Position of points on the output grid

[qweqwasdazE]

Will using Poisson reconstruction result in changes in the position of points in the point cloud, making it impossible to match their pose? Because I want to do texture mapping on the output mesh, but there is very little code available in this area, and the code I can find is not very good after use. I don't know if it's for this reason. Is there anyone who understands or can someone provide follow-up programs and executable files? Thank you very much

[mkazhdan]

The output mesh should be in the same coordinate frame as the input points.

[qweqwasdazE]

The output mesh should be in the same coordinate frame as the input points.

So is there a change in the position of a single point? Can we still use the original pose from pixel to vertex in the image?

[mkazhdan]

You should be able to. However, the reconstruction method is not interpolatory, so the vertices in the output are not in one-to-one correspondence with the input samples.

[qweqwasdazE]

You should be able to. However, the reconstruction method is not interpolatory, so the vertices in the output are not in one-to-one correspondence with the input samples.

I attempted to reconstruct images from the DTU dataset. I used patchmatchnet to generate depth maps and fused them into point clouds, and then used Poisson surface reconstruction to obtain a mesh. But after that, I found two texture mapping projects (3DTexture and G2LTex) from github, but I couldn't get normal output. The error display of G2LTex shows that the vast majority of vertices cannot be seen in one view. So I really don't know how to achieve this last step. Can you give me some advice? Thank you very much

[mkazhdan]

I am afraid I am not familiar with these texture-mapping applications.

On April 9, 2023 2:34:37 AM EDT, qweqwasdazE @.***> wrote:

You should be able to. However, the reconstruction method is not interpolatory, so the vertices in the output are not in one-to-one correspondence with the input samples.

I attempted to reconstruct images from the DTU dataset. I used patchmatchnet to generate depth maps and fused them into point clouds, and then used Poisson surface reconstruction to obtain a mesh. But after that, I found two texture mapping projects (3DTexture and G2LTex) from github, but I couldn't get normal output. The error display of G2LTex shows that the vast majority of vertices cannot be seen in one view. So I really don't know how to achieve this last step. Can you give me some advice? Thank you very much

-- Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/255#issuecomment-1501054196 You are receiving this because you commented.

Message ID: @.***>

[qweqwasdazE]

I am afraid I am not familiar with these texture-mapping applications. … On April 9, 2023 2:34:37 AM EDT, qweqwasdazE @.> wrote: > You should be able to. However, the reconstruction method is not interpolatory, so the vertices in the output are not in one-to-one correspondence with the input samples. I attempted to reconstruct images from the DTU dataset. I used patchmatchnet to generate depth maps and fused them into point clouds, and then used Poisson surface reconstruction to obtain a mesh. But after that, I found two texture mapping projects (3DTexture and G2LTex) from github, but I couldn't get normal output. The error display of G2LTex shows that the vast majority of vertices cannot be seen in one view. So I really don't know how to achieve this last step. Can you give me some advice? Thank you very much -- Reply to this email directly or view it on GitHub: #255 (comment) You are receiving this because you commented. Message ID: @.>

Ok，thank you for your recovery，it solved some of my confusion.Anyway, I would like to express my sincere gratitude for your prompt response!