4708ac11@news.povray.org...

> POV-Ray allows you to do that, not as a side-effect of the SDL, but
> precisely BY DESIGN.  Drop that, and you kill POV-Ray (there are
> already many nice and free pure renderers out there).

Really, this is what should be discussed in depth in those threads, because 
it should drive the development of POV 4.0.

POV-Ray is a unique association: a modelling script + a high-quality 
renderer. The modelling script is intuitive enough to be used by 
non-programmers, while still powerful. That's what made POV-Ray a success at 
a time where there was little competition in terms of free, powerful 3D 
applications. This period more or less peaked in 1997-1998, with the 
integration of radiosity. Now, the times have changed.

- The shortcomings of the script as a modelling tool have become more 
evident: 3D users expect to be able to create certain things - character and 
industrial design, animation, particle systems etc. - that typically require 
a GUI to be done efficiently. Also, including external models into POV-Ray 
isn't straightforward, which makes POV-Ray currently ill-suited as a 
rendering-only application.

Using script (and only script) for modelling is generally a personal 
choice - intellectual and/or aesthetical - and there are niches where it's 
very efficient. Particularly, POV-Ray is relevant in sectors like education, 
scientific illustration, processor benchmarking and as a testbed for 
experimental CG techniques (see http://www.wikipov.org/ow.asp?PovSpotting 
for examples and feel free to add some).

- The shortcomings of POV-Ray as a renderer have also become obvious to 
anyone who browses the galleries of other applications. Just have a look at 
Vray, FinalRender or Maxwell galleries, for instance.

So where do we go from there?

Here are the 3 basic options (which could be mixed/nested of course):

1. Keep POV-Ray as it is, with a relatively accessible SDL, with a few 
additional bells and whistles (better mesh support, better integration of 
3rd party objects, more complete shading language...) and an improved 
rendering engine.

2. Turn POV-Ray into a "pure" rendering library, that could be controlled 
through different programming languages and used by external packages.

3. Turn POV-Ray into a standalone modelling+rendering package (using Moray 
as a starting point for the GUI, for instance), with a deep integration 
between the modeller and the script.

Option 1 is certainly the easiest/cheapest. It's main drawback is that it 
maintains POV-Ray where it is now: a hobbyist toy with some niche 
professional applications. Not a bad thing per se, but not very attractive 
to coders and 3rd party users. This kind of thing needs some momentum and 
I'm not sure that a "hobbyist toy" with a smallish user base has it.

Option 2 is intellectually attractive for coders, and, if done well, could 
promote POV-Ray as a valid open source, free high-quality rendering tool.
Problem 1: there's already some lively competition (Yafray, Indigo, 
Kerkythingie...) in the "pure" renderer field.
Problem 2: will it be possible to keep intact the original simplicity and 
human-writeability of the SDL without excluding the non-programmer users?
Problem 3: the attractiveness (to users and programmers) of the software is 
linked to its attractiveness to artists able to create WOW pictures. Now, a 
language too "programmy" is going to limit those to the smaller population 
of good artists that are also serious programmers.
Of course, if POV-Ray become a 100% pure renderer that only reads SDL 
written automatically or by programmers, the question becomes moot (mosts 
artists will use it through a 3rd party GUI), but something will be lost I 
fear.

Option 3 is clearly a mammoth undertaking, and there's already (since we're 
talking pachyderms) that big elephant in the room of open source 3D that is 
Blender, which is probably already sucking up the free time of many CG 
programmers. Now Blender, thanks to its idiosyncratic interface, is a 
love-it-or-hate-it application, and POV-Ray is all love, so there's a 
possibility there ;)

G.

Post a reply to this message

Shay <Sha### [at] cccc> wrote:
> nemesis wrote:
> > next you'll tell me the pure SDL scene writers out there are
> > not geeks...
>
> ahem
> http://tinyurl.com/ynpfeo

yes, I saw that one before.  Mighty impressive!  and, manual laborer or not,
you're a geek. :)

Post a reply to this message

nemesis wrote:
> 
> next you'll tell me the pure SDL scene writers out there are not geeks...
> like the chemical geek rendering molecules out of matlab and SDL...
> 
You talking to me? Huh? wanna step outside?

