irrlicht + ogre3d mesh format

[Guest]

hi,

a feature suggestion:

why not support ogred3d's mesh format(s) and benefit from their exporters and mesh optimizers and other tools?

would be nice i think.

[elander]

Ogre is in the path of abandoning suport for that stuff in favor of a generic scene format that is also suported by the big guys. It's called COLLADA and FXComposer2 is going to suport it as well as GLSL. Think big not small.

[Guest]

ogre won't give up its formats for collada. collada is an exchange format and not optimized for game engines!

[elander]

Being an interchange format doesn't make it bad on the contrary. Is it missing something ? If FXcomposer2 will use it for advanced shaders i don't see what the problem is. Besides COLLADA can be extended which is much better than build an entire new format from scratch.

[sRc]

ogre won't give up its formats for collada. collada is an exchange format and not optimized for game engines!

o_O? it's being used as the standard format for the PS3...

http://www.gamespot.com/news/2005/07/29 ... 30037.html

[Guest]

"Masami Chatani said that the PS3's development environment will support COLLADA."

the development evironment! ...but ps3 games will surely not directly use plain xml collada files!

collada is an exchange format! i didn't say it's bad and it's nice that irrlicht directly supports it but most engines will still use their own binary formats which can be read much faster and which have other advantages!

[elander]

"Masami Chatani said that the PS3's development environment will support COLLADA."
the development evironment! ...but ps3 games will surely not directly use plain xml collada files!

They haven't said they won't be using COLLADA for their game 3d scene formats either. I think you are interpreting the defenition literaly. When in fact it just means that COLLADA is a complete scene format unlike your common game 3d scene formats that are very incomplete for the most part.

The only limitation being text/xml can be elimininated by creating a binary encoding. Something trivial for a good programmer. And this is only necessary if you need to stream game 3d data.

[buhatkj]

so how about we talk some sense then. If collada is so widely supported, then do all your art in collada, then if you really must have a binary format for your actual in-game art, define that, and write a collada->your format convertor
there are a number of existing binary formats for this, such as md3, md2, or the binary version of .x maybe.
to be honest though, if you are doing things really right, you are loading and caching geometry at startup, so the impact of having slightly slower to load files is only loadup time, and wont affect in-game performace much, which i think is what matters most....
also, you would be surprised how fast some xml parsers are...
I would be interested to see a real comparison of loadup times, maybe render the same geometry in md2 and collada and see how the load times compare in irrlicht...

[Murphy]

Not that I think this is a major concern, but an XML format will also be probably at least four times larger than a binary representation.

[bal]

Not that I think this is a major concern, but an XML format will also be probably at least four times larger than a binary representation.

And x times slower.

[elander]

there are a number of existing binary formats for this, such as md3, md2, or the binary version of .x maybe.

I like this better:

COLLADA + evolving specification + shaders support + 3d tool creators support + major game publishers and creators support + text/xml or binary encoding

[Guest]

Collada isnt a choice in the industry.

We have looked at collada, and what they say is nice, but not realy interesting. And also they saying it for a long time.

The normal way at the moment is, if we want to transfer data we use the fbx format and then convert it into our own formats.

For design time it could be nice to work with only one format, but as we need a special way for every game an open file format isnt what we use.

We have tested collada and the xml hype format, but our result is killing every collada positive idea.

our binary format against collada:
collada size 6 times bigger
collada speed 10 times slower
loading and saving collada 8 times slower

As we need streamed data from cd or harddrive and also handling a huge ammont of models, preloading collada first, compiling it into usable engine format and then using it isnt realy good, as the player needs realy a high end pc for runing our game then.

And we are not the only company that isnt using collada for the same reasons. All other developers i know have decided the same way.

[elander]

Of course that a xml text format cannot compete with a binary format. I already said that, sort of, in my previous post.

Why don't you try creating a binary encoding for collada like i have sugested instead of making yet another 3d format for your game. It doesn't make any sense that everyone doing a game must create its own binary format.

[Guest]

Creating a own format is faster and easyer, than creating a binary encoder and reader for collada.

Why first making a stop at the collada format if you can directly reach the finish?

[elander]

You will have to use professional tools in order to create your game. Most well known tools support collada or plan supporting it: Maya, 3dstudio, Nvidia Fxcomposer, and others. These are the kind of tools professional artists work with. So if you plan to use collada as an interchange format you will have to build at least some software to convert collada for your own format. Then you have to read your own format and setup your game system. So this was a step you would have to take anyway. The other advntages is that it's easier to do a binary encoder and support only collada features your game needs than to create a completely new scene format, with the bonus that it's already documented and most likely another programmer would know how to work with it.

This is all relative of course. The game needs to have at least a certain complexity and ambition or collada won't bring any benefits.