Kitware Scene Graph Meeting April 6, 2010 KHQ

Kitware Scene Graph Meeting April 6, 2010 KHQ

1. Presentation by Michel of slides briefly describing JHU and existing Kitware requirements for KWScene. [1]

2. Description by Julien Finet at KRS of requirements for ParaView for transforms, which led to a description by Luis of transforms in KWScene. The possibility of reusing the IGSTK spatial relationships model, where every object is connected to a parent was evoked. IGSTK uses rigid transforms, which could be expanded here. He presented two potentially conflictint use-cases, namely a large collection of small objects, such as the Computer Vision group's requirement, or a small collection of large objects, as is the case for JHU.

3. Amitha presented a Computer Vision group usecase in more detail, which would involve a reprojection on a 3D space onto where an image came from on the ground. Also a possibility of affixing avatars to 3D space. A rapid refresh rate, 30 Hz, would be needed.

4. Berk boiled down the Scene graph use-cases to two: i) representation of a hierarchy of objects, and ii) input/output of that hierarchy. He suggested that the Scene graph should not compete with VTK. It was later pointed that VTK flattens the representation, and that KWScene would bring something complementary.

5. Aashish indicated that a requirement of his is to make the scene graph fast, in cases where the scene data does not change (often).

6. Will listed 4 requirements of the Scene Graph: i) representation, ii) rendering, iii) attributes, and iv) association with annotation (e.g. labeled anatomical atlas).