juyong/fast_rnrr

Source code for the paper "Quasi-Newton Solver for Robust Non-Rigid Registration" (CVPR2020 Oral).

fast_rnrr's Issues

Error: Some points cannot be covered under the specified radius, please increase the radius

I use this code for face rigistration. when I ran this code, the error occured.
this is the source img

this is the target img

I would be appreciate it if you can tell me how to fix this error.

The format of point cloud to input

It says the code supports non-rigid registration from a triangle mesh to a mesh or a point cloud. Then if I want to input a point cloud, what format is supported in this code? And is there any part of the code has to be modified please? Thank you so much

Which one is the Nonrigid registration transformation matrix,Smat_X?

About test model

hello!it's me again. I checked the test data and noticed that the source and target mesh has the same vertex\face\edge\half edge. so my question is did you guys use some methods to initialize the model? if so ,which method did you use or how you accomplish it?

Because for my design model which has
Vertex:8 face:12 edge: 18 half_edge: 36
and I change it a little bit, the changed model has
Vertex:3645 face:7286 edge: 10929 half_edge: 21858

It dosen't working for my designed model.

numerical issues

I tried your old version code and got error: numerical issues,then the output.obj is the same as src.obj What's wrong?

About dynamic scenes.

Congratulations on your results, we are very interested in incorporating your results into our work.

What we want to know is, does the mesh-pointcloud registration support complex dynamic scenes?

If the source and target used for registration only partially overlap, can accurate registration be accomplished?

Doesn't the first vertex need to be used?

Fast_RNRR/src/tools/nodeSampler.cpp

Line 61 in 9c6e336

VertexNodeIdx[firstVertexIdx] = 0;

Different results from src and src_cvpr????

Thanks for sharing the codes. @Juyong @yaoyx689
I want to register the head of SMPL human model into my 3d scan data for hole filling. The back of a head is hard to scan, and my model has a big hole there. I think non-rigid registration could help. (Do you have any suggestions for it?)

Currently, I am making some tests on synthetic data using Fast_RNRR and NonRigidreg with different methods.
With the same set of source and target data, Fast_RNRR and NonRigidreg with method 3 don't give the same result. Is it caused by a mistake in building or parameters in the scripts????
Fast_RNRR gives a very long straight line. What is this issue?

Unexpected end of file while reading.

hello，thank you for your work!
I made a model myself for testing, but it seems a little problem. I use your test data and runs a perfect result. But use my own design model it suggests:
Custom face Properties defined, before 'vertex_indices' property was defined. They will be skipped.
Custom face Properties defined, before 'vertex_indices' property was defined. They will be skipped.
Unexpected end of file while reading.
Can you tell me how to solve it? Thank you!

Can this tool be used for face registration?

Hi, thanks for your outstanding work. I have a series of 3D face scans and I want to do face registration by fitting a mesh template(like BFM model) on the scans. Can this tool be used for this process? Thanks.

How to get the corr_point

i want to get the corr_point or convert matrix?how can i get that>?

Usage of fixed_vertices

Hi, I'm trying to understand your method and run your code about non-rigid registration, I found a parameter named fixedvex in RegParas but never used. would you mind me asking the usage of the parameter? To fix the mesh boundary or skip finding the correspondence step?

face registration result is very bad

the result in face registration is very bad, like follow, 134 landmarks is used:
source:

target:

nricp result:

fast rnrr result:

How to get the transformation matrix for each point in source surface?

Hello, I want to get the transformation matrix for each point in source surface? what should I do?

The runtime of the function Prune::Initialize()

when I try to use the old version with the SHOT feature and diffusion pruning, the function Prune::Initialize(hgeodesic* geo, VPairs& corres, double c_0) takes about 30s to Initialize. Is this normal?

the "corres_pair_ids_" does not reflect the corresponding relationship between x and y

Fast_RNRR/src/Registration.cpp

Line 129 in b522fd0

rigid_T_ = point_to_point(rig_src_v, rig_tar_v, corres_pair_ids_);

I think the "corres_pair_ids_" does not reflect the corresponding relationship between x and y (rig_src_v, rig_tar_v), or maybe I have a slight misunderstanding?

how to increase radius

i want to know how to increase radius

Hello, how did you get the code of the comparison algorithm you used named RPTS and SVR-L0? I can't find it online, could you share them to me?

weird result in the example data

Hi, @Juyong and @yaoyx689 , thanks for kindly sharing the source code. There are so many brilliant ideas and the results are stunning!

However, when i run the code, there some weried results.
I run this code at the example data "partial overlap" , mesh_0018.obj to mesh_0020.obj, which shown as follows:

1. mesh_0018.obj:
   

2. mesh_0020.obj
   

3. the result
   

4. the provided result
   

I run the code utilize the code under src_cvpr dir, which utilizes the nearest point as the initial correspondence as defaults.
could you tell me why I generate a weird result?
Thanks in advance, your reply will vary appreciated.

关于更新代码的deformation graph node生成相关的问题

hi，大佬，新的代码非常给力，学习到了很多东西，这里关于node生成我有一些问题请教

1. 更新的测地线计算方法对于非闭合mesh生成的node，形变时仍有非常差的结果，尤其是对于人脸数据
2. 如果我使用别的方法生成node会对算法最终结果造成多大影响？比如使用fast_marching

After adjusting the parameters, why the running time of the algorithm is still so long？

Hi, @Juyong, thanks for kindly sharing the source code. However, when I run the new version of Fast_RNRR, the running time of the algorithm is still long after I adjust the parameters. I don't know what caused it.

Undefined symbols error when make

good jobs!

i meet undefined symbols error when make, can you give some suggestion?