Since there have been several attempts to understand the C1 Crush data, I’ll document it here, as far as I’ve worked out.
There are two header sections marked with
// CRUSH DATA. The first might be used only in low-mem mode, as it doesn’t contain any crush positions, so I ignore it in OpenCarma.
Heres what the second header looks like:
Next, a section for each of the 93 vertices (in this example) which are checked for crushing:
Next, a section for each child vertex.
And so on for the other 71 children (in this example)
Hope that finally clears it up :)