[Papervision3D] collada vs. *.3DS file format PERFORMANCE ???

[andysk8er]

I agree that the results are a little strange. You would expect that once the
models are imported, the meshes would be the same. As I said, I will try to
clean up the files and post the source for people to play with. 

I would expect the 3DS model to be a cleaner format though. Collada is still
evolving somewhat as a standard (as evidenced by the fact that some DAE
files "don't work"). I personally think that Collada has some room for
improvement and I'll probably be using 3DS models for the forseeable future.
The file load speed alone is worth it. Even so, the Max3DS parser has some
weird quirks. I may try to compile some tips for how to use it to avoid the
headaches I have had with it.

-Andy



jsremi wrote:
> 
> Yes, the fact that one format is using XML has nothing to do with the
> rendering performance at all. 
> There are number of possible issues that this demonstrate.
> + the two exporters do not export the same data. Even if it looks the same
> the models can have more polygons in one export than the other export
> + the loader are not creating the same data to render. Maybe the 3ds
> loader is forgetting some polygons, or the Collada loader create too many,
> or the materials are not optimal. 
> + the models could be the same but not organized the same way. Maybe the
> ordering of polygons is better from the 3ds loader or exporter. 
> 
> Very interesting results that show there is definitly an issue to be
> investigated. 
> 
> One thing I recommend looking into for papervision is the oportunity to
> write a PV COLLADA conditioner. This would be an application that takes a
> COLLADA file and optimize it specifically for PV usage and rendering,
> producing a new COLLADA file. This would also enable processing files that
> won't work directly with Papervision., for example they contains polygons
> rather than triangles. This would be a better solution than going back to
> exporting the model again with the right options, especially if the
> original model, or the tool used to create it, is not available to the one
> wanting to use it in PV. 
> In addition this will be the place where all the optimization tricks could
> be made available to the PV community, so there will be less learning
> needed to create optimal models, regardless of what tool is used to create
> the models. 
> 
> Regards
> - Remi
> Sent via BlackBerry by AT&T
> 
> -----Original Message-----
> From: "Mark I. Ross" <nospam at therossman.org>
> 
> Date: Wed, 08 Jul 2009 21:00:38 
> To: <papervision3d at osflash.org>
> Subject: Re: [Papervision3D] collada vs. *.3DS file format PERFORMANCE ???
> 
> 
> In all honesty I thought they would have performed the same, once 
> loaded.  The fact that Collada is XML and the 3DS is a binary file 
> format shouldn't really have any bearing on the way the model performs 
> once it is loaded in.  What are the stats saying?  Are there more poly's 
> in the Collada versions of the models, somehow??
> 
> Bizarre.  Anyways, nice work!  Always good to have this sort of 
> investigation going on.
> 
> Cheers,
> mark
> 
> andysk8er wrote:
>>
>> So, I have to give the advantage to the Max3DS parser. Not only are the
>> loaded files smaller, but they perform better. Kind of what I expected.
>> Collada is XML, so it has a lot more overhead. I'll try to clean up the
>> files this weekend and post them on my website for people to play with.
>>
>> -Andy
>>
>>
>>
>>
>>   
>>
> 
> _______________________________________________
> Papervision3D mailing list
> Papervision3D at osflash.org
> http://osflash.org/mailman/listinfo/papervision3d_osflash.org
> _______________________________________________
> Papervision3D mailing list
> Papervision3D at osflash.org
> http://osflash.org/mailman/listinfo/papervision3d_osflash.org
> 
> 


-----
===============================================
Check out my blog for Flash, AS3, and web development stuff:
http://www.wastedpotential.com
-- 
View this message in context: http://www.nabble.com/collada-vs.-*.3DS-file-format-PERFORMANCE-----tp24289928p24409751.html
Sent from the Papervision3D mailing list archive at Nabble.com.