8:30 – 5:30 FP

• Adding environment methods that will coordinate the interactions of the objects.
• Cleaning up Arvo’s so that only 3D collisions between solid objects are calculated, with a resulting unit penetration value.
• Collision detection and response are now working!
• Adding a few more cubes to collide with. Then work on global to local mapping so that I can have the cubes at arbitrary angles.
• Added the cubes, now I need to clear and sum the collision forces each time through. Done.
• 
• After spending the entire afternoon unrotating objects in my head, I got the projections onto the object in it’s coordinate frame working. So bounding boxes work at any arbitrary angle. Whee!

8:00 – 10:30 FP

• Working on collision detection. It just needs to be convex hull. At this point I’m thinking of doing the following:
• For each polygon
• calculate the shortest distance to the plane
• If the square of the distance is less than the collision radius
• See if the collision point is within the polygon (project the plane onto X, Y or Z, if it crosses an odd number of line segments, it’s outside – for convex polygons we can stop at one crossing)
• If the distance from the “center” (average of all points) to the collision point on the plane is less than the distance from the “center” to the test point, then the collision is happening outside the hull, and the force can be calculated based on the difference between the collision radius and the surface. Otherwise, it’s within the hull, and the force is based on the collision radius + the distance from the polygon.
• Well this is convenient – The Game Programming Wiki
• Looks like we can use Arvo’s algorithm. Going to see how this works. Discovered Arvo here at Gamasutra.
• Graphics Gems code repository
• www.geometrictools.com: source code for real-time computer graphics and physics, mathematics, geometry, numerical analysis, and image analysis
• http://www.realtimerendering.com/intersections.html

10:30 – 12:30 Interview

• Went well. The work would be entirely on site. Ah well.

1:30 – 5:00 – FP

• Got Arvo’s in. I can clean it up so that the collision test is only for solid shapes, which will allow me to get penetration values out of the collision. There was one tricky bit – I had to get the viewpoint matrix multiplies out of the collision calculations. So now there is a render() pass which draws, and a calculateCollisionPoints() that is called at some other time without any glBegin()/glEnd() blocks, and starting with an Identity matrix. I then pull the result of all the calculations off the matrix stack and save for the collision calculations. This does need to be cleaned up, but it’s not bad.
• Collisions!
• 

8:30 – 5:00 FP

• Wire up the force sliders
• Index finger is done
• Thumb finger is done
• Ring  finger is done
• Middle finger is done
• Pinky finger is done
• Added text within the 3D environment, partially to show Dong how it’s done, but also because it could be quite useful. It’s in the DrawableObject base class.
• Add collision detection
• Add force output display (Small GL windows, drawing only 2D?, slider?)

8:30 – 5:30  FP

• More on the HandControlUi and RightHand proper.
• Added a clock to calculate elapsed time between frames for physically-based modeling.
• Got the basics of the hand working, but I need to rework the eyepoint/model control. Adding a popup menu to select the mode.
• This is a great source for FLTK coding examples (Erco’s FLTK Cheat Page)
• Done with the popup menu
• Adding the matrix analysis to get the position  of the fingertips. Done for one finger. Need to add methods to store and access all the position information. May need orientation. If so then I could subtract the position from the current matrix, then multiply a vector by the resulting matrix.
• Might need this: http://www.cprogramming.com/tutorial/3d/rotationMatrices.html

8:30 – 2:30

• Well, I was hoping to use the MODELVIEW_Matrix to determine the orientation vector of the camera, but no luck. Time for classical spherical coordinate transforms. Transposing Y and Z, since OpenGl likes to have Y as the up vector.
• phi = angle from vertical, theta = angle from x axis
• z = cos(theta)*sin(phi)
• x = -sin(theta)*sin(phi)
• y = cos(theta)
• And that is now working. Note that we had to multiply the x component of the view vector by -1.
• Oooo – new shader book: http://www.packtpub.com/opengl-4-0-shading-language-cookbook/book
• Started on the HandControlUi fltk class.