texrecon script takes as input several images, their camera
poses, camera intrinsics for one or more sensors, a mesh, in .ply format,
and creates a textured mesh as an .obj file. This tool is a wrapper around
the third-party MVS-Texturing software.
See also the
voxblox_mesh program (Section 16.68) which
can produce a mesh that is usable with this tool.
has an example of how a set of camera images acquired with a rig were
co-registered using the
(Section 16.55), and how a mesh was created.
With that data and this tool, a textured mesh can be obtained as follows:
texrecon --rig_config rig_out/rig_config.txt \ --camera_poses rig_out/cameras.txt \ --mesh rig_out/fused_mesh.ply \ --rig_sensor sci_cam \ --undistorted_crop_win '1000 800' \ --out_dir texrecon_out
The inputs to this program need not be created with
rig_calibrator. What is important is that the camera poses and the
mesh be co-registered.
16.65.2. Command-line options¶
- --rig_config <string>
Rig configuration file.
- --rig_sensor <string>
Which rig sensor images to texture. Can be more than one (bound by quotes then). Must be among the sensors specified via
- --camera_poses <string>
Read images and camera poses from this list.
- --mesh <string>
The mesh to use for texturing, in .ply format.
- --subset <string>
Use only the subset of images from this list.
- --undistorted_crop_win <string>
The dimensions of the central image region to keep after undistorting an image and before using it in texturing. Normally 85% - 90% of distorted (actual) image dimensions would do. Suggested for Astrobee images: sci_cam: ‘1250 1000’ nav_cam: ‘1100 776’. haz_cam: ‘250 200’.
- --max_texture_size <int>
The maximum size (in pixels) of each texture file created for the produced textured mesh.
- --out_dir <string>
The directory where to write the textured mesh and other data.
- --texture_alg <string>
Use one of the two texture creation modes: ‘center’ (for a surface patch choose the image in which the patch shows up closer to the image center; this is the default), or ‘area’ (for a surface patch choose the image whose camera view direction is most aligned with the surface normal).
If set, skip a postprocessing algorithm which may remove some seams but which on occasion can cause a crash.