Perspective Grid (GUI for Image-Based Modeling)

Excerpt from Chapter 5 (full text)

The success of the TIP technique rests largely on the fact that the user can model the scene quickly and easily. A good GUI is the key. The spidery mesh used in the TIP technique can only shows a vague sense of depth and that the information given by it is insufficient to determine all the viewing parameters in the modeling. We attempt to improve the spidery mesh and we call our improved interface the perspective grids. The spidery mesh and the perspective grids are shown in figure 5-1 for comparison. Essentially, perspective grid is a spidery mesh with uniform grids on it.

Figure 5-1. A comparison of spidery mesh and perspective grids.

The perspective grid GUI has two advantages over the spidery mesh: visual adjustment of camera parameters and visual image feature matching to produce a more accurate model. Although in our current application of animating pictures with a rough model, a more accurate box model may not make a meaningful difference, we believe that this GUI is suitable for applications that require accurate modeling of box-like shapes or, in general, structures with orthogonal surfaces from existing images.

[...]

Figure 5-5. Dragging the corners of a grid cell to match some known shapes in the image using our GUI. (a) Selecting a grid cell. (b) Control points appear. (c) Dragging a control point to match a known pattern on the ceiling. Notice that other cells on the ceiling are aligned too.

- See full text for more information.

Copyright (C) Nelson Chu, 2001. All Rights Reserved.