Re: Cooperation and COLLADA

[Henrik Bennetsen]

Cross posted to RealXtend Dev, Kyoryoku Mailing List and Sirikata Dev List

On Wed, Sep 16, 2009 at 12:06 PM, Ryan McDougall <ryan.mc...@realxtend.org> wrote:

I know you guys are balls->wall right now, but I overheard some chat
about your guys' collada plans. That would be good content for
kyoryoku.

Maybe when you've more time we can restart engineer-engineer
cooperation over COLLADA?

We would be thrilled to collaborate with you guys and anyone else who wanted a COLLADA implementation.

As some of you may have noticed we have Mark Barnes working on COLLADA support. This code as with all we do is BSD licensed so there should be plenty of room for collaboration and reuse.

You follow Mark's progress right from Github and he recently posted to this list about his approach and requested comments (sent to the Sirikata dev list which is where Mark reads) which anyone should feel free to pipe in on.

Henrik
 

--
Henrik Bennetsen
Associate Director
Stanford Humanities Lab
Stanford University

Wallenberg Hall, 450 Serra Mall
Building 160, Stanford University
Stanford, CA 94305-2055, USA

benn...@gmail.com
Cell: +1 415.418.4042
Fax: +1 650.725.0192

[Antti Ilomäki]

Apparently Mark net with some of our representatives in the US recently and I answered some questions he had about the realXtend project. Without knowing the exact technical details I think it would make all the sense in the world to colladorate if possible. Mark suggested we take a look at OpenCOLLADA project (opencollada.org), which will apparently be used in Sirikata as well? Anyway, the engineer to engineer discussions sound like a very good idea.

2009/9/16 Henrik Bennetsen <benn...@gmail.com>

[Ryan McDougall]

On Wed, Sep 16, 2009 at 1:29 PM, Henrik Bennetsen <benn...@gmail.com> wrote:
> Cross posted to RealXtend Dev, Kyoryoku Mailing List and Sirikata Dev List
>
> On Wed, Sep 16, 2009 at 12:06 PM, Ryan McDougall
> <ryan.mc...@realxtend.org> wrote:
>>
>> I know you guys are balls->wall right now, but I overheard some chat
>> about your guys' collada plans. That would be good content for
>> kyoryoku.
>>
>> Maybe when you've more time we can restart engineer-engineer
>> cooperation over COLLADA?
>
> We would be thrilled to collaborate with you guys and anyone else who wanted
> a COLLADA implementation.
>
> As some of you may have noticed we have Mark Barnes working on COLLADA
> support. This code as with all we do is BSD licensed so there should be
> plenty of room for collaboration and reuse.
>
> You follow Mark's progress right from Github and he recently posted to this
> list about his approach and requested comments (sent to the Sirikata dev
> list which is where Mark reads) which anyone should feel free to pipe in on.
>
> Henrik
>

When you asked if you could reply publicly I assumed you'd remove any
reference to any testicular anatomy -- but that's ok, no one would
mistake me for someone with any class anyway. ;)

I understand that the code is open and available from github, but my
concern is without two-way participation these kind of initiatives
usually die quietly. That's why I mentioned waiting until the
deadlines subside somewhat, so there is the time/energy to actually
have a conversation.

What I'd like to see is something along the lines of what kyoryoku is
supposed to do, ie. discussions about RFCs, design, and interfaces.

So to start things off, why have you decided to use COLLADA, and in
precisely what roles will it be used? I understand it will be an asset
type, transferred over the network as static binary blob, to replace
raw ogre mesh/materials, but aren't there performance limitations with
converting to and from COLLADA to "native"?

Do you have any overview documentation so we can follow your reasoning and plan?

That said, I understand completely if you guys need to wait until your
deployments are over before you can catch your breath and have this
dialogue.

Cheers,

[Toni Alatalo]

Antti Ilomäki kirjoitti:

> suggested we take a look at OpenCOLLADA project (opencollada.org

> <http://opencollada.org>), which will apparently be used in Sirikata
> as well? Anyway,

Ali told me today that he had tested converting Collada meshes to Ogre
meshes using OgreColladaPlugin at some point earlier. That tool uses the
FCollada library, which apparently is not open source anymore nor even
freely available?, but anyhow it was cool to learn that OgreCollada
easily works as a conversion tool (one of the demos of the lib is such
an app) -- not only as a plugin to Ogre that is (connected to) a
scenemanager which I knew earlier, and hence wasn't interested in this
point 'cause we feed our scene ourselves.

