I was wondering if anyone has any benchmark info on using linked libraries to create large terrains. I have created a planet for my project The Garden which I have broken down into 226 quadrants, each a simple plane of 1,000,000 quads (2,000,000 tris). Each quadrant is its own blend, linked to the primary blend. Does anyone have any insight on whether organizing these quadrants into groups will provide any sort of efficiency boost? For example, will 113 blends consisting of 2 quadrants be less taxing than 226 individual blends? I have it set up so that the quadrants are only loaded as needed, which seems to work fine so far. Any advice or foreseeable issues you can offer are appreciated.
When you do linking, all scene content and linked scene content will be loaded into the scene. This means it is still loaded in a static way. No this will not be faster in any way (no faster loading, no faster running).
If you want a sort of content streaming, have a look at LibLoad. It allows you to dynamically load and unload content from other blend files. But you need to write your own system (Python controller). It is not done automatically.
Thank you for the insight, I had suspected I would need LibLoad and now it is confirmed. I am still learning but I think I can manage this. I am going through the forum looking for any info on using LibLoad and if anyone has any links to interesting reads or useful info for LibLoad, please share them.
If you want load a blend with LibLoad without blender stuck, use:
This: bge.logic.LibLoad(path, ‘Scene’, async=True)
You need add the: async=True
You are right. You need to separate managing content and managing status.
A) managing content = LibLoad
B) managing status = Save/Load (e.g. via globalDict and SaveActuator/LoadActuator) you can find some solutions in the resource forum. Search for saveload. (even my signature contains a thread for this).
Thanks again, I will make use of the resource found in your signature. This thread is now solved.