koide3 comments

Results 458 comments of


                                            koide3

Performance issues with smallgicp, while the old fast-gicp version does very well with the same data. PCD is too small warning.

@junwei1999 Can you print out the number of points just before running `example1`? https://github.com/koide3/small_gicp/blob/2c5e9e6092f06bd8b4a0c618c9714e39e4b315dc/src/example/01_basic_registration.cpp#L66

Ubuntu 20.04 --> build error "catkin_make -DCMAKE_BUILD_TYPE=Release"

It seems that there are some issues about the C++ version. Can you try adding `set(CMAKE_CXX_STANDARD 17)` in `CMakeLists.txt`?

Fast-gicp DO NOT

Sorry!! We are reducing the support for fast_gicp. Please consider using [small_gicp](https://github.com/koide3/small_gicp), a new faster library with better portability. Meanwhile, feel free to open a PR if you have any...

Fast-gicp DO

seperate intensity and color vertices?

Sorry for the late response. I think it would be good to add the colormap mode to iridescence. Are you willing to open a PR?

Realsense support / Multi-sensory support

> I'd like to know if GLIM currently supports: > Realsense D435 (depth camera) It depends on if the sensor is equipped with an IMU. GLIM should work for any...

how to adjust the transparency of a texture mesh?

In `TEXTURE_COLOR` mode, alpha values are obtained from the texture data, and material alpha is ignored. You can make textured objects transparent by adjusting the alpha values of texture pixels.

When I build Example Code , a cmake errors occur !!!

Maybe the version of cmake is too old. I recommend upgrading your ubuntu as Ubuntu 18.04 is no longer supported and it is dangerous to keep using it.

2-D points registration

You can customize the scan matching algorithm with [RestrictDoFFactor](https://github.com/koide3/small_gicp/blob/8e665814a9fcbc3b8a3e5bde135917f586abed6e/include/small_gicp/factors/general_factor.hpp#L43) to ignore some dimensions. See [this example](https://github.com/koide3/small_gicp/blob/8e665814a9fcbc3b8a3e5bde135917f586abed6e/src/example/03_registration_template.cpp#L258).

The linearize function in plane_icp_factor.hpp, the Jacobian matrix should be a 1 by 6 matrix.

I think your 1x6 Jacobian corresponds to the error function `(err.x + err.y + err.z)^2` while the 4x6 Jacobian corresponds to `err.x^2 + err.y^2 + err.z^2` (the last row is...