Algorithmic symmetrization (mathematics in Ghosh et al. and tested in Tallman et al.) has been fully implemented in the R package morpho.

Hi Peter,

I suggest you use the R package Morpho (the mantainer of the package is Stefan Schager) to retrodeform deformed specimens. The method is based on the work published by Ghosh et al and Tallman et al.

Here, you can find two applications of this method. We applied the full procedure to retrodeform two human remains: Ceprano (Homo heidelbergensis) and Saccopastore 1 (Homo neanderthalensis) respectively.

Article Retrodeformation of fossil specimens based on 3D bilateral s...

Article Digital reconstruction of the Ceprano calvarium (Italy), and...

if you need help with R please don't hesitate to contact me.

I would very happy to help you!

Best,

Antonio

Thank your for your answers and the literature. Looks good!

I have maybe just the additional problem that my specimens (newt skulls) are very flattend and likely only a 2D-reconstruction is possible to symmetrize them but not really to recover their original 3D-shape.

I will likely come back to you ;)

Hi Peter,

the retrodeformation procedure should be used only when there is a distortion along the midsagittal plane (symmetric plane). If your specimen is flattened dorso-ventrally or anterior-posteriorly there is no chance to retrodeform it.

Best,

Antonio

Hello everybody,

You can use the free software MorphoDig, downloadable from the open-access MorphoMuseum platform (https://morphomuseum.com/morphodig), implemented by scholars (Renaud Lebrun) for scholars!

Antonio provided me a R-script applying retrodeformation using basically the functions 'symmetrize' & 'retroDefromMesh'. But for some reasons the outcome in R-'Morpho' is not as good as in LandmarkEditor. Specimens are still skewed and just wider than the original mesh, while in LandmarkEd. this is fixed. It is even better when ticking the option "Is Articulated" in the "Fossil Fixing Parameters". But for some inexplainable reasons LandmarkEd. crashes during retrodeformation in some specimens, althoug everything is done exactly the same way??? Therefore, it would be good if R would produce similar good results. Maybe someone has any experience in that kind, too?

Hi all, Peter takes into account that there are some parameters to be set... Maybe the issue is the scale, in your specimen if I remember correctly the vertex coordinates are expressed in micron and not in millimeters... In the next days I'll edit the R script :)