CGAL users discussion list

[SirM2X <>]

Sebastien,

Thank you for your reply. The target position of each control vertex is the
result of only translating the control vertex along some vector.

To demonstrate our goal, I have attached two images below. Given a frontal
view of the cactus (first image), we determine the object's silhouette and
project into 2D. We then deform the 2D silhouette, to the silhouette shown
in the second image, our goal is to deform the mesh in a way that it matches
the deformed silhouette as best as possible. The 3D translation vector of
each control vertex is determined by backprojecting the 2D deformation
vector into 3D space, and intersecting the ray with a plane passing through
the control vertex and oriented towards the camera.

The whole idea is inspired by  this paper
<https://www.cs.ubc.ca/~van/papers/2009-SBIM-contourDrawings.pdf>  .
However, they are using  Mean-Value Geometry Encoding
<https://pdfs.semanticscholar.org/726d/4edc8aef7880e16eb2b98f0ef82657b2bb0e.pdf>
  
for mesh deformation. Since we are heavily using CGAL, we were hoping to be
able to have a quick prototype ready by using CGAL's modules.

<http://cgal-discuss.949826.n4.nabble.com/file/n4662943/cactus1.png> 

<http://cgal-discuss.949826.n4.nabble.com/file/n4662943/cactus-2.png> 



--
View this message in context: 
http://cgal-discuss.949826.n4.nabble.com/Mesh-deformation-produces-strange-results-tp4662924p4662943.html
Sent from the cgal-discuss mailing list archive at Nabble.com.