Post a reply to this message

Gilles Tran wrote:
> 2. Turn POV-Ray into a "pure" rendering library, <snip> 
> Problem 1: there's already some lively competition (Yafray, Indigo,
> Kerkythingie...) in the "pure" renderer field.

IMO, any renderer with a shader language is not a "pure" renderer. "Drag 
and drop" might be great for some surface shader effects, but will not 
create as many WOW shaders as are possible with programming.

Geometry tweaking is, with some models, as important as shader tweaking. 
I have created nice models which would look like crap with *any* kind of 
algorithmic vertex normal scheme. Yafray, AFAIK, doesn't even allow 
explicit definition of vertex normals!

I'm assuming from your Problem 2:

> Problem 2: will it be possible to keep intact the original simplicity
> and human-writeability of the SDL without excluding the non-programmer
> users?

... that by "Turn POV-Ray into a "pure" rendering library", you in fact 
mean "Turn POV_Ray into a 'pure(ish)' rendering library with a bit of 
accessible tweakability." Doing so would keep POV in a different 
category from the competition.

And as far as Problem 2, yes it will be possible if all potential 
additions to the SDL are very carefully measured by their utility in 
creating WOW pictures.

> Problem 3: the attractiveness (to users and programmers) of the software
> is linked to its attractiveness to artists able to create WOW pictures.
> Now, a language too "programmy" is going to limit those to the smaller
> population of good artists that are also serious programmers.

Exactly. I believe that POV-Ray should be a "pureish" renderer with nice 
Blender or Wings integration and the additional feature of accessible 
tweakability and scene scripting. The hand-coding contingent will figure 
out a way to do what we want to do with whatever the SDL becomes.

  -Shay

Post a reply to this message

andrel <a_l### [at] hotmailcom> wrote:
> > next you'll tell me the pure SDL scene writers out there are not geeks...
> > like the chemical geek rendering molecules out of matlab and SDL...
> >
> You talking to me? Huh? wanna step outside?

come on, confess you're a geek... :)

Post a reply to this message

Shay wrote:
> Gilles Tran wrote:
>> 2. Turn POV-Ray into a "pure" rendering library, <snip> Problem 1: 
>> there's already some lively competition (Yafray, Indigo,
>> Kerkythingie...) in the "pure" renderer field.
> 
> IMO, any renderer with a shader language is not a "pure" renderer. "Drag 
> and drop" might be great for some surface shader effects, but will not 
> create as many WOW shaders as are possible with programming.
> 
> Geometry tweaking is, with some models, as important as shader tweaking. 
> I have created nice models which would look like crap with *any* kind of 
> algorithmic vertex normal scheme. Yafray, AFAIK, doesn't even allow 
> explicit definition of vertex normals!
> 
> I'm assuming from your Problem 2:
> 
>> Problem 2: will it be possible to keep intact the original simplicity
>> and human-writeability of the SDL without excluding the non-programmer
>> users?
> 
> ... that by "Turn POV-Ray into a "pure" rendering library", you in fact 
> mean "Turn POV_Ray into a 'pure(ish)' rendering library with a bit of 
> accessible tweakability." Doing so would keep POV in a different 
> category from the competition.
> 
> And as far as Problem 2, yes it will be possible if all potential 
> additions to the SDL are very carefully measured by their utility in 
> creating WOW pictures.
> 
>> Problem 3: the attractiveness (to users and programmers) of the software
>> is linked to its attractiveness to artists able to create WOW pictures.
>> Now, a language too "programmy" is going to limit those to the smaller
>> population of good artists that are also serious programmers.
> 
> Exactly. I believe that POV-Ray should be a "pureish" renderer with nice 
> Blender or Wings integration and the additional feature of accessible 
> tweakability and scene scripting. The hand-coding contingent will figure 
> out a way to do what we want to do with whatever the SDL becomes.
> 
Not much to add to my fellow-geek here ;)
I only want to stress again that POV as it is now, and even more is it 
will be as POV4 is a fantastic tool to use in highschool math and 
physics classes. Nobody likes geometry (well nearly nobody except for us 
geeks) but if you want to build a chandelier for your virtual barbie 
hous, you do want your two torus sections to match up perfectly. If you 
build a car (or import one from a library) you do want to let the wheels 
behave physically correct when driving and when turning a corner.

