3D mesh manipulation

Oct 22, 2009 at 9:08 AM


Is it possible to transform the mesh of the model, after it is rendered. Suppose I have a mesh of triangles, and I need to manipulate some of the pixels on a specific area. Is that possible?

Thanks in advance.

Best regards,


Oct 22, 2009 at 9:24 AM

Do you want to change pixels within a texturemap or change the model itself?


Oct 22, 2009 at 9:31 AM
Edited Oct 22, 2009 at 9:33 AM

Well, I think I'll have to change the model itself. This mesh will have a texture which should be transformed as I change the model. Look at this


In this example, they construct a 3d model by defining key points on a photo. Then, (I think) it is applied a triangulation and a 3d mesh is generated. For lip sync and expressions they have to change the model itself, right ?

Thanks for the quick reply


Oct 22, 2009 at 10:32 AM

Its mesh modifications you're looking for, I guess. 

I'll get back to you on the subject, just need to figure out the best way for you to do this. 

Oct 22, 2009 at 6:10 PM

I still can't find the answer. Do you have any news ? Thank you very much!

Oct 22, 2009 at 8:08 PM

Sorry about that. I haven't had the time yet to look it over, due to fixing the source repository. 

One way, from the top of my head without looking deeply into it, would be to create your own Mesh type. 
If you look at the Mesh class, you could inherit from it. Through the Mesh class you will have something called a GeometryContext, from that you can manipulate what is known as Vertices, the vertices are the 3D coordinates for the 3D model.

The vertices are not exposed publicly, since they are device dependent - but I'm considering a way of doing this. So if you could live with the GeometryContext till then, it would be great.

Also, look at this thread: http://balder.codeplex.com/Thread/View.aspx?ThreadId=72698

It has some pintpointers to how to create your own mesh type.

Oct 22, 2009 at 8:19 PM

Thank you, thank you very much. I'll try that way.

best regards,




Oct 31, 2009 at 2:32 PM

One thing I have noticed is the disconnection between the position and scale fields in a node from the actual World matrix.  It took me  awhile to understand that not only do I need to set the position or scale, but that I also need to then perform the matrix transform and assign it to the world.  Is this correct or am I missing an easier step?  If this is correct, would it be wise to add the transform into the setter for each field to be transformed?

Oct 31, 2009 at 6:18 PM

This is a convenience thing that has been going in and out of the class, apparently it is left out right now. It should be as you describe, I'll create an issue on GitHub for it - so its not forgotten. Thanks for the input.