[Bf-committers] Patch for constrain-to-plane

[Bill Baxter]

Martin Poirier wrote:

>--- Bill Baxter <baxter@cs.unc.edu> wrote:
>
>Really nice patch. If you're interested on working on
>the transform function again, I'll be reworking in up
>in the next month, cleaning the functionnality,
>splitting in sub functions for cleaner code and add a
>lot more constraining (in relation to edges, vertices,
>faces, ...). Maybe you can help me.
>
>
>  
>
Yes, perhaps.  Let me know.  I like working on little UI/usability things.

>  
>
>>- I think there should be better smarts in
>>translating mouse motion for 
>>GKEY and SKEY modes in the constrained case. 
>>Currently it just projects 
>>mouse motion onto the constraint axis/plane, but
>>that behaves poorly 
>>when the axis is nearly perpendicular to the screen.
>>  3DSMAX and those 
>>guys have a bit more complicated behavior to handle
>>those cases.
>>    
>>
>
>Trivial fix for edit mode, you just need to insert a
>matrix multiplication before the snapping code call.
>
>  
>
Yeh, but what matrix?  It depends on the situation.   In Maya I think 
(it's been a while since I used it) if you're constraining to a plane 
that's reasonably parallel to the screen then it tries to keep the 
object under the mouse as much as possible (so 1-1 projective 
tracking).  But as the constraint plane goes more and more nearly 
perpendicular to the viewplane, at some point that becomes silly, 
because a tiny movement in X can give you a HUGE motion in Z, eg.  So at 
some point you want to switch to doing 1-to-1 absolute mapping, where 
one unit of x motion gives you 1 unit of z motion.    It's not as 
difficult as remeshing, but still it's not totally trivial to get it to 
feel just right.  

--bb