a scalar value 
pattern and a normal{} block modification method. Maybe calling it 
facets2 or something?

- I've still not played much with the existing facets pattern...

- The larger the accuracy setting the smoother the result, which I 
confess, I don't really understand at the moment. But, there is stuff I 
think not completely right in the way sampling gets done today - and 
there is that magic stuff sitting there in the code too. I need to dig 
into it someday.

Bill P.

Aside: Yep, it might be just using crackle as the value pattern works 
too and twiddling with accuracy works too - I've not tried it. I'm 
trying for something less expensive the crackle(z), but who knows where 
it would all come out cost wise.


(z) - The existing normal{} facets pattern itself uses the crackle 
pattern internally.


Post a reply to this message


Attachments:
Download 'facetplay.jpg' (558 KB)

Preview of image 'facetplay.jpg'
facetplay.jpg


 

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.