RenderMan & RISpec, The RI Spec as an open standard

Greetings!

I may be coming dangerously close to asking questions which are
already answered in the RenderMan FAQ, including answers which can be
"read between the lines". I apologize if these things have already
been discussed ad nauseam on this newsgroup, or else if they are best
not discussed publicly.

I have in the past been interested in whether the RenderMan Interface
Spec would be a good standard to adopt for a new renderer or modeller.
Previously I came to the conclusion that it was not, given the lawsuit
by Pixar against Exluna and its founders, and the fact that requests
for renderer licenses were being ignored by Pixar. As the FAQ states:
"Most observers interpret this as a definitive end to any view of RI
as anything but a proprietary API for Pixar's products." Of course,
I'm sure many people disagree with this point of view, given the large
number of free and commercial renderers and modellers that support the
RI in one way or another.

Now I see that at some stage Pixar has updated the RISpec3_2.pdf
document from version 3.2 to version 3.2.1. (I'm sorry if this is very
old news to everyone, but I haven't seen much mention of it.) The
visible difference to me is that the "Statement About Pixar's
Copyright and Trademark Rights" on page 214 now no longer requires
that a no-charge license be obtained from Pixar for anyone who wishes
to write a renderer that uses the RI interface.

My first question is: in the opinion of the readers of this newsgroup,
did this change improve the viability of the RI as a public standard,
signaling a change in Pixar's attitude towards it?

To me it seems that there are still dangers for the (non-PRMan) users
of RenderMan:
1. Pixar asserts numerous times in the RI Spec 3.2.1 (On p. i, in the
Preface on p. x, and on p. 214) that the RI Spec may not be copied.
The statement repeated on pp. i and x is, "No part of this publication
may be reproduced, stored in a retrieval system, or transmitted, in
any form or by any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior permission of Pixar."
Version 3.2.1 appears to have added the sentence (p. 214), "The
foregoing statements of permission expressly do not include permission
to reproduce the specifications for the RenderMan Interface." This
means that any modeller or renderer following the RI Spec cannot
reproduce the standard governing its use other than to link to Pixar's
online copy of the Spec at https://renderman.pixar.com/products/rispec/rispec_pdf/RISpec3_2.pdf
. Should Pixar decide to take that document off its site, a conforming
program is left completely stranded in terms of documentation.
2. Pixar can decide to close up subsequent versions of the RI Spec,
making them available only to licensed PRMan users. Is it the case
that this is already happening with the draft version 3.3, or has
Pixar stated that it will become publicly available?

Have requests been made for Pixar open up the standard a bit more, or
are they not interested in doing so?

Also, have there been any attempts to make a more "open" standard in
the same mould as the RI? In the document "Ten Things About NVIDIA
Gelato that May Surprise and Delight You" (http://www.nvidia.com/
object/gelato_ten_things.html), I see the following statements made:
"Gelato APIs are also truly 'open.' The APIs and how they are used,
the header files, and all the example scenes, shaders, and source code
that we ship with Gelato are covered by the BSD License. With the
exception of the trademarks on the names NVIDIA, Gelato, and Mango,
you are free to use all the header files and examples, copy them,
modify them, redistribute them, and extend them. You can also write,
distribute, or sell the readers and writers and compatible tools,
including renderers."
Do the authors of Gelato envisage it's API as a possible alternative
standard to the RI?

Thanks in advance.

Michael

n May 23, 11:08 am, Michael Ben-Yosef < XXXX@XXXXX.COM > wrote:

For a renderer, I would say it depends a lot on your project. There
are certainly upsides to using it: you get to use a lot of existing
tools and get some guidance as to how things should be done. But it
could also be overly broad for your renderer or become a burden to
support.

For a modeller, supporting a single spec is probably a bad idea. The
way to go is to have a fairly neutral API on top of which plug-ins can
be written for various standards. Even if you're only going to support
RenderMan, it's a good idea to not let the rendering API "crawl" too
deeply in your app.


