Screen‐Space Constraints for Camera Movements: the Virtual Cameraman

Abstract

This article presents a virtual cameraman which allows us to obtain the whole set of camera movements satisfying user defined constraints specified in the image space and/or constraints on the objects of the scene. This research follows the "Declarative Modelling" approach, which focuses on a 3-phase modeller concept : description; generation; result exploration. Our tool is based on a solver using interval arithmetic. The time dimension is treated as another variable, thus constraints can be specified for the total duration of the animation or could last only for a given amount of time. There is no keyframing and no interpolation, thereby, for the solutions obtained, the satisfaction of the specified constraints are guaranteed. Several ways to include time dimension efficiently are discussed. We claim that the method is simple enough to be implemented easily without the need of any external solver.After a presentation of previous work, the declarative modelling concept will be exposed. We then explain the necessary mathematical models for the scene objects, the camera and the treated movements. After this, the description process and the properties used to specify a desired sequence will be described. The constraint solver based on interval computation will be exposed before the solution space navigation process. finally, some results will be shown. The paper ends with some future research directions.

A Constraint-Based Approach to Camera Path Planning

“…A precise comparison with the former approach [7] is difficult: the paths computed by the hypertube approach are more complex; the solutions are natural since established camera movements are used; and good quality solutions are propagated. Nevertheless we can state that our results are very encouraging: the previous approach needs important execution times (more than 10 minutes for the first solution) for problems defined over more than 7 variables.…”

“…Recently, several constraint-based approaches [7,1,6] have been investigated. In this framework, the aim is to give the user declarative tools for animation modeling.…”

“…The main feature of Jardillier and Langunou's work [7] is the specification of camera movements by a high-level intelligent process. However, a number of drawbacks can be identified.…”

“…Our research follows Jardillier and Languénou's work [7]. Their approach is based on the following ideas: (1) the user provides a high-level cinematograhicbased description; (2) each property given in this description is written as a constraint; (3) a constraint solver, which implements pure interval methods, computes solution sets in which the animator navigates.…”

Automatic Camera Control: A Dynamic Multi-Objective Perspective

“…position and rotation. The first seminal works addressing automatic camera control according to this model [1,13,15] defined the concept of frame constraint and camera optimisation. These approaches require the designer to define a set of required frame properties which are then modelled either as an objective function to be maximised by the solver or as a set of constraints that the camera configuration must satisfy.…”