Artifacts on reconstructed mesh

[mrchalk]

Great work!
Most of the reconstructed mesh surfaces on pirate ship demo are flat and smooth, but I’ve noticed some artifacts—mostly tiny spikes and some pits.
What might be causing these artifacts, and do you have any suggestions for how to fix them?

[Luo-Yihao]

Hi, try closing clamp_anchors and see what happens? But there is still an intrinsic problem when one voxel intersects with a multi-component. We will try to solve it soon

[mrchalk]

Set clamp_anchors to False helps, there is no spikes. What did clamp_anchors do? Will disable clamp_anchors cause other problem?

[mrchalk]

Hi, try closing clamp_anchors and see what happens? But there is still an intrinsic problem when one voxel intersects with a multi-component. We will try to solve it soon

This paper mentioned a problem of dual contouring that "non-manifold geometry generates a undesired point outside of the square" in page 4, figure 5. Is this the multi-component intersection problem you said?
TRELLIS.2 use a regularization term in QEF that encourages computed dual vertex to stay close to the average of the intersecting points, which is inside voxel.

[Luo-Yihao]

There are two issues. 1. Even for manifold cases, the correct solution of QEF will jump out of the box, but still be located on the raw mesh (usually on sharp edges or corners). 2. For multi-component intersections, one solution can never represent the local geometry of more than two parts, where the solution is wrong, and shows the undesired out-of-range.

[mrchalk]

There are two issues. 1. Even for manifold cases, the correct solution of QEF will jump out of the box, but still be located on the raw mesh (usually on sharp edges or corners). 2. For multi-component intersections, one solution can never represent the local geometry of more than two parts, where the solution is wrong, and shows the undesired out-of-range.

Thx, the explanation make sense.

[ydayda]

Set clamp_anchors to False helps, there is no spikes. What did clamp_anchors do? Will disable clamp_anchors cause other problem?

A few holes hanpen if set clamp_anchors=False

[ydayda]

@mrchalk @Luo-Yihao Could you open this issue again, the above problem solved I will get the message! FC convert mesh so fast and without loss! I think it is the most feasible way!

[Luo-Yihao]

@ydayda Hi Sir, have you managed to solve the issue? To be honest, I don’t think there’s an excellent strategy to handle the clamping yet. But a more delicate design is coming soon to address this issue.

[jsnln]

@ydayda @Luo-Yihao I observed the same small holes regardless of whether clamp_anchors is True or False. @Luo-Yihao Do you know what might be causing this? Thanks!

[ydayda]

@ydayda Hi Sir, have you managed to solve the issue? To be honest, I don’t think there’s an excellent strategy to handle the clamping yet. But a more delicate design is coming soon to address this issue.

Not yet, I'm sorry for my bad english. It's much difficult for me to solve it. And I have no idea what happen in it.