Maxwell Render | Official public forum

Actually Maxwell is using single precision floats in Geometry ( in RAM ) but in the MXS it depends if:

void setSinglePrecisionOfGeometry( );

method is called. By default MXS saves geometry in double precision, if setSinglePrecisionOfGeometry is called, it saves data in single precision.

Next versions will support maxwell rendering with single and double precision too.


Albert.

Hi Timo,

I'm pretty sure that MAX exports single precision, infact all of it's geometry is single, hence the accuracy problems encountered with either very large or small objects, or models that are a long way from the origin.

Maxwell is supposed to optionally use double precision in later versions, so it does make sense to save the data in this format, too.

Though doubles need some more memory, calculations in double are usually faster (at least on a G5), and single precision is obviously not be enough for complex and extended scenes.

Kabe

Exactly, maxwell accepts double precision for future pourposes. SOme platforms returns geometric data in doubles.