3D Scene Control Research Based On Through-the-Lens

For decades, with the rapid development of computer technology, the applications of Computer Graphics also become more prosperous. Among the application fields of CG, 3D scene design is of great importance .It can be found in such fields as 3D animation system, video design, architecture design and metropolitan planning, etc .Therefore, it is necessary how to construct a properly 3D model and project it on screens as the designers' intention.The thought of Through-the-lens Control (TTLC) improves the efficiency and interactivity of 3D scene design largely, by allowing users to operate the projections of objects directly in order to set the components in 3D scene.This research is based on the thought of TTLC and existing research results, declaring a more integrated set of control techniques based on Through-the-lens. The purpose of our research is that the system can recalculate the parameters of multi components, including cameras, objects, lights and shadows, according to the inputs from users. The ultimate goal is to improve the efficiency and interactivity, so that users can both focus more attention on designing and conceiving and make more perfect composition in a short time.This research declares how to manipulate cameras, objects, lights and shadows, taking advantages of the theories and methods in TTLC, as well as the principal of constraint planning .The constraints to the four components above are set up under certain models. The final constraints virtually express the relationship between component parameters and user inputs. Then we can get the solution space about the parameters by solving the equations. However, after setting up the constraints, the constraint equations are non-linear. Here, we convert the non-linear problems to linear ones by introducing Jacobi Matrix, which decreases the complexity of solutions. What's more, we also discuss the problem of under-constraints and give out the solutions to it.This paper contains some experimental results which are explained immediately. Remember that we improve the efficiency via converting the non-linear problem to linear ones, which decreases the accuracy of the results to some degree. However, in contrast to the improvement of the efficiency, errors are absolutely tolerated.Considering the results, the theories and methods in the paper can be applied to the realistic 3D scene design.