SandboxConcept
From Moviesandbox
| (6 intermediate revisions not shown.) | |||
| Line 32: | Line 32: | ||
* Splitting the directorial part from the engine | * Splitting the directorial part from the engine | ||
* Creating an output stream that could be interpreted with a core infrastructure (a player basically) | * Creating an output stream that could be interpreted with a core infrastructure (a player basically) | ||
| + | |||
| + | [[Category:Projects]] | ||
Current revision
the Sandbox concept is a project that tries to look beyond Moviesandbox and investigate the possibilities of a new set of tools for realtime animation production.
Contents |
the general idea
The Sandbox Concept tries to use a 3D engine as its realtime renderer, or display. The virtual environment that is being displayed is created and modified using a number of independent modules, that communicate with the 3D engine by a standardised language such as COLLADA[1] or something yet to be created.
The individual Tools can consist of highly specialised programs such as PlayViz[2] or the Multi Track Input Recorder[3] from GeorgiaTech's TUI3D[4] group, the Panda Puppet project[5] or the VoxelEditor and some of the MSBTools that are already implemented.
By exchanging data with the 3D engine in realtime, you can work cooperatively and simultaneously in the same virtual environment.
a common language
In order to establish communication between the different modules and elevate the Sandbox Concept to a non-proprietary idea (not linked to a particular 3D engine), it is necessary to find a protocol that is capeable of dealing with all the different tasks and function of realtime filmmaking - be it modelling and lighting a scene, staging a character or defining a camera path.
A language like COLLADA seems to be a good starting point for this, especially for the animation, modelling and lighting modules.
The directing module however needs to provide a higher level language that is closer to common sense and film language than xml. It will be necessary to find a basic set of commands and grammar that can be used by filmmakers to direct virtual characters and link the appropriate resources from the other modules in a reasonable way.
Moviesandbox right now uses linked graphs as its grammar and a basic set of actions as its commands.
Interfaces and Modules
With realtime modification comes the possibility to rethink our content creation tools both in the way we interact with them (interfaces) and in how specialised they are. Below, you'll find two examples of relatively low-cost interfaces (for modelling and camera control respectively) within a 3D environment:
This video shows the Milkscanner as a low cost scanning technology for characters. It has since been revised to also work with balsamic vinegar. The tools takes still images from a webcam and layers them on top of each other according to the amount of milk poured into a tupperware bowl with the target object inside. The layered image then serves as a displacement map that can be used together with the Tracer Tool to create a character in the VoxelEditor.
This video shows the use of MSBTools as camera control for a Moviesandbox camera. It uses the GameTrak Device[6]] and a WiiMote (not in demo-video) to track 3 axis of rotation and 3 axis of movement and sends it to Moviesandbox.
the story so far
Moviesandbox is a first step in trying to embrace the sandbox concept.
the road ahead
Some of the things that are needed right now are:
- A timeline editor and central data storage/recorder
- Splitting the directorial part from the engine
- Creating an output stream that could be interpreted with a core infrastructure (a player basically)
