SuperHack/SuperHack CD.bin/CODING/GRAPHICS/IMPACT21.ZIP/IMPACT.DOC

Text File | 1995-01-27 | 10.5 KB | 185 lines

            Documentation for IMPACT Dynamics Simulator v2.1

Copyright (c) 1993,1995 John Henckel
Permission to use, copy, modify, distribute and sell this software
and its documentation for any purpose is hereby granted without fee,
provided that the above copyright notice appear in all copies.
----------------------------------------------------------------------------
First let me introduce myself.  I'm John Henckel ([e-mail]).
Currently, I work for IBM in Rochester, Minnesota as a computer programmer.
I like to write fun DOS programs as a hobby, and this is one of them.

Impact simulates the collision of 2-dimensional convex polygons with graphical
animation.  The equations for collision of irregularly shaped objects are
taken from "Dynamics" by Pestel and Thomson, 1968.  With the mouse, you can
grab an object and drag it or throw it.  You can stack objects on top of one
another.  You can design experiments interactively, or you can use a text
editor to make scenario files.  As you watch the bodies bouncing off each
other and the the walls, you can get energy reports of the total rotational,
translational, and potential energy.  You can turn on trace and sound to see
the amusing "kinetic artwork".  You can set physical constants restitution
(elasticity), gravity, air friction, and sliding friction.  It works on CGA,
EGA, VGA, or Hercules using Borland BGI. (Help text not readable on Hercules.)
Recommended but not required: mouse and 286 or better.

-----------------------------------------------------------------------------
Tutorial

Experiment 1:  Kinetic doodling
  Start the program and type "a 2 3 4", this will create two tall tri-
  angles.  Position the mouse in the middle of the screen and
  click button 1.  The nearest triangle will move toward the mouse.
  Now hold down mouse buttons 1 and 2.  All the triangles are drawn
  toward the mouse (but not as quickly).  Press 't', now as the bodies
  move they leave a colorful design behind them.  Press 't' to turn off
  trace, press 'c' to clear the screen.  Press 'd' twice to delete the
  bodies.

Experiment 2:  Inclined plane
  Type "a 1 3 10 2" to make a long incline plane.  Type "a 1" to make
  a little box.  Type "g 5" to enable gravity.  Type "r 0.5" to make
  things not so bouncy.  Now using the mouse, pick up the little box
  and set it on the inclined plane.  Then watch it slide down.

The Impact program take one command line parameter which is the name
of a scenario file.  See notes below for the file format.  Some
scenario files are included in this package, for example, type
    impact exp1.dat
to load experiment 1 scenario.

------------------------------------------------------------------------
  R E L E A S E    N O T E S
------------------------------------------------------------------------

1.0 93/11/16 - John Henckel, first release

Room for Improvement:
* The program defines impact as whenever the corner of one body enters
  the interior of another body, or enters a wall.  This sounds simple
  but it has many difficulties.  (a) It is possible for two bodies to
  overlap, but neither has a corner inside the other, this is very
  common with triangles because of their sharp corners. (b) It is
  possible for two bodies to each have a corner in the other, in
  this case I pick one arbitrarily. (c) It is possible for a body to
  hit a wall or another body with two of its corners, in this case I
  choose one arbitrarily.   There are certainly better ways to handle
  these cases.
* When gravity is enabled, one would like to stack objects on top of
  one another.  This doesn't work though because the acceleration
  of gravity is constant, but the acceleration of impact becomes zero
  as the bodies slow down.  Thus, they slowly mush into each other.
  To fix this, I added the 'j' parm for center adjustment factor.
  If s=1.0, then after computing the acceleration of impact, the bodies
  are separated from each other so that they no longer overlap.  This
  opens up another question:  what direction should they be separated?
  I use the normal to the impact surface.  This has the nice side-
  effect of simulating sliding.  However, the sliding is only a program
  artifact and not based on physics theory.
* The center adjustment algorithm sometimes fails because of choosing
  the wrong impact surface.  When a body is moving quickly, the impact
  edge can be found by reversing the relative velocity of the corner.
  However, if a body is moving very slowly, instead I choose the
  edge nearest to the corner as the impact surface.  Sometimes this is
  not right.
* "Dynamics" mentions two possible assumptions regarding forces acting
  parallel to the plane of impact.  One is that they cancel relative
  velocities, the other is that the forces are zero.  I have implemented
  the later, i.e. the bodies are perfectly slippery during impact.
* The so-called air friction is not very realistic.  It is a simple
  constant deceleration v' = v * (1-f).  It doesn't take into account
  the shape of the object for aerodynamics.
* Sometimes, if things get moving too quickly, the program will go crazy
  and shapes will fly off the screen into never-land.  When this happens
  a floating-point error usually halts the program in a few seconds.
  Type "mode co80" in DOS to fix the video mode.
* If you try to recompile the source code, you may have some difficulty
  with the registerBGIdriver().  You need to run the BGIOBJ program
  and change the project options.  Or if you prefer, I think you can just
  delete the line with the registerBGIdriver, just make sure EGAVGA.BGI
  is in the current directory at runtime.

