HTDerekLiu/intrinsic-simplification

Issue with Coarsening

Opened this issue · 3 comments

Hi, I am running the first Example 00 where I attempt to refine the mesh with 80% compression. I encountered this visualization with orange spots in between, accompanied by text stating "wrong edge length." Could someone please explain what this means exactly, and why there are patches of yellow color?
error

Oh they are two meshes overlapping with each other. The blue one is the original mesh, the yellow one is the simplified mesh with slightly incorrect visualization. For a proper visualization, please refer to 03_visualization.

Thank you, I will check it out. I also have some additional questions regarding this method, as opposed to QEM, which is often considered the go-to compression technique. It would be great if you could provide further insights into the following points:

  • Regarding speed, I understand that QEM tends to be significantly faster. However, is this difference consistently noticeable across all types of meshes, or is it particularly pronounced when dealing with more complex ones?
  • With respect to accuracy, does this new approach ensure near-complete and superior performance compared to established methods such as QEM, currently employed in compression?

As I consider implementing this approach for optimizing game assets, I have a few concerns:

  • Since game engines like Unreal Engine utilize intricate optimization techniques that generally perform well even in complex gaming scenarios, can we expect consistent performance improvements using this novel approach?
  • To what extent does this new method outperform QEM or alternative approaches in terms of processing time? I've heard claims that QEM can reduce polygon counts from 2 million triangles to 30 thousand within just 3.5 seconds. Any relevant data or comparisons would be greatly appreciated.

Hi,

Sorry in advance for the terribly delayed reply. I didn't noticed that the issue was reopened until now.

Re: Speed & Processing time
The best answer I can comment on this is about the implementation. Technically, our method has the some time complexity as QEM. But the way one implement QEM and our method sometimes can make a 100x difference in the speed. We didn't optimize our code to the extreme, so it is likely that the current implementation has room to further improved in order to catch up the speed of well-optimized QEM.

Re: Accuracy
The answer to this question depends on which accuracy you are aiming for. Our method is superior than QEM in terms of accurately preserving intrinsic geometric properties. But QEM will be superior in preserving visual appearance of the mesh.

Re: Unreal
The answer may depends on the type of "improvement" you are aiming for. If the answer is for intrinsic computation, I would say yes, but if for preserving appearances, I would say no.