I don't think use of the RI spec was ever *the* major point of the
lawsuit. Now I'm not a lawyer and I haven't read the actual filings,
but from what I heard and read, it had more to do with Larry leaving
Pixar with lots of knowledge about how PrMan works and starting his
own competing company. There are a lot of trade secrets in a renderer,
many of which are quite simple but hard to obtain, so I could
understand if some people at Pixar got pissed. Now that's just my
reading between the lines so take it for what it's worth ;-)


There are printed versions of the spec, probably a truckload of online
copies, etc. But it's still a valid point. I read that as "don't copy
our documentation for your own software", which brings us to the next
point...


I don't think they have (or ever will) make any statement about that.
And later versions of the "spec" are pretty much the documentation of
PrMan. I don't know if there will ever be an effort to make later
versions public. Pixar certainly has very few good reasons to do that.
Places which buy "the whole pipeline" have little reason to ask for it
either.


That's marketing :-)


I'm sure they did (it is very similar) but right now it doesn't seem
to be getting anywhere. Nvidia, being squeezed from all sides by
integrated video solutions, is probably quite busy enough finding a
future for itself.

Olivier

Thanks very much for your reply, Olivier.



Ah, that was also my impression from trying to make sense of it. On
the other hand, Pixar doesn't exactly seem to promote the independent
use of the RI Spec, even by those who have no insider knowledge of
PRMan. On the third hand, why should they? :-)


I find their general purpose GPU offerings quite exciting. We'll see
what the future holds.

Michael

They are for some people. But here's my take on the future: nvidia's
major problem is that with transistors shrinking and multi-core CPUs
having reached a point of diminuishing returns for the mass market,
the CPU makers are going to look at new ways to spend the transistors.
In fact, hey already have... and one obvious way is integrating a GPU
on the same die as the CPU. Or some very specialized core which can
act like a GPU yet do a lot more. So while nvidia is trying to
generalize their GPUs, the other guys are going to specialize their
cores and eventually both will meet. Then it could get messy for
nvidia if they haven't found new markets for their chips.

Olivier

I, for one, had not noticed that. I'll update the FAQ appropriately.
Note, though, that this doesn't explicitly say anything about the
Shading Language, only the RI procedures and RIB requests. I'm not
sure if that's intentional or not.



I am not a lawyer, but I think the interpretation of this is that they
may not reproduce the RI Spec *document* (the PDF file we're talking
about). You may write your own documentation, in your own words,
describing the spec. Copyright only protects a particular expression,
not the ideas therein.



That was the idea, yes. We weren't actively promoting it to be widely
adopted, but we wanted to be clear that we were tearing down all
possible roadblocks to other people using the APIs that we had
developed, and in particular wanted to repudiate the notion of
documented APIs being used as fodder for litigation.

Well, I guess they updated it on the sly then, so to speak. Not very
honorable of them.


Understood. Thanks for your reply!

Michael

Not on the sly. Appendix I (differences between 3.1 and 3.2) clearly states:

The statement about Pixar's copyrights and trademark rights has been
altered to permit use of the interface by rendering programs without a
written license.

I noticed this years ago. Are you sure it's new in 3.2.1? Seems like
it was earlier than 2005 when they changed it.


Rick LaMont
Dot C Software, Inc.
http://www.dotcsw.com/

I have a copy of 3.2 from june 2003 sitting on my desktop and it does
not have that statement. Instead, its appendix I contains:

Written License for Use of the RenderMan Interface by Renderers
A no-charge license is available from Pixar for anyone who wishes
to write a renderer that
uses the Pixar RenderMan procedure calls or RIB requests. This
license must be in writing.

That statement is also present in all versions of the PrMan manual I
have access to. So it looks like something new from 3.2.1... but I
have no idea how old 3.2.1 is.

Olivier

I am sure because I myself am the author of the 3.2 spec, and it was
issued in 2000, and contained no such statement.

It's a welcome change, in any case.