CS 655 Advanced Computer Graphics - PowerPoint PPT Presentation

1 / 18
About This Presentation
Title:

CS 655 Advanced Computer Graphics

Description:

If we take several images from different locations, and get a correspondence of ... Free. 5. 1995. Plenoptic Modeling. Free. 7. 1991. Plenoptic Function ... – PowerPoint PPT presentation

Number of Views:108
Avg rating:3.0/5.0
Slides: 19
Provided by: cs6
Category:

less

Transcript and Presenter's Notes

Title: CS 655 Advanced Computer Graphics


1
CS 655 Advanced Computer Graphics
  • Image Based Modeling and Rendering

2
What is it?
  • Image Based Modeling (IBM)
  • using images to assist in modeling
  • Image Based Rendering (IBR)
  • using images to assist in rendering

3
Traditional Graphics
Model
Image
Synthetic camera
4
Traditional Computer Vision
Real Scene
Model
Real cameras
5
Combined
Real Scene
Real cameras
Model
Synthetic Camera
Image
6
Problem Vision Falls Short
Real Scene
Real cameras
Model
Synthetic Camera
Image
7
As Does Graphics
Model
Synthetic Camera
Image
8
Image Based Rendering
Real Scene
Real cameras
Model and images
Synthetic Camera
Image
9
Image Based Modeling
  • Images are used to determine
  • Scene geometry
  • Scene appearance
  • Kinematic properties
  • Reflectance characteristics
  • Modeling photorealistic scenes is easier

10
Image Based Rendering
  • Appearance in available views is used to
    determine appearance in novel views
  • Rendering is faster

11
Is This New?
  • Texture mapping
  • Environment mapping
  • Shadow maps
  • Range images
  • Etc.

12
Models From Images
  • How do we derive models from images?
  • Camera calibration
  • Structure from motion
  • Stereo correspondence

13
Camera Calibration
  • If we know camera parameters, it is much easier
    to recover 3D information from the images
  • Intrinsic parameters
  • Focal length
  • Radial distortion
  • Extrinsic
  • Position
  • Orientation

14
Structure From Motion
  • An image consists of a set of points in 3D that
    have been projected to 2D
  • If we take several images from different
    locations, and get a correspondence of points
    between images, we can deduce 3D locations

15
Stereo Correspondence
  • Take two images of a scene from positions close
    to one another
  • Knowing the distance between camera positions
    (the baseline) we can determine 3D shape

16
IBMR Papers
Light Field
Concentric Mosaics
Lumigraph
View Interpolation
View Morphing
Layered Depth Images
3D Warping
Texture Mapped Models
No Geometry
Implicit Geometry
Explicit Geometry
17
Plenoptic Function
  • Function which defines all light through a point
    in space
  • with
  • (Vx, Vy, Vz) the viewpoint
  • q the azimuth angle
  • f the elevation angle
  • l the wavelength
  • t the time

18
Approximations to the Plenoptic Function
Write a Comment
User Comments (0)
About PowerShow.com