[MINC-users] xfm2param -scale option does not do anything

[Vladimir S. FONOV]

Hello,

I think it is here:

https://github.com/BIC-MNI/libminc/blob/develop/volume_io/MNI_formats/gen_xfs.c#L65
- if( .... && compute_transform_inverse( ...

so, if compute_transfor_inverse fails  , then identity transform is
allocated instead of supplied one

On Wed, Jul 3, 2013 at 9:08 PM, Andrew Janke <a.janke at gmail.com> wrote:
> Hi all,
>
>> indeed.... param2xfm -scale 0.2 0.2 0.00001 test_scale_2D.xfm works though
>>
>> it might have something to do with the fact that affine transformation with
>> scaling of zero is not invertible.
>
> I'd say Vlad is right but on a quick look I can't see where this
> happens.  My guess is that this is an artifact of the NR code (with
> associated +1 indicies) in minctracc that does the transformations.
> The multiplication of such things is done here:
>
>    https://github.com/BIC-MNI/mni_autoreg/blob/master/minctracc/Numerical/matrix_basics.c#L810
>
> From when the affine matrix is built here:
>
>    https://github.com/BIC-MNI/mni_autoreg/blob/master/minctracc/Numerical/make_rots.c#L236
>
> I would argue that while this makes sense regarding generating an
> invertible xfm for registration purposes it's not the entirely correct
> answer. The resulting transformation should have been just on a plane
> not the identity matrix. Seems that more investigation is needed on
> this one.
>
>
> a
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users



-- 
Best regards,

 Vladimir S. Fonov ~ vladimir <dot> fonov <at> gmail <dot> com