In reply to dave_59:
Thank you Dave. I might have to end up doing matrices upto 12x12 which can cause quite a overhead. I’m thinking of just randomizing it (without a constraint) and doing a check in post randomize to make sure if it’s inverse exists (redo it in case inverse doesn’t exist) in case the constraint solver takes too long.
If you have any suggestions on a implementation, what would be great!
Thanks again!