Le_Forgeron <jgr### [at] freefr> wrote:
> What if SSLT, if it does not fit finish, was an additional (or two ? does SSLT have
a
> meaning to have a different "value" from outside/inside ?) piece of data ?

Again you're not following discussions in these newsgroups.

I already mentioned that I do think it is most suited to be considered an
"interior" attribute.

Its key parameters - aside from index of refraction - are the "(reduced)
scattering coefficient" and "absorption coefficient", both representing the
probability per unit of distance (in this case normalized to per-mm) that a
photon is scattered or absorbed, respectively, as it travels through the
material.

Note that these parameters are basically equivalent to absorbing / scattering
media density (except that media gives proper results only for low-scattering
substances, whereas SSLT is designed for highly-scattering materials).

Surface events are involved as well, but they play only a secondary role; the
main coefficients, at least, do not describe properties of an interface between
materials, but of one material only.


There might be reason to consider it a pigment property, as in real life the
events involved in subsurface scattering *cause* the pigment color, and the
SSLT mechanism supersedes the classic pigment&diffuse thing. But in my opinion,
this only shows that pigment is at the wrong place, at least from a theoretical
point of view.

An additional point is that pigment might still represent an additional surface
coating, which is too opaque to be worth modelling with an added layer of
SSLT-enabled material.


From your musings, I must assume you have no idea what SSLT actually does. I
recommend reading the original 2001 Siggraph paper "A Practical Model for
Subsurface Light Transport" by Jensen, Marschner, Levoy and Hanrahan.

I also wonder whether you have any idea what a "surface normal" is, that you
think SSLT could fit anywhere near.


> last note: should we keep using sslt, or a proper noun (simpler) could be used ?

I for one think "subsurface" is quite a good keyword to use. The proper
technical term is "Subsurface Light Transport" (SSLT), while some refer to it
as "Subsurface Scattering" (SSS), so "subsurface" fits both.

Also note that I'm not in the mood to discuss the whole smash with you. You may
ask me questions, and I may answer those. And you may make comments, and I may
be inspired by those. But I'm positively *NOT* going to *consult* you on how to
implement SSLT, what SDL statements to use, etc.

Honestly, there are lots of other people on these newsgroups I'd ask first.

I feel treated by you as if I had not much of an idea of what I'm doing, or
where to go from where SSLT currently is - and I know for sure that I do NOT
like to be treated that way. If that is really not the case, then I must ask
you to make more effort to convey your true attitude.

Otherwise, any further discussion about SSLT between the two of us ends right
here.

Post a reply to this message

"clipka" <nomail@nomail> wrote:
> Otherwise, any further discussion about SSLT between the two of us ends right
> here.

I'm sure we all hate to see 'points of order' getting in the way of genuine
creative discussion, and this seems to be one of those moments.

I wouldn't like to see developers' time wrapped up in answering FAQs, answering
points that have been adequately answered in the past, or be sidetracked into
peripheral arguments, and instead have them doing what they enjoy: the creative
stuff that helps future POV-Ray users to create great-looking images.

I was going to suggest that a short statement be composed about the new SSLT
implelementation, but remembered that was the first thing that clipka did:
stating that it was *a first attempt, at alpha quality, in a beta release*. We
also discussed priorities that would enable SSLT to be more useful in general
scenes. I wonder if that information could be better-placed, so that we don't
get into today's situation so easily?

So yes, I would suggest that anyone wanting to pitch in - with key ideological
statements, or converse opinions about the correctness of an agreed approach -
read up and be well-versed in the background material, including having a go at
reading the referenced SigGraph paper, understanding what's happening with light
in real life, and how SSLT approximates it, and also having a play with POV-Ray
to understand the limitations. I would consider any less understanding - at
least when throwing major curveballs - to be a little disrespectful to others
here.

Having said that, I see no harm in people pitching shorter questions that can be
quickly answred, in a way that does not take much time for all parties (who are:
the curious posters, the knowledgeable answerers, and those who plough through
the material in search of something they can contribute towards).

Post a reply to this message

"clipka" <nomail@nomail> wrote:
> Mike Raiford <"m[raiford]!at"@gmail.com> wrote:
> > Attached is a minimal scene that demonstrates the artifact. Essentially
> > where CSG ojects join, there are seams. At first I thought this was due
> > to union, but this occurs even with merge.
> [...]
> This is a basic problem of the algorithm that needs to be sorted out in the POV
> environment: To compute the surface color of a certain object at a certain
> location, it needs to find nearby points on the surface of the very same
> object.
>
> Define "same object" in a CSG setting... tricky.

Does this raise some fundamental design questions about POV-Ray? We have the
original direct lighting model, the radiosity model, and photons, we have
different attributes, like surfaces/finishes (ambient, diffuse, specular),
interiors (media, SSLT, fog), and we have different modelling constructs
(primitives shapes, implicit surfaces, CSG, closed composites).

The problem is that they do not all talk to each other like we would casually
expect, An example relevant to this post would be SSLT and 'media scattering'
currently only working with direct light sources.

Is it worth asking a fundamental design question about standards and interfaces
for the very top-level of (classifications of) POV objects? (surfaces,
interiors, models), so that all compliant features, new and old, will operate
with all the others? I appreciate that new features might necessarily introduce
new elements to the interfaces, if they are radical enough.

I'm going to cross-post this to the POV v4 thread, and encourage responses
there, rather than here.

Post a reply to this message

"MessyBlob" <nomail@nomail> wrote:
> > Define "same object" in a CSG setting... tricky.
>
> Does this raise some fundamental design questions about POV-Ray? We have the
> original direct lighting model, the radiosity model, and photons, we have
> different attributes, like surfaces/finishes (ambient, diffuse, specular),
> interiors (media, SSLT, fog), and we have different modelling constructs
> (primitives shapes, implicit surfaces, CSG, closed composites).
>
> The problem is that they do not all talk to each other like we would casually
> expect, An example relevant to this post would be SSLT and 'media scattering'
> currently only working with direct light sources.

As for this particular issue, that's just due to an outright hack that's still
in the SSLT code, which circumvents the standard light source handling stuff
until I've found the time to integrate it properly.

> Is it worth asking a fundamental design question about standards and interfaces
> for the very top-level of (classifications of) POV objects? (surfaces,
> interiors, models), so that all compliant features, new and old, will operate
> with all the others? I appreciate that new features might necessarily introduce
> new elements to the interfaces, if they are radical enough.

The interfaces between the various components are still a bit more fuzzy than
I'd like them to be, but integrating SSLT is a good opportunity to address
these issues, and I'm about to do just that.

Post a reply to this message