[Bf-committers] Re: Re: Bugtracker lost me again....

[joeedh]

Robert Wenzlaff wrote:

>The way I read it, it all comes down to is he prefers his default spheres 
>inscribed instead of curcumscribed.   (He wants his default sphere to fit 
>inside the default cube, not the other way around.)
>
>If you do your example viewed from the top, you will see that the corners of 
>your plane (maximum extent of the shape) align perfectly with the radius of 
>your sphere (also a maximum extent, as is evey point on a sphere).  The same  
>is true of a default cube.  All 8 corners align with the surface of the 
>sphere.   The sizes are chosen such that the point (1,1) is on the radii.  
>Other apps tend to think more in terms of unit bounding boxes (it's probably 
>the more common philosophy).
>
>As far as changing the internals, no need.  If all he is talking about is the 
>default shapes, then the data could simply be changed for the default shape.  
>Any previous .blends already have their size and data recorded, and any calls 
>to Add->Sphere in a newer version would simply have verts added a little 
>closer to the center.  It would not affect spheres that already exist.  It 
>would only become a problem in tutes that used the fact that a default cube 
>fits inside a default sphere (or some other assuption that follows from that 
>fact)...
>

I personally think that this is a useful and important thing.  You never 
need any of this in most renderings, true, but for example for a camera 
matching experiment or something where you have to perfectly replicate 
real objects in 3D it can be fairly useful to know that the default size 
for a cube with size (1,1,1) is 1 x 1 x 1, and that a sphere with size 
(1,1,1) has a radius of 1.  Still, this is more of a functionality 
issue. . .it's not hard to make the change, though, just normalize the 
vectors that construct the UVSphere in addPrinitiveMesh in editmest_add.c

joeedh