# Difference between revisions of "Stitching a photo-mosaic"

(→Understanding the mosaic-lens model) |
|||

Line 14: | Line 14: | ||

== Understanding the mosaic-lens model == | == Understanding the mosaic-lens model == | ||

− | The panotools [[lens model]] has been expanded with addition of three new parameters that describe the camera position in cartesian coordinates, | + | The panotools [[lens model]] has been expanded with addition of three new parameters that describe the camera position in cartesian coordinates, TrX, TrY, and TrZ |

− | The [[yaw]], [[pitch]] and [[roll]] angles now apply on the camera around the camera's | + | The [[yaw]], [[pitch]] and [[roll]] angles now apply on the camera around the camera's TrX, TrY and TrZ position in space |

## Revision as of 14:05, 27 April 2010

This section is a stub.

## Definition of photo-mosaic

A photo-mosaic is an image that is stiched together by photographs all taken from different viewpoints. This differs from the panorama where all images are taken from one single viewpoint but with different angles.

The typical case of a photo-mosaic is the linear panorama, where one large subject like a wall, is photographed by taking photos of the wall, each covering a proportion of the wall, and moving the camera parallell to the wall with each shot.

The classical way to stich these kinds of mosaics was to lay the prints out on a light-table and glue them together. Now with the mosaic-mode in the panotools lens model, it is possible to use the software estimate the camera position including angles, lens distorsion etc. and to reproject and stich the images to a photo-mosaic.

## Understanding the mosaic-lens model

The panotools lens model has been expanded with addition of three new parameters that describe the camera position in cartesian coordinates, TrX, TrY, and TrZ

The yaw, pitch and roll angles now apply on the camera around the camera's TrX, TrY and TrZ position in space

Legend:

- The unit-panosphere is drawn in gray.
- "The panorama camera" is centered in Origo, X,Y,Z =0
- Three photos are taken with camera positions 1, 2 and 3

Note that I have given these cameras different local Yaw angles

My assumption was

- The model uses separate camera positions 1, and 3 that have different X and Z values in this case
- The photos are reprojected to a plane that touches the panosphere in (0,0,-1) and then onto the panosphere using the "Panorama camera"
- The camera angles (y, p ,r) are applied in each camera position