Project done at Visgraf, IMPA, for Globo TV network, Brazil, to enable video compositing with camera movement, integrating actors and virtual scenario. The software implemented allowed the control of a robotized camera, and the planning of takes, using different scales for the virtual and real parts. Ideally, a robotized camera has several sensors and step motors, to control various camera and position parameters, such as position, pan, tilt, roll, zoom and focus, some of which are shown below. The robotized camera in question had four basic controls: pan, tilt, zoom and focus.
The multi-scale compositing allowed the integration of virtual scenarios, either from miniature models or computer-generated, with real actors. The movement of the camera had to be reproduced at different scales to allow for a seamless final composited take.
For a static camera, it is necessary to scale all measures, and to adopt reference points that can be reproduced in the different scale to attain registration, since small displacements in different scales can result in error multiplication. For a moving camera, the same principles of camera positioning hold, but new problems, such as parallax, occur when the camera direction is changing.
The robotized equipment, or the memory head, has to be positioned such that its center of rotation coincides with the nodal point of the camera. A device to allow for the proper calibration of the memory head was devised to help in this part of the process. The software and the equipment have been used at Globo TV network in the production for scenes where repetition of the camera movement is necessary.
This page is under construction.