RBF scaling questions

Hello everyone,

I have two questions about the RBF scaling process, which are:
(1) I have read the tutorial Lesson 2 of "Pesonalizing your model", and I wonder if I can scale my model just by AnyFunTransform3DLin2 and AnyFunTransform3DRBF, without using the function AnyFunTransform3DSTL. Because at most time, I wouldn't have the stl model of my testing subject.
(2) If the answer to (1) is yes, how many points at least should I used to implement the RBF scaling? Usually I only place 3 or 4 markers on one segment, so can RBF scaling work well in such situation?
Thank you for your attention and hope to get the answers.



  1. Yes. You can use just the 3DRBF version or if you are happy with the linearized scaling - then 3DLin2.

  2. Depends on the bone. The landmarks help to construct interpolation/deformation of the source bone into the patient specific bone. An example in the tutorials uses about 8-15 points for femur. A pelvis would take around 50 to reconstruct it well and capture most of the anatomic features.
    The general "rule of thumb" is to provide landmarks for each anatomical features representing a prominent feature, e.g. lesser trochanter peak point, etc.

Kind regards,

P.S. Please update your profile with name and affiliation - this is a professional forum and we would like to know whom we are helping. Thanks!


Thank you so much for your reply!
And I still have 2 questions: Do I need to define a bounding box for each bone segment, like the image given below, which is raised in the article of Morten Enemark Lund(Scaling of musculoskeletal models from static and dynamic trials,2015)? And if I only have 3 or 4 markers, can I scale a bone segment by the RBF method, even if the result may not be so good as more markers?

Best regards,
Chiho Ng

(BTW: I have update my personal information)