Triangle winding incoherent

[neurolabusc]

Your software does a nice job of reducing the triangle count. However, it does not retain the triangle winding. This is revealed with viewers that use this information to determine front and back face. To see this issue, run your program
./main.exe ./dragon.obj ./sim.obj 0.25
and then open the result in MeshLab. Notice that the surface appears wrong. You can fix the problem with MeshLab's Filters/Normals.../ReorientAllFacesCoherently (potentially followed by Filters/Normals../InvertFaceOrientation.

It would be great if your software could retain triangle winding - it really is a nic project.

[lihy96]

Thanks for your advice.
The issue on the triangle winding is ignored in the design of this project. So I just saved the information of vertexes and edges,but left out the information of faces.
Now I am thinking if I can use some data structure(rather than ”set“ as before) to save adjacent points around the vertex so that it can keep the order of adjacent points to maintain the normal vector of the face.
Glad to receive more suggestions from you.

[wonderly321]

我的实验结果和这位仁兄一致，我使用bunny的obj文件感觉简化后感觉表面就像破碎的玻璃，不知道你是否还打算优化个项目。在关于存储邻接点组的信息是，if(i<it*) break 这一句也很令人费解

[GuDuJian-J-Zhang]

The same algorithm I got from vcg works very well