The RI Spec as an open standard - Graphics

On May 23, 11:08 am, Michael Ben-Yosef <septa...@mweb.co.za> wrote:
> 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.

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.

> 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?

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 ;-)

> 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 athttps://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.

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...

> 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?

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.

> 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’s APIs are also truly 'open.' The APIs and how they are used,

That's marketing :-)

> Do the authors of Gelato envisage it's API as a possible alternative
> standard to the RI?

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.

On May 26, 3:31 pm, Olivier3...@hotmail.com wrote:
> 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 ;-)

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? :-)

> > Do the authors of Gelato envisage it's API as a possible alternative
> > standard to the RI?
>
> 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.

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

Michael

On May 27, 1:05 pm, Michael Ben-Yosef <septa...@mweb.co.za> wrote:
> > 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.
>
> I find their general purpose GPU offerings quite exciting. We'll see
> what the future holds.

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

On May 23, 8:08 am, Michael Ben-Yosef <septa...@mweb.co.za> wrote:
> 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.

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.


> "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 athttps://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.

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.


> Do the authors of Gelato envisage it's API as a possible alternative
> standard to the RI?

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.

On Jul 1, 8:06 pm, Larry Gritz <goog...@larrygritz.com> wrote:
> On May 23, 8:08 am, Michael Ben-Yosef <septa...@mweb.co.za> wrote:
>
> > 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, for one, had not noticed that.

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

> > Do the authors of Gelato envisage it's API as a possible alternative
> > standard to the RI?
>
> 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.

Understood. Thanks for your reply!

Michael

Michael Ben-Yosef <septagon@mweb.co.za> wrote:
> Well, I guess they updated it on the sly then, so to speak. Not very
> honorable of them.

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/

On Jul 16, 10:06 pm, Rick LaMont <lam...@dotcsw.com> wrote:
> 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.

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

On Jul 16, 7:06 pm, Rick LaMont <lam...@dotcsw.com> wrote:
> 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.

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.