Hi, i need a bit of help solving this issue of the final reconstruction as you can see in the 3a.mp4 it is full of mistakes, but on 2a.mp4 before the final reconstruction it seem to be OK, and on the 1a.mp4 you can see where the cameras overlap there is this black band… isn’t it suppose to function like one single camera? is this the cause of the problem with final reconstruction?
3 intel real sense D435 - win 10 32Gb - gtx1080Ti 12Gb
the final textured model is generated by mapping the keyframes onto the mesh. When the transformations are not estimated ideally, the observed situation like in video 3a may occur. You can try to improve the texturing by removing the keyframes that don’t overlay with the mesh well. Another thing to keep in mind is that the person needs to remain completely immovable during the scan.
The black band on your first video is resulting from the fact that the depth image is not aligned with the color image, therefore we get 3d measurements for some points without the color. This is expected behavior and this doesn’t affect the final reconstruction.