As to what my solutions to some of Gilles problems are: Create 
(maintain) a simple POV4 SDL understandable for highschool kids. Add 
features of ordinary languages (assignment (or binding), functions, flow 
control things etc). Add support for multiple camera's, our beloved 
raytracer but also a fast scanline/DirectX/whatever previewer that 
accepts the same scenes but without reflections and such. Make it 
flexible enough to build a third party GUI or enhance Moray. That should 
do it.

So who is going to write that book and interactive course for POV4?

Post a reply to this message

nemesis wrote:
> andrel <a_l### [at] hotmailcom> wrote:
>>> next you'll tell me the pure SDL scene writers out there are not geeks...
>>> like the chemical geek rendering molecules out of matlab and SDL...
>>>
>> You talking to me? Huh? wanna step outside?
> 
> come on, confess you're a geek... :)
> 
It has been a long time since anybody said that to me, but I will ask 
some selection of colleagues tomorrow if I fit that description. :)

FWIW I am not chemical, but a physicist working for nearly 20 years in 
cardiology. None of the patients ever ran away screaming. OK that does 
not prove much, I know.

Post a reply to this message

andrel <a_l### [at] hotmailcom> wrote:
> I only want to stress again that POV as it is now, and even more is it
> will be as POV4 is a fantastic tool to use in highschool math and
> physics classes. Nobody likes geometry

no doubt!  Math is much more fascinating when we can visualize equations as
graphs...

povray has always been to me an excellent tool to teach people both geometry
and introductory computer programming.

> Add support for multiple camera's, our beloved
> raytracer but also a fast scanline/DirectX/whatever previewer that
> accepts the same scenes but without reflections and such.

you do know that there are quality parameter both via the command-line and
from the Render Settings GUI dialog that allow for very fast rendering,
don't you?  Only thing lacking, I guess, would be allowing the mouse to
manipulate that render window by zooming in/out and rotating freely.
Wouldn't that be a cool application for the real-time rendering in povray
3.7? :)

Post a reply to this message

nemesis wrote:
> andrel <a_l### [at] hotmailcom> wrote:
>> I only want to stress again that POV as it is now, and even more is it
>> will be as POV4 is a fantastic tool to use in highschool math and
>> physics classes. Nobody likes geometry
> 
> no doubt!  Math is much more fascinating when we can visualize equations as
> graphs...
> 
> povray has always been to me an excellent tool to teach people both geometry
> and introductory computer programming.
> 
>> Add support for multiple camera's, our beloved
>> raytracer but also a fast scanline/DirectX/whatever previewer that
>> accepts the same scenes but without reflections and such.
> 
> you do know that there are quality parameter both via the command-line and
> from the Render Settings GUI dialog that allow for very fast rendering,
> don't you?  
Yes but still not fast enough to me, unless I missed some.
> Only thing lacking, I guess, would be allowing the mouse to
> manipulate that render window by zooming in/out and rotating freely.
> Wouldn't that be a cool application for the real-time rendering in povray
> 3.7? :)

I don't use the beta ;) .

Post a reply to this message

andrel <a_l### [at] hotmailcom> wrote:
> Yes but still not fast enough to me, unless I missed some.

With level 0 or 1 you don't even get basic shading, just the shapes of
objects in your scene.  Couple that with a small resolution and enough CPU
and you get almost real-time-like rendering speed, even for complex scenes.
 It'll get better if the SDL doesn't have to be reparsed each frame...

> I don't use the beta ;) .

I'm not saying that kind of visualization is planned, but it'd surely be
cool...

Post a reply to this message