On Sun, Mar 16, 2014 at 3:51 PM, Paul Doyle <[email protected]> wrote: > What is unclear is how the ICE approach (as a high-level visual programming > paradigm) meshes with Bifrost as publicly shown to date - I expect that is > driving the questions people are raising. Because of that, I think it is > problematic to say that Bifrost is the spiritual successor to ICE. Nobody is > really explaining how that's the case, beyond it's also going to have a > visual programming system - but that's like saying Maya and Softimage are > the same because they both have a scene graph.
Here is the link between ICE and Bifrost. When we worked on ICE at Softimage, there questions about how ICE could be used in other markets such games or also used in the rest of Avid as an fx module. However, the implementation ended up being 100% tied to XSI's innards and not a separate engine. But the idea endured that it would have been the right thing to do, to have a "ICE runtime". As you know very well, after making some game middleware authoring tool prototype with XSI, the team then went on to work on project Skyline, where they created an stand alone evaluation engine that could have been the successor to ICE, and there had been talked in the first year of the possibility to plug that BACK into Softimage at a later date. Marc Petit was hoping to get a portable ICE engine as a side effect of Skyline, but that could not happen because the team's focus was animation (i.e. processing vector, matrices, and doing new things like state machines) and targeting single-threaded game consoles, and not particle work on a PC. There was just a load of other things to do . But the principles of the graph, with get/set data and polymorphic nodes, for example, are similar. So now we're at bifrost and the siggraph 2013 demo of it, which you can see here at the 8 minute mark http://tinyurl.com/nsswz4n is running on work that is evolving from these previous efforts. The prototype shown was even using the skyline graph, although that won't be what we ship.
