If one is using the
mathcam function for one's own work he may find some of the following discussion helpful. I found some idiosyncrasies that degrade graphic quality that are most likely related to PostScript and/or the use of
PlotRegion to magnify the image. The animation was originally developed under Version 2.2 so some of the comments may not be valid under current versions.
Theoretically, the function performed by
LOSAffineShape should not be necessary. Simply setting the viewpoint to be very close to the view volume should give the same image (with the proper setting of the plot range and box ratios). For the luge run, which required a lot of depth, this gave a very large magnification. Practically, during development, I found that if I did not perform the
LOSAffineShape step, I would receive polygons seemingly rotated by 90 degrees. Sometimes they would be rendered on the wrong side of each other. I even had the image render upside down on occasion (on a PC, but not UNIX workstation)! The final straw was when individual pixels mapped nonlinearly to the viewport. This resulted in very "crinkled" polygons. I suspect this was due to numerical round off error.
Similarly, the use of
Clip3D at the front clipping plane should not be necessary. Practically, I found that if I did not perform this step, polygons near the front clipping plane (the slider in my case) would sometimes be colored (filled) on the wrong side. Clipping at the front clipping plane seemed to reduce but not eliminate this. If you look at the limb function you will see three plot points where only two are required for the cone. The extra intermediate points seemed to give the rendering process a definite point to work with, which resulted with the proper image.
When I have completed an animation, I will typically make a graphics array for hardcopy display. I was not able to do this with the luge ride. The
PlotRegion specification is also used by
GraphicsArray which overrides the specification by
mathcam. All 12 frames were larger than desired and overlapped each other.
Under Version 2.2, I had to use a zoom factor of 1.06 to make the viewport fill the window provided by the notebook front end. Subsequent versions allowed the expected zoom factor of 1.0 to be used.
Converted by Mathematica
[Prev Page][Next Page]