OpenCOLLADA seems fine, the new collada stuff in Blender uses that. It
was made as a Summer of Code project in 2009, now in a branch but on
it's way to trunk - I'll build the branch probably this evening for my
tests. I read the code how it's used in Blender and seemed certainly
nice enough and reportedly works well,
http://dl.getdropbox.com/u/1769373/seymour_blenderender.png and
http://dl.getdropbox.com/u/1769373/bikefromxsi2blender.png are read with
it. From earlier there is a py based AFAIK no lib using Collada suite
for Blender, it worked pretty much ok I think, at least some
Papervision3d authors / flash folks were happy?

I built 'ColladaDotNet' today, taking the Collada Document object from
'Collada for XNA', seemed to be Sony C# code. Tested reading the dining
room example using that, with the idea that could do Collada scene
import within Opensim & ModRex with that. The Collada mesh to Ogre mesh
conversion would be missing there, but as we have a working pipeline for
meshes already from 3d apps, but not for full scenes, I figured testing
with scenes would be interesting. And perhaps if OgreColladaPlugin is
ported to OpenCOLLADA or if the FCollada lib is still usable we can do
conversions of meshes too, we'll see.

Also took a look at PyCollada which is a nice new simple little thing
with a clean API. Might do some little test with that within Naali for
fun, 'cause placing objects there works so could trick a scene sync ..
e.g. load Collada scene locally directly to viewer, for previewing when
have created it in a 3d app and exported. Would kind of be a nice
shortcut now that no server has Collada scenes yet .. it's not there in
Sirikata either (but they currently use csv instead .. and if i
understood correctly, the DotScene Ogre .scene files as another
temporary measure in the pipeline :) This is how scene stuff is in
PyCollada,
http://in4lines.com/in4linescms/documentation/pycollada/collada.scene-module.html
.. it uses ElementTree which is in 2.5 py stdlib and NumPy for numeric
data, which can be Very Cool (I hope we get numpy array support to Naali
at some point so can do wave effects etc in scripts efficiently for
meshes, like a water mesh etc :) .. version 0.1, has scene support,
http://collada.in4lines.com/

The idea of loading some collada binary directly to vertex buffers in
Ogre, and not using the Ogre format at any point between the server and
Ogre rendering, sounded like an interesting idea, and like said on irc,
something that is easily shared by different projects (i.e. a project /
thing on it's own, like the Sirikata folks had planned it).

~Toni

> 2009/9/16 Henrik Bennetsen <benn...@gmail.com
> <mailto:benn...@gmail.com>>

>
> Cross posted to RealXtend Dev, Kyoryoku Mailing List and Sirikata
> Dev List
>
> On Wed, Sep 16, 2009 at 12:06 PM, Ryan McDougall
> <ryan.mc...@realxtend.org

> <mailto:ryan.mc...@realxtend.org>> wrote:
>
> I know you guys are balls->wall right now, but I overheard
> some chat
> about your guys' collada plans. That would be good content for
> kyoryoku.
>
> Maybe when you've more time we can restart engineer-engineer
> cooperation over COLLADA?
>
>
> We would be thrilled to collaborate with you guys and anyone else
> who wanted a COLLADA implementation.
>
> As some of you may have noticed

> <http://groups.google.com/group/open-platform-community/browse_thread/thread/63f4c5d8778131f0>

> we have Mark Barnes working on COLLADA support. This code as with
> all we do is BSD licensed so there should be plenty of room for
> collaboration and reuse.
>
> You follow Mark's progress right from Github

> <http://github.com/mcbarnes/sirikata/commits/master> and he
> recently posted
> <http://groups.google.com/group/platformtalk/browse_thread/thread/36e4c917ef0bf22b/35d53b248de773c0?lnk=gst&q=barnes+collada#35d53b248de773c0>

> to this list about his approach and requested comments (sent to
> the Sirikata dev list

> <http://groups.google.com/group/platformtalk> which is where Mark

> reads) which anyone should feel free to pipe in on.
>
> Henrik
>
>
> --
> Henrik Bennetsen
> Associate Director
> Stanford Humanities Lab
> Stanford University
>
> Wallenberg Hall, 450 Serra Mall
> Building 160, Stanford University
> Stanford, CA 94305-2055, USA
>

> benn...@gmail.com <mailto:benn...@gmail.com>