What level of Numerical Precision does Blender operate on?

Hello Everyone,

I have a question about the basic computation engine of Blender. Does Blender operate on Single or Double precision? If I query a single vertex from a mesh using the Active console I get values that are double precision, but is all the math that is done in Blender done at Double precision? Can I put in a scaling value of 0.999999996723 will I get that scaling or will I get 1 or 0.9999999?

If Blender does work at Double precision then why does the importing routine like Mesh.extend seem to import at single precision? Basically when using the Autocad importing routine supplied in the scripting section, the vertex from the Autocad .dxf file of value:
-0.8660254037844389
becomes in Blender:
-0.866025388241

This is still higher than single precision but the value breaks from the original around the single precision mark (8 digits).

If Blender does work at double precision is there a way to import data in Double precision?

And do the different computations take place at single or double precision when adjusting values of the scaling, translation and rotation?

I hope this question make sense. If it doesn’t please ask for clarification.

Thanks
Bleiner

with normal scale it works with 7 digits
like 9999.001 hope it helps

salutations

You may find some help here.

http://www.rab3d.com/tutorial.html

According to a Dev at IRC #Blendercoders Blender runs on single precision.

Thanks everyone!