superai-20240310-3d-reconstruction-two-important-representation-and-camerapose.txt

Representation:
	implicit:radiancefield: noNN(Plenoxels)
	implicit:radiancefield: mlp(NeRF)
	implicit:radiancefield: plane/triplane/multiplane(triposr)
	implicit:radiancefield: tree(PlenOctrees)
	implicit:radiancefield: hash(InstantNGP)

	explicit:discrete: points(dust3r)
    explicit:discrete: points++(GS/GS++)
	explicit:discrete: voxel(Leap)
	explicit:continuous:combinatorial: sdf/udf
	explicit:continuous:combinatorial: mesh(FlexiCubes)
	explicit:continuous:functional: *

	other:hybrid:gs+surface(SuGaR)
	other:hybrid:volume-surface(BakedSDF)
	other:combination:misc(Brep)


CameraPose:
    keypoints-correlation:hard:colmap
    keypoints-correlation:soft:dust3r

    pose-nothing:SplatterImage
    pose-render-only:LEAP
    pose-differential-equation:RobustLoc
    pose-auto-gressive:TripoSR
    pose-joint-optimize:nerfmm
    pose-joint-optimize:nerftrinsic-four 
    pose-explicit-geometric+video-continuity:colmap-free3DGS
    pose-space/perspective-warping:nerf-f2
    pose-monocular-depth-prior:nerf-nope
    pose-relative-rotations:realpose++