Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Image-based rrendering (ibr) is a technique used to produce a new image from real images by combining and interpolating them. This process requires correspondences between the images to sensibly combine them. In this document, we explore various methods for getting correspondences, matching windows, morphing, and interpolation. We also discuss advanced topics like light-field rendering and non-photorealistic rendering.
Typology: Papers
1 / 18
This page cannot be seen from the preview
Don't miss anything!
Produce a new image from real images. Combining images Interpolation More exotic methods
What’s the most realistic image? A photograph. But photographs lack flexibility. Can’t change viewpoint. Can’t change lighting.
Image-based rendering is mostly combining images to get a new image. Correspondences needed to sensibly combine images. If viewpoint has changed this can be hard. If not, it’s trivial.
By hand: Works if few correspondences needed By matching intensities This is really ~ ½ of computer vision.
Often done by hand.
Interpolate each point. Position and intensity.
Also use interpolation for more correspondences.
Also interpolate intensities.
Interpolate to find other point correspondences.
Sample the set of light rays in the world.
Then generate an image by selecting the right rays.
Mosaicing: simpler, just sample rays through one focal point.
If one has all rays then camera can also move.
(slide from Mark Ollila)
Light Field Rendering (Levoy and Hanrahan; paper and slides)
Light Field Rendering (Levoy and Hanrahan; paper and slides)
Light Field Rendering (Levoy and Hanrahan; paper and slides)
Light Field Rendering (Levoy and Hanrahan; paper and slides)
Light Field Rendering (Levoy and Hanrahan; paper and slides)
Example, generating views for teleconferen cing, with depth from stereo.
Generating using two views in which speaker isn’t looking at camera.
(^00 1 2 )
1
(^00 1 2 )
1
2
r
Z
Y
X
2 !(Z -X -Y )^2 2 2 !(X -Y )^2 2 ^ XY ^ XZ ^ YZ
Note, this can also be done with 9 real images, because this is a basis that contains real images In 3D, real images aren’t in the 3D space, we have to take the max with 0 to get real images.
Take a photo and turn it into a different kind of image.
