Hi, first of all congrats on the fantastic work!
I want to run dust3r on 2 sets of images and then merge the output of these 2 sets to have a complete scene. in other words, reconstruct a scene that is split into two distinct chunks of images.
my idea is to save important data from scene for each set of the images like:
objects = [scene.imgs, scene.get_pts3d(), scene.get_masks(), scene.get_focals(), scene.get_im_poses()]
AND then try to merge them and finally pass to _convert_scene_output_to_glb unction to have final output.
BUT it does not work and the algorithm loses camera position. i will appreciate it if you share your ideas with me.
thanks in advance.
Can you make sure that two scenes are in the same scale? In other word, the pts3d of two images is may have difference scale even they have some common images.
Hi, first of all congrats on the fantastic work! I want to run dust3r on 2 sets of images and then merge the output of these 2 sets to have a complete scene. in other words, reconstruct a scene that is split into two distinct chunks of images.
my idea is to save important data from scene for each set of the images like:
objects = [scene.imgs, scene.get_pts3d(), scene.get_masks(), scene.get_focals(), scene.get_im_poses()]
AND then try to merge them and finally pass to
_convert_scene_output_to_glb
unction to have final output. BUT it does not work and the algorithm loses camera position. i will appreciate it if you share your ideas with me. thanks in advance.