<div dir="ltr">Hi all,<br><br>I think there is some inconsistency between comments in the code of the composite transform and its (doxygen) documentation.<br>The documentation states that a transformation is composed in the following way:<br>T_0 o T_1 = T_0(T_1(x)), with T_1 the transform added latest.<br><a href="https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Core/Transform/include/itkCompositeTransform.h">https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Core/Transform/include/itkCompositeTransform.h</a><br><br>However, some (very insightful) comments in the code state that:<br>T is composed of T1(p1,x), T2(p2,x) and T3(p3, x) as: T3(p3, T2(p2, T1(p1, x0)))<br><a href="https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Core/Transform/include/itkCompositeTransform.hxx#L620-L659">https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Core/Transform/include/itkCompositeTransform.hxx#L620-L659</a><br><br>The confusing part is that the ordering is reversed (and starting from 1 instead of 0). This may also explain the next comment:<br>// to do: why parameters are listed from N-1 to 1???<br><a href="https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Core/Transform/include/itkCompositeTransform.hxx#L674">https://github.com/InsightSoftwareConsortium/ITK/blob/master/Modules/Core/Transform/include/itkCompositeTransform.hxx#L674</a><br><br>The first definition is consistent with the code as far as I can see. Can anyone confirm this finding? If so, I'm happy to provide a patch via gerrit to solve the comments.<br><br>Best,<br>Floris<br><br><br> <br></div>