Realisation (see also technology and other relevant sketches: [image gallery] [online video from some earlier project] Note: one should remember that the project is realized as steroscopic image i.e. some of the scenes are perceivable in depth - behind the surface and popping out of the projection surface - filling the room with multiple layers of images and fluid shapes. Virtual (3D) scenes from a meta-database, audiovisual assets (or data streams) and locative information, mixing common 2D, 2.5D, 3D and special stereoscopic imagery, synthesized 3D sound. Images and video sequences in the data base are analysed for: 1) average color properties 2) texture / pattern segmentation (at lest basic indexing) 3) pixel properties across the edges and a 'connection map' options are generated 4) access statistics (material usage, references in other projects) arbitrarily defineable 'connections' are visualised as spatial moving sculptures, consisting of dense 'tissues' of the illusory space. The surface of image grows into spatial objects, materiality. The stereoscopic effects is also created by mapping images onto cell-like, convex and concave 3D shapes and skewed planes, so that sometimes somewhat paradoxical spatiality is achieved through combination of various perspectives and stereoscopc depth illusion. Multipe virtual cameras generate also "endo perspective" views - where some angle of the scene part can becomes a new dynamic image re-used in nearby scene. Sound channels from various sources are remixed into spatial soundscape. 3D scene are constructed from "cells" of related, juxtaposed and superimposed images and video seqences and sound tracks. For chosen sequences perform video tracking process (e.g. follow some pattern presence or movement registration), convert data to space/time coordinates and create spatial structures and animated modifications of the surface properties (e.g. a "butterfly to earthquake" effect in the VR world) The sound material also can be mixed, panned (moved) in 3D space, apply real-time effects, related to image modifications. scenes are "bridged", "nested into each other", real footage mixed with abstracted computer structures - extracted paths from moving objects leave patchwork of traces, creating fluid landscapes, where the "activity" of object detail gets "inverted" into a "frozen spatial mutation". The user can initiate image mutations in several "dimensions", that get calculated "in background" and gradually update (change scenes). The navigation possible with help of touchscreen (which indicates a "map" of explored and unexplored variations, or switch to new scene generation, and adjust parameters) and/or conventional game pad/joystick. Alternatively(or additionaly) as a performance with sophisticated input devices like data glove by author(s). ( In this case it resembles "virtual surgery" of "audiovisual tissues" of the VR scenes) Concluding: the proposed work is an experiment without an entirely predictable result. It could make an impression ranging from of a fascinating cinematic 'ambient audiovisual landscape', live painting, to a portrait of a breathing organism, or to extreme (overloaded, explosive) game experience. Due it's integration in the shared database, and more or less direct manipulation of several users, the work could have dual or triple presentation form: as a stand-alone installation, as a side-by-side extension to connected installation (game environment, for instance), and a (streaming) video content generating "tool". Some of the media cross conversion might be done "manually", some "automated", but the interaction/navigation in the VR would be smothly realtime. other relevant material : [image gallery] [online video (other earlier project)]