Chris Cason <del### [at] deletethistoopovrayorg> wrote:
> On 1/10/2025 02:04, yesbird wrote:

> >
> > All of us know that our lovely oldest but goldest ray tracer has

> > community for a couple of reasons: complicated SDL, lack of dedicated

> >
> > Bill works hard to improve this situation, but the skills and time of

> > would like to assist him all ways I can and have a following proposal.
> >

> > (https://povlab.yesbird.online/all2pov/) now we can try to attract
> > people from most active CG communities, at least:
> > https://www.zbrushcentral.com/
> > https://www.artstation.com/
> > https://blenderartists.org/
> > https://polycount.com/
> >

> > advertising to to POV-Ray there. The main problem I see here is a
> > lack of modern and convenient communication channel - newsgroups are
> > outdated and people don't want to take a step (or more) back.
> > Realtime chats, fast graphics and video exchange is very important when
> > discussing technical graphics details, so if we need to move forward,
> > appropriate service is a must have.
> >
> > You will laugh, but I guess that Discord channel will be the best

> > and satisfies all our needs. We can use it not instead, but as addition

> > answering questions.
> >
> > What do you think about all of the above ?

>
> I have a list of things I want to add to POV once we get 3.8 out of the way (I need
someone to do that as I can't, at
 least not right now). Most of the things in my list touch exactly on the topic of
getting more traction. Primarily, th
e way I see we could get most traction is by having not just the full POV-Ray in
releases (i.e. what we have now), but 
also a libPOV - i.e. POV-Ray as a library, without a parser, specifically intended to
be integrated with third-party ap
plications.
>
> This will provide the means for platform-independent raytracing (whether intending
to really raytrace or just to crea
te simple graphics) in applications via a carefully-defined API implemented in a few
officially-supported major languag
es (at least C++, C# / dotnet and Python - the latter is critical). The API would
allow direct creation of objects with
out using SDL, plus permit calling existing functions such as trace etc. It would also
allow for animations to avoid ha
ving to re-parse the entire source when only moving the camera (or a few objects as
long as bounding is re-calculated).

>
> Apart from making it easy to add sophisticated graphic generation to just about
anything, from a simple python script
 up to a huge C++ application (a good example would be rendering circuit boards in PCB
editors).
>
> Additionally it would make it much easier for existing 3D tools to add POV-Ray
support since rather than having to ge
nerate SDL they can call our API directly.
>
> I have a few libraries that I have squirreled away somewhere that would add value,
such as generic object import; I'l
l have to dig them out and list them here.
>
> As for engagement in terms of talking / forums - I have a dislike of walled gardens
like Discord as the knowledge sha
red there is locked away from search and thus does not benefit the general public. I
would consider endorsing one provi
ded we have a path to extract the chats / threads into a form we can have on our
website, thus allowing it to be indexe
d via search engines.
>
> As for our forums themselves - the entire forum system here on news.povray.org needs
replacing with a much more moder
n alternative. I have a (commercial, but affordable) candidate I would like to use,
but the one thing that is bugging m
e is it would require dropping support for NNTP (unless I want to write a plugin for
it, which I don't). If we went thi
s way basically the current forums would be made available on the new site as a
read-only copy of what's here now, sepa
rate from the live forums. There's a bunch of issues with moving to a new forum (like,
do we make the effort to match u
ser accounts with the archived posts) but overall I consider it essential in the long
term.
>
> I'm interested in hearing from existing NNTP users as to whether the move to a
web-based forum will be an impediment 
to you.
>
> Finally: all the above is a bit of a brain-dump that I came up with in the period of
a few minutes; a much more detai
led reply would take me hours as I would need to look at all my notes and saved links
accumulated over the last few yea
rs. Hence some of what I said here may be not quite right, but the overall meaning is,
I hope, understood.
>
> -- Chris

I love these ideas !
About the slowness of Blender's exporter, my bad for it, but some margin does
exist to make it faster, though indeed having a Python module to just import
into the Blender add-on seems grand! and probably way faster, I would assume, if
it has a way to modularly load data, but my own compiled languages skills are
too limited so far if this requires something else than Python to assemble: my
immediate strategy would just be to load blender structures from numpy array
buffers or subsequent alternatives (some alternatives to numpy can already work
on GPU, but these are not yet cross-hardware compatible. Maybe one universal
enough will appear one day or even numpy itself get the feature). Persistent
data and differential file rendering for animated scenes is indeed a key
feature. Meanwhile I'll keep my effort refactoring POV@Ble but I could use some
more expert devs opinions about it. Even when staying within Python only-realm.

Post a reply to this message