This flashcard is just one of a free flashcard set. See all flashcards!
17
What are Layered Depth Images? How to render them with correct occlusion?
Input: Set of images with colors and depth information.
Layered Depth Image: All of these input images combined into one image (from a new viewpoint), where along each viewing ray there may be several pixels with different depths and colors.
We can render an LDI by splatting all points. Points are transformed by the camera matrix:
can be decomposed into 4 row vectors:
If we know a transformed point
, new points in the vicinity can be computed incrementally:
Correct occlusion can be ensured by projecting the camera position into the LDI. This point splits the LDI into 4 quadrants which are rendered row-wise outside-inwards towards the epipolar point.
Layered Depth Image: All of these input images combined into one image (from a new viewpoint), where along each viewing ray there may be several pixels with different depths and colors.
We can render an LDI by splatting all points. Points are transformed by the camera matrix:
can be decomposed into 4 row vectors:If we know a transformed point
, new points in the vicinity can be computed incrementally:Correct occlusion can be ensured by projecting the camera position into the LDI. This point splits the LDI into 4 quadrants which are rendered row-wise outside-inwards towards the epipolar point.
Flashcard info:
Author: janisborn
Main topic: Informatik
Topic: Computergrafik
School / Univ.: RWTH Aachen
City: Aachen
Published: 18.05.2022