Anim8or Community

Hello again Cooldude.  We had such fun with the 3d calc last time, I thought I'd chime in again.  Actually, just wanted to say that if you google "line plane intersection code" you'll a few useful hits.  Easier than me typing it all out, anyway.

I may be a bit old school in terms of my gaming, but because collision detection can be quite computationally expensive there are a couple of tricks that tend to be employed to reduce the number of calculations.

Invisible bounding boxes around objects.  Where each detailed model has an invisible, crude model associated with it that is used for collision detection.  doing precise collision detection on a detailed 10k+ model is far more expensive than for a 500 poly blocky model, so it kinda makes sense.  You see this in games all the time, where you can see a bullet's tracer missing/hitting something very clearly, and the game reacting to it wrong - what you don't see is how the bullet is striking the bounding box.

When shooting through objects games like railings or chain link fences, games will often treat the whole object as a 2 poly collision object and roll randomly to see if the bullet goes through unimpeded.  E.g. the alien robot zombie is behind a metal railing, the bullet goes through railing space, then the enemy space, so at the railing space there is a 10% that the bullet stops. Rather than going through the expensive calculation of detecting the space of each rail individually.

Do an initial calc of which objects are even in the right area/direction and if they're not then don't even bother trying to do collision detection.  Just makes sense, but it does need to be done.

You probably know all of this already, but I thought I'd just add it just in case.  If you are having trouble with your frame count, then proper collision detection can have a pretty significant impact.

I've never seen this to actually do anything.  I think the window is there because Steve wanted to set it up for future functionality.  But if it works now, I'm clearly missing something.

I've poked around with it a bit.  It actually has my favourite pose editor, but I haven't found much actual use for it.  Not that I've had any significant progress in any 3d project for a while.  The poly count tends to be outrageous with their models, I think, because of the way that they overuse morphs for customization.  As for cheating, remember the adage:  "Plagiarism is the highest form of flattery."

Yes, I think I know what is going on there.  I sort of took the calculation of the yaw for granted as it's pretty straightforward on paper, but in computer speak there are a few cases that you need to take care with.  So I have a new formula for you to try out to get your yaw angle that should take care of everything.  I couldn't get it all into one calculation, but this is the most graceful I could get it:

if dx=0 and dz=0 then
   yaw is unchanged from its last orientation
else
   yaw=sin^-1(dx/sqrt(dx^2+dz^2))
if dz<0 then
   yaw=(pi/2)-yaw

The first part is to take care of situations where the yaw component is non-existent:  straight up, straight down, or not moving.  This will require your keeping track of the last position, or building up functions for special cases.  Movement along the y-axis is problematic, so one way to deal with it is to keep anything from being able to move to an angle which is perfectly perpendicular to the ground -> pitch max of 89 degrees.  Otherwise, if pitch exceeds 90 degrees you'll need to invert the yaw 180 degrees and roll the object around its z-axis 180 to get the pitch below 90 again - if you follow.  Best to just not let them get there.

The second part is the best way to calculate the angle without running the risk of getting a DIV0! error at the x or z axis.

The last part is what takes you from a 180 degree angle to a 360 degree one - except in radians, of course.  I couldn't see a way to integrate it into the other calc, so I think you may have to break it out into an if statement as I have done.

edit:  forgot to mention that this will calculate the angle as increasing while rotating clockwise.  If you want it to increase counter-clockwise then replace dx with -dx in the formula.

Without going into too much trig, I can say that your second formula works out to be the same as the one I quoted - with the obvious exceptions that you've scaled it from a normal vector to a vector of length 2.5 and you are using a different base orientation.  That one is just fine and shouldn't be causing you any problems.

In your first set, where you are converting to radians, your rady2 formula is ambiguously parsed.  Depending on how your code is being assembled the value could be (pitch/180)*pi or pitch/(180*pi), so I'd recommend a pair of brackets to be sure - with the first one being correct.  That difference can affect your vector length, so maybe start there and see if it helps.

ps  currently I'm working on a Mongolian helmet, a Lithuanian helmet, and a pair of gauntlets.

I've studied 3d calc and 3d physics a few years back, so I can probably dig up an answer for you, but I think I will need to know more details about what you are trying to do and what numbers you want to end up with.  The problem could be a few things.  Are you using lines or vectors?  Is distance important or just relative orientation?  The world, the camera, and the objects all have their own coordinate system, so which angle is being compared to which?  E.G. "line between camera and its focal point in relation to the world's z-axis."  And of course, a Google search for "angle between two 3d lines/vectors"  will tell you more than I ever could on the subject.

I'm running 7-64bit at home.  Let me know if you want me to test something out for you.

Or you could just go to www [dot] darkwoodarmory [dot] com or www [dot] deltin [dot] it .  Oh, and if you are going to do it by hand (which, being an amateur sword/armour maker myself, I don't really recommend) I'd at least use an angle grinder rather than a jigsaw - they're relatively cheap and would save you an awful lot of time.  Hilts are tricky unless you have some rudimentary welding skill.  Plain crossbars can be fairly easy, but if you are planning any sort of swept hilt then I hope you know a good blacksmith.  Regardless, it is a lot of fun but takes an awful lot of time.  There is reason why I don't have the time to be more involved in 3d animation .