wfpokorny

Update on blob (solve_quartic) issues seen with my solver updates plus the changes in this pull req. Found the solve_quartic() tolerance on the roots it finds is somewhat large. In...

I'll be checking the solve_quartic() - Newton-Raphson root polishing change into my solver branch sometime in the next few days. It held up well across all my test cases. In...

Blast it all... I was running through my test cases with the coefficient reduction off for everything today because that is still where I'm aiming to go with the solver...

Yes, Ubunutu 16.04. The image was with 4 threads. Thanks much for the ideas, but woke up this morning with the thought it was likely the shadow cache mechanism -...

Thanks. Yes, might be part of the problem. I've been digging on and off over the weekend and earlier today. No overall conclusions yet, but found and - I think...

OK. The lemon and ovus shadow cache sensitivity is due not filtering near zero roots to a min intersection depth on the torus roots. I believe, but have not verified,...

Update. Resumed working this and the solver updates last week. In my local space I've updated ovus.cpp, lemon.cpp and parametric.cpp to filter

Quick update. I've *not* gotten to looking at the shadow cache obstacle to generally more accurate returned roots. Got hung up on the polysolve/sturm adjustments for collapsing dual roots with...

Busy in about 15 minutes with real life, but in that time a quick update. On the shadow cache issues both old and with more apparent ones with more accurate...

Happened today to run the benchmark scene against the changes proposed herein, plus the shadow cache fixes and solver updates. Not tested often because it's slow to render and with...