I've been putting geometry through the sh/conversion.sh script like  
wheat through a combine harvester and am now starting to get a good  
sense of our conversion rate.  For the past couple days, the script  
has been chewing through about two hundred real target description  
models and the failure rate is higher than our sample geometry as one  
would expect.

It's still churning and not even 10% done, but so far more than 70000  
objects have been processed.  Of those, there are about 4000  
failures, 63000 successes, and 3000 timeouts when restricting  
conversions to a 5 second time limit.  That implies a worst case  
failure rate of 10.1% and a best case failure rate of 5.7%, which is  
considerably higher than our sample geometry's 2-4% failure rate.   
Moreover, the failures cascade up the model hierarchy meaning you're  
almost guaranteed to have a failure on top-level objects for a given  
model.

The quasi good news is we now have a pretty stable framework for  
tracking progress, preventing regressions, evaluating conversion  
performance, and addressing the most common failure patterns first.   
One such common failure seen is NMG triangulation, which constitutes  
approximately 5-10% of the failures.

If you want to test the script out on some geometry, the usage is  
very simple.  Just run:  sh/conversion.sh help

I'll likely send a note out to the users list with a simplified  
summary of this information later today.

Cheers!
Sean


------------------------------------------------------------------------------
Increase Visibility of Your 3D Game App & Earn a Chance To Win $500!
Tap into the largest installed PC base & get more eyes on your game by
optimizing for Intel(R) Graphics Technology. Get started today with the
Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs.
http://p.sf.net/sfu/intelisp-dev2dev
_______________________________________________
BRL-CAD Developer mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/brlcad-devel

Reply via email to