This home page is part of my final project of the 3D Graphics Systems Course, taught by Luiz Velho at IMPA from March to June 2009.
Here you'll find the slides and movies that I made for my final presentation, the references that I used, the results that I got and also the entire code that I developed to make the software that I'll give some details below.
The most known images from real world scenes are the ones that are got via a perspective transform. For example, pictures taken from a simple digital camera.
Although they have many desirable properties, the limitations of perspective pictures gave rise to multiperspective pictures developed by artists long time ago. With the use of more than one perspective on the same image, they could express more feelings and ideas and could improve the comprehensibily and representation of a scene. For example, objects that were too distorded on a simple perspective, could be corrected using other local perspective or objects that didn't have their geometry well represented, could be corrected also.
On this project, we follow the ideas from reference , where the authors developed a software for Multiperspective rendering but in computer made enviroments. Here, we'll develop these ideas for real world scenes.
I developed an interactive software using the FLTK (Fast Light Toolkit) on C++ language. The user gives as input a set of views of the scene, represented by spherical images.
The fisrt window allows the user to select a center of projection and the level of details for each view:
On the second window, the user chooses the occlusion between the objects:
The third window is where the user selects a camera for each object:
And the fourth window displays the final result:
For more implementation details, please check the Presentation section. There you find also some demonstation videos of the program working.
And for more results, please check the Results section.
 M. Agrawala, D. Zorin, T. Munzner: Artistic Multiprojection Rendering . In Eurographics Rendering Workshop, 2000.
 V. Kolmogorov, R. Zabih: Computing Visual Correspondence with Occlusions using Graph Cuts. In International Conference on Computer Vision, 2001.