Yeah... I changed everything to double precision and now I don't get the
warning anymore (andit was expected, using ascii to represent the number
has nothing to do with precision). But I think I might still get some
errors in very specific situations (if I make too many calculations, small
errors might sum up).

Ah! Is there a similar way to specify object position? without specifying a
rotation (and using directing vectors instead)?

Post a reply to this message

Cesar wrote:
> (andit was expected, using ascii to represent the number
> has nothing to do with precision).

Be carful: The (decimal) ascii representation is not equal to the binary 
representation (you can't, for example, represent decimal 0.1 as a 
binary floating point number) - see 
http://docs.sun.com/source/806-3568/ncg_goldberg.html

But still, I did not have any "Camera vectors are not perpendicular" 
problems when using double precision to compute them.

> Ah! Is there a similar way to specify object position? without specifying a
> rotation (and using directing vectors instead)?

You can use a transformation matrix directly. If your external 
application uses matrices, that would make sense. Check the POV docs for 
the "matrix" keyword.

Post a reply to this message

> Be carful: The (decimal) ascii representation is not equal to the binary
> representation (you can't, for example, represent decimal 0.1 as a
> binary floating point number) - see
> http://docs.sun.com/source/806-3568/ncg_goldberg.html

But you can get so close it's the same (unless, like I said, you use the
values again and again until it degenerates) :o) It doesn't matter. My
problem is solved and I can only say thanks for the help.

Post a reply to this message