Re: [eigen] Rigid transformations in eigen: use of dual quaternions

[ Thread Index | Date Index | More Archives ]

2009/9/13 Hauke Heibel <hauke.heibel@xxxxxxxxxxxxxx>:
> On Sat, Sep 12, 2009 at 8:22 PM, Benoit Jacob <jacob.benoit.1@xxxxxxxxx> wrote:
>> float data[4];
>> reinterpret_cast<Quaternionf*>(data)->slerp(....);
> Why can you not do this?
> float data[4];
> Quaternionf( Map<Vector4f>(data,4) ).slerp(...);
> For me the code seems to vectorize - at least the slerp() part. Even
> when the code would not be vectorizing don't you think it is a bit
> hard-core to pre-emptively optimize via a reinterpret_cast?

You can do this, but what it does is to copy the data into the new
Quaternionf's coeffs.
Rohit was trying to avoid paying for that copying.


Mail converted by MHonArc 2.6.19+