Introduction

Methods and tutorials for projection mapping 3D objects where a physical mesh of the object is available. It basically allows to retrieve the extrinsics and instrinsics parameters of a projector based on manually defined correspondences between the 3D scene (3D world points) and the projection space of the projector (2D points).

See also the contribution page of this tool for the DX11 version.

CalibrateProjector

Using OpenCV's CalibrateCamera method to find the intrinsics and extrinsics of the projector (i.e. the projection and view matrices respectively).

Inputs

• 7 or more World to Projection correspondences

Controls

• S: Orbit
• D: Pan
• F: Zoom
• 1: Next point
• 2: Previous point
• Shift: Zoom to current point

Prerequisites

• vvvv24.1+
• Elliot Woods' VVVV.Packs.Image that wraps OpenCV (EmguCV)

Notes

• Probably due to some misunderstandings from my part, I got the correct pose estimated with the following configurations:
  • Points in projective space (XY) converted to OpenCV's image coordinate using the node "MapNormalisedToCamera"
  • Input Space: pixels
  • Inside "CalibrateProjector": Coordinates: OpenCV
  • Rotate the view matrix by (0,0,0.5)