Acknowledgements:
 This program was developed using Borland Turbo C++ 3.0.  The speed and
 versatility of the graphics of this program is a tribute to the talented
 programmers at Borland.  Also their online help is fabulous.
 The equations for collision of irregularly shaped objects are taken from
 "Dynamics" by Pestel and Thomson, 1968, I haven't found such a detailed
 analysis of this problem in any other book.

-------------------------------------------------------------------------
 1.2 93/01/20 - Some improvements:

*   The collision algorithm has been substantially improved.  The
  dynamics equations have not been changed, but the method of
  determining which body hit which.  When a body hits a wall, the
  corner deepest in the wall is used for collision.  When two bodies
  collide, all corners on both bodies are searched to find the one
  with the deepest penetration and it is used to compute collision.
  These two changes REALLY helped the stability of stacking objects.
*   I added sliding friction.  When the velocity normal to impact
  between two bodies or a body and a wall is less than some threshold,
  then velocity is changed in each body parallel to the surface inversely
  proportional to mass to decreases the relative motion.  Sliding
  friction is also important for stacking objects.
*   Now you can create scenario description files which can be loaded
  from the command line.  Three scenario files are included in the
  package.  The first line in the file contains 7 physical constants
      friction gravity center-adjust restitution trace-mode
        sliding-threshold sliding-friction
  these are all floats, except trace which is 0 or 1.  The remaining
  lines have the same format as the interactive "a" command.
      count sides aspect-ratio radius color noise mass
        moment pos-x pos-y pos-a vel-x vel-y vel-a
  Trailing data may be omitted.  If you specify 0 for color, mass,
  moment, then the program will generate reasonable values.
  Moment is actually I/m.  Angular data is in radians.  Be careful
  about large numbers, the program does very little checking.
  The virtual screen is 30 by 40, with origin in bottom left.
*   I added a new feature to get an ENERGY REPORT while the program
  is running.  It is fun to watch angular, translational, and
  potential energy change while the total remains about the same.
  The total varies some, I think it is because of the center-adjust
  algorithm.
*   I changed air friction to be v' = v/(1+f*v) which seems more
  realistic.
*   I added some checks to catch things before they crash the
  machine.  You still need to be careful not to set gravity too high.

-------------------------------------------------------------------------
 2.1 95/01/26 - Some improvements:

*   I added some performance timers and discovered that 60% of the time was
  spent drawing the screen border!  So I fixed that.
*   I tried to add joints but did not succeed.  You can try the joint.dat
  file and see.  The rotational inertia is not conserved as the joint moves
  and the bounce is not correct when the limb hits a wall.  You're welcome
  to try and fix it, I have decided to start over using a new design in c++,
  (someday).  One note:  in the joint file, if you make one part jointed to
  another, then its position and velocity are ignored.  The position and
  velocity of the dependent part is based on the root part.  The "root" and
  the "joint" parent part are not necessarily the same, in case you have
  an arm with several joints.  I haven't tried this, though.  The rest of
  the program works fine, only the joints are buggy.
*   I improved the help text, and other parts of the user interface.
*   Impact does not adjust itself to your CPU speed, it runs as fast as it
  can!  If you have a screaming 90 MHz Pentium then you may need to adjust
  the sample files -- gravity and velocities.  The faster your CPU, the
  smoother and more accurate the simulation.

I do not want you to send me any money for this software, but I do like to
get email.  If you make any improvements, please send them to me.  Thanks.

Romans 9:19ff
One of you will say to me: "Then why does God still blame us?  For who resists
his will?"  But who are you, O man, to talk back to God?  Shall what is formed
say to him who formed it, "Why did you make me like this?"  Does not the
potter have the right to make out of the same lump of clay some pottery for
noble purposes and some for common use?  What if God, choosing to show his
wrath and make his power known, bore with great patience the objects of his
wrath--prepared for destruction?  What if he did this to make the riches of
his glory known to the objects of his mercy, whom he prepared in advance for
glory--even us, whom he also called, not only from the Jews but also from the
Gentiles?

John Henckel      [e-mail]