I thought that was only a big coordinate problem.

But it happens everywhere.

I am not sure if i am modelling correct or not.

modelling or report error.3dm (3.2 MB)

I thought that was only a big coordinate problem.

But it happens everywhere.

I am not sure if i am modelling correct or not.

modelling or report error.3dm (3.2 MB)

My guess is the very small discrepency is due to truncation/round off error in conversion between radians and degrees. My understanding is most algorithms used for trig functions use radians, at least internally. Conversion from degrees to radians or radians to degrees always has a truncation/round off error (other than from 0 degrees or 0 radians). In addition errors in converting fractions between base 2 used by computers and base 10 used by (most) humans can introduce errors.

It appears that in at least some bit of code used in Rhino a decision was made that accuracy of angles in degrees of 0.000001 degree was sufficiently accurate. That corresponds to an error of 1 mm in 57 kilometers.

I donâ€™t care about microns. I have read this article many times. It confuse me a bit to tell you the truth. I am not sure when i intersect two lines if the result is an approximation or an absolute operation. I believe its an absolute.

I have also understand that tolerance does not apply to every operation.

I donâ€™t think in my case this is a tolerance problem.

The report tool has a problem. I donâ€™t think when i rotate something 90 degrees i am actually rotate it 90 plus minus tolerance.

Maybe you are right about radians to degree conversion. But thatâ€™s something that should be fixed and it sound easy to me.

Why i use these tolerances? There are many cases here in the forum, especially when we discuss make2d errors that people from mcneel refers that problems occurs due to poor model tolerance. Take a look at this answer i get.

Truly i am in a situation i donâ€™t know how to model properly.

Best practices for absolute tolerance in Rhino:

The absolute tolerance should be an order of magnitude or more than the tolerance needed for the final results in the physical world.

At the beginning the project set the absolute tolerance in the recommended range: 0.01 to 0.0001. Too loose can cause failures in operations. Too tight can cause excessive complexity with the results of some operations, and may be incompatible with some built in tolerances.

If the absolute tolerance needs to be tighter than 0.0001 for physcial world reasons then change to smaller units.

**Do not tighten** the absolute tolerance after starting modeling. It is okay to loosen the tolerance but if it has been loosened do not revert to the original tolerance.

1 Like