1. Introduction
1. Use Gargoyle for curve display? relationship to general plotting, other Cedar plotting?
2. Get images into IP whiteboard.
4. Key goals for Jan 88:
a. having entered a trivariate poly in CaminoReal, a user can select the MakeDisplayData operation, either supply parameters or accept defaults, this forks offline generation (or retrieval) of all needed data for display, notification of completion is given. Anytime display data exists for a trivariate polynomial (this check is done by looking up the current CR trivariate poly in global cad indices, the CheckDisplayData operation), user can invoke a Display operation.
b. Beretta (4?) standard views, or Dobkin standard flybys, leading to automatically generated and stored frames, individual frame or video playback. Generation of both Beretta standard views and Dobkin standard flybys should be the default for GenerateDisplayData.
There could be multiple display operations, e.g. PlaybackStandardFlyby, or StandardViewX, StandardViewY, StandardViewZ, StandardViewSlant.
spec of Dobkin flyby: find bounding box for interesting behavior of surface, find its radius r, then for each axis (X, Y, Z) through the center of bounding box, step back an amount alpha*r for some alpha, look at center, snap, then take a step along the circle of radius (1+alpha)*r, look at center, snap, etc. Do this for each coordinate direction. Save all the images, use ThreeDWorld playback.
c. wireframe display option for 3-d objects.
d. plane curve, 2 space cad, 1,2,and3-space sas display.
5. "Topologically reliable" use of ray tracer is probably a research problem, i.e. being able to guarantee that any image displayed to the user in which one or more objects (clusters) are displayed ray traced doesn't tell lies.
6. Paper outline.
a. Introduction - the problem: hair, isolated points, complex signularities, other possible complex behavior of algebraic surfaces.
b. Summary of cad's, pointer to other published works (Arnon)
c. passing from the algebraic to the graphical world (Arnon & Rauen)
1. we do display of individual cells, rather than of clusters.
2. The issue of "precision" (fineness) of triangulation mesh for cells.
3. The issue of boundaries - topologically reliable "knitting together" along boundaries.
4. the user interface - we have a data base, you can make different triangulations for a given cell, how does this relate to the boundary knitting issue. Ideally the triangulation software browses the database, each triangulation of a cell is Object-type data and gets written back into the database.
5. the user interface - the goal of "lazy evaluation", "lazy adaptive triangulation", i.e if user zooms in, you may do a finer triangulation.
d. Empirical results (Rauen)
1. description of the actual user interface.
2. pictures.
7. Interface design goal.
a. defining formulas should not be needed at all, only specifications of which root of a given polynomial