Here’s a graph of the memory usage during export, with my new technique, on a very large model:
It’s now pretty much flat while generating the mesh, but there’s a big spike when it hits Apple’s export code at the end, which is too big for many iPads. I’ll be trying to get that down, but it may be up to Apple to optimize their code.
Here’s the graph of OBJ export:
Because OBJ is such a simple format, I use my own code to export. Notice there’s no big spike. Also, notice how the usage in the end is higher. This actually because Sculptura isn’t being given memory warnings by the OS (in response it unloads everything it can).