search

nmoehrle / mvs-texturing

Algorithm to texture 3D reconstructions from multi-view stereo images
Other
931 stars 328 forks source link

Texture from RGB-D camera is not aligned to 3D mesh #177

Closed wodndb closed 3 years ago

wodndb commented 3 years ago

I want to texturing 3D mesh using RealSense D415 camera.

RealSense D415 Camera is RGB-D camera which has two IR camera to generating stereo depth image and one color camera.

  1. I captured a point cloud and color image from D415 camera.

  2. I reconstructed point cloud to ply format 3D mesh.

  3. I prepared cam file which include intrinsic and extrinsic parameter of D415 color camera.

  4. I build https://github.com/andre-schulz/mvs-texturing to use texrecon in windows.

Extrinsic parameter of color camera is below matrix.

image

And result is this (below image). Texture is not aligned.

image

And I translate the mesh to (-0.015, 0, 0) and change the extrinsic parameter to identity matrix.

And a result like this. (below image) Texture is aligned.

image

Why these two things are differ?

I attach two data set what I show in upper images. default.zip mesh_translated.zip

Thanks.

wodndb commented 3 years ago

I'm sorry. I mistake to make data set. This issue is not correct.

lihui1998 commented 2 years ago

I'm sorry. I mistake to make data set. This issue is not correct.

Can you tell me how to use cam file,thanks

wodndb commented 2 years ago

@lihui1998 I generated cam file based on this description. https://github.com/nmoehrle/mvs-texturing/blob/master/apps/texrecon/arguments.cpp#L33