math & physics
smokey97 at August 4th, 2003 00:15 — #1
hey, can someone explain ot me how todo world oriented billboarding, i read a tutorial on it, done everyhtign it said, but still didnt work... :wtf:
So i was wondering if someone could quicly explain the math of it all, or if oyu have to some sample code.... (explaining the math behind it all is prefereable)
davepermen at August 4th, 2003 09:17 — #2
uhm.. just from brain, for some ideas..
your billboard has a point, where it rotates around, and a "prefered up vector"..
now, you want the point-to-cam vector, simply to_cam = normalized(cam - midpoint).
next, you want a "right" vector, wich simply is perpendicular to your to_cam and the up vector, so you use a crossproduct.. to_right = normalized(cross(to_cam,up)) (or the other way around.. don't know from brain).
finally you want to adjust your up vector (but you let the original up vector still exist..), by crossing your current to_right vector with the to_cam vector..
to_up = cross(to_right,to_cam) (again, or the other way around.. )
with those three, you can transform all your billboard points of that specific billboard by these 3 vectors (they build a rotational 3x3 matrix.. look up orthogonal bases.. or so).
i hope that helps a bit..
unknownstranger at August 4th, 2003 09:45 — #3
I bet that within the next few hours at least one person comes in here and says something like "Hey cool davepermen, why don't you write an article about it?"...
davepermen at August 4th, 2003 09:52 — #4
because i only write snippets..
now, the question will be "why no snippet?"
answer: because right now i could not get arsed to write one..
julio at August 4th, 2003 11:46 — #5
check out the billboarding tutorial at nate.scuzzy.net. it's opengl, but the theory and math carries over.
smokey97 at August 4th, 2003 19:53 — #6
ok, well thats umm *checks code* (which i seem have to deleted sience it didnt work) *checks the tutorial he leaarned from* ok, i was doing exactly what you said, except your cross products are around hte wrong way (accordign to the billboarding tutorial on flipcode)
So either my matrix transformations are wrong (which they arent, i checked that) or my computer hates me. :lol: ...oh or i'm nto constructing the matrix correctly.
Anyway, i'll start again and if i have no luck i'll post here again with big capitol letters saying 'argh' (but alot longer )
P.S. i'm doing world oriented billboarding, so you use the up vector of the world, or in my case (due to the architecture of my engine) the particle system up vector; not the camera up vector
davepermen at August 5th, 2003 00:32 — #7
as long as you don't overwrite your original up vector..
davepermen at August 5th, 2003 00:35 — #8
possibly you'll have to transform by the transpose rotation matrix in the end (inverse rotation..)
i've set up some code for my raytracing cam yesterday and had to do exactly that..
davepermen at August 5th, 2003 02:21 — #9
would be helpful btw to see whats wrong.. demo, or pics, or what ever..
starboarder2001 at August 17th, 2003 20:08 — #10
I have been doing this so far...is there anything wrong with it??
It seems to work fine.
modelview = 1;
modelview = 1;
modelview = 1;
// -> all transformations here
// -> must do scaling here
// -> render here
baldurk at August 19th, 2003 15:08 — #11
if it works, why complain?
off-topic: this is my 666th post!
davepermen at August 19th, 2003 16:45 — #12
it depends on what you do.. it essencially sets the billboarding point, the one you rotate around, to the current 0,0,0 point of your world matrix. thats about it.. if you can life with that, fine