Would love some feedback

Hi, I’ve just began doing some research for collada and judging by the information I’ve gathered I’ve come up with the following aproach:

Store the 3d modells and all its data, including specialized gameplay data in a 1 .dae per modell and have references to external textures, shaders, sound etc. This .dae would also contain profiles for diffrent platforms.

Let the game levels be a .dae with game play related custom data which references the external .dae previously mentioned.

One thing which I’m thinking about is the last pipeline step. The most obvious way would be to parse the .daes and save them to a binary format for integrity/speed/size reasons. However, I’m thinking perhaps there might be a better way or if perhaps someone smart already have done a custom tool for this last step.

Grateful for any input!

Have you looked at the COLLADA Refinery on sourceforge? It is an example of a tool to create a data processing pipeline for .dae content that could handle your use case.