Meeting Date: 02.08.1999
What I did last week: I finally got around to writing the mouse-controlled input device I always envisioned.I am not using xev code though. Shaun thinks that an output-device should't be serving as an input-device (which my appraoch does, since I use the Mousefunctions already provided by openGL, which in this case meant connecting my VRMouseGlideInputDevice to the VRSink). But apart from that it showed that I by now understand CoRgi quite well and was able to integrate an new input-device smootly into the system. Matt liked it and following his suggestions I added a form that allows you to Scale the movement and brings you to a pre-defined position in space if you feel the need. On the weekend I did quite some extensive surfing to find out more about gesture recognition on the web. There is tons of stuff out there but usually only using finger positions or video input. What I did find though was some handwriting recognition software called REMUS. It's only a demo-version but works quite nicely. If I could get the source code for that I could incorporate it into CoRgi instead. This way I could define my normal gestures as more characters and write in thin air. I wrote an email to the people that developed the system, but no reply yet. The next Honours presentation is on Wednesday and so I did a lot of preperation for that. Making things look pretty and stuff like that :-)
What I'll do this week: Waiting more for the reply to my e-mail request. Other than that I want to start preparing some stuff for my actual object Manipulation application. I want to write a VR Command Input Actor (should be one word, but don't try that in Frontpage express - did I ever mention that that must be the worst case of software ever written? Even for Microsoft its quite appalling) which takes input from diverse input devices like Speech / Hand / Wdgets / Gesture Recognizer / etc. processes the data internally and pipes out predefined "commands", so that in the end a user will be able to connect to one of those Command Input Actors and receive input-commands from all sorts of input devices without having to bother where they came from or how.
I'll also have to do some cleaning up of my existing code, because the OFFTransformShape class moves vertices absolutely and shaun thinks one should be able to move the objects around relative to some origin and not by moving each and every vertex by hand. He might have a point there (or a vertex?) so I will have to find a place to convert vertices from
absolute coords to relative ones and then back. This will probably be done by the OFFTransformShape, since it knows best about its position and orientation and it is not really the job of the SelectionSplineActor to deal with this kind of thing anyway (In my humble opinion).
Then I'll have to see that I get my Linux up to date, so I can re-start compiling things. Since Mike broke everything (;-)) by incorporating Magick++ into the system. Too bad for me.