View-based Rendering: Slide 13 of 17.


Click slide for next, or goto previous, first, last slides or back to thumbnail layout.

Click slide for next, or goto previous, or back to thumbnail layout.

When we scan the object from different directions, we typically reorient the object in front of a stationary scanner between scans. With fixed light sources, the lighting conditions with respect to the object change from view to view. This leads to visual artifacts at mesh boundaries where suddenly one of the views ceases to contribute to the image. This is illustrated on the top right.

To remove or at least diminish these artifacts we smoothly decrease the influence of a mesh toward its boundaries. This is illustrated in the image at bottom left: bright means full contribution, darker means that a pixel contributes less to the final image.

Image on the bottom right shows the effect. Like the previous weight, this weight does not change when the virtual camera moves.