PDA

View Full Version : radio wave bouncing



Bill Nelson
12-13-2004, 04:16 PM
Okay kids, here's a good one... my client wants to show the effect of multipath radio waves in an urban environment. concentric rings emanating out of a transmitter, bouncing off high building walls in a 4 or six block urban "God's Eye" view, with the receiver being hit by many of the reflected waves down several blocks and around a corner. Any ideas? Could be a 2-D solution, but the 3-D could be pretty effective.

prospector
12-13-2004, 09:15 PM
How about particle ring?

by making a disk say .0005 in thickness and just round enuf to encircle your tower, and making it many sided so that the particles come from many vertices but only 1 particle from each, you would end up with an ever expanding ring that would be set to collide with buildings and bounce off of them into others untill they either 'die' or make it thru the city into the countryside.
So you could have a disk of say 500 sides (which would only spit out 500 particles, so it should run reasonably fast), set to go off every 10 frames or so depending on the freq you are transmitting.
you could also simulate light the same way by moving the disk up and down to simulate the 'wave' in sound or lightwaves.
And if you want to see them hit a reciever, then follow the paths a few times (from top view) to see where they seem to gather after all the bouncing and build your reciever there and the signals would arrive at different times showing the effect of (echoing,as we called it int the 'OLD' days of aluminum antennas).
I might even go 2 disks for the emmiter to make a more solid wave with some sprite HVs starting very small and gradually increasing in size and changing color (red or blue shifting),and dissolve the disks so they aren't seen.

just a thought off the top of my head.

toby
12-13-2004, 10:01 PM
you could use the ripple procedural texture, set to one 'source' on a large flat transparent polygon, clone the polygon and place the center of it's ripple in places where you want it to bounce - (if you need more control make a ripple texture map instead of using the procedural) then ray-trace refraction. I'd start by plotting out exactly where and when I want each ripple to eminate

No wait - I'd start by seeing how it looks in the render :eek:

Bill Nelson
12-13-2004, 10:46 PM
some cool responses! I've used Strata since the early 90's, and only last year jumped into LW. Not up on the potential. I'm running LW on my Powerbook (7.5c on my 800Mz) Will those hardware/software limitations, do the collision suggestions still hold? Can 7.5c do the collision calcs?

prospector
12-14-2004, 02:10 AM
yep..7.5 ok for that

good idea Toby, cept I see 1 problem unless you have done it cuz I haven't tried that way.

If you use either way you described, and the 'wave' hit a building at an angle, would the 'wave' bounce back off building in same fasion as it hit it?
In other words, if the leading edge hit the building at the left corner and moved down to the right corner (like a wave on a beach), would the reflected wave follow the 'hit' or would it just eminate from the center of the flat poly?
But here's what I thought was wrong when I thought of that first...if you make a poly for each building, would you not have to cut it in whatever direction the wave would bounce so that the poly wouldn't show behind any other buildings?
Because the wave would follow the poly to the end without falloff and ya can't make falloff in the shape of a city as far as I know. So somewhere it would seem, the wave would go thru a building as it spread off into the distance.

Just tryin ta go thru that in my mind, as small as it is ;)
But then the physics may not be THAT important, and just a representation is ok. In which case your way is faster.

prospector
12-14-2004, 02:32 AM
OHH OHHH (sounding like Horseshak) ;)

Just thought of the fastest and most reliable way to get the correct look.

In modeler build yer buildings and put into background, or use a painting as a background.

Now make a poly long and skinny (like a flat plane as toby mentioned) and subdivide it a few times (like 3, then knife the short way between each of those polys so they are doubled).
Now while it is straight map it with a UV map in the direction looking down on it (depending on how you model it).
Now make a 'NEW' Endomorph, and bend and rotate it thru the city as you would expect it to bounce off walls (ie; it hits wall at 45 deg then the bounce is 45 degree) and continue till you want to stop or as long as you made your poly (which you would probably have to remake a few times to get the right length).

Now in Layout, open the morph mixer, load the morph to 100%.
Now when you add texture, it will follow the poly exactly along it, bouncing with the buildings.
And if you really want realistic bounces then you would have to rotate the poly 180 where it hit the buildings (like folding a dollar bill to look like a 'V'), give the texture double sided texturing, give it transparency where the color bands are (clip mapping I think tho it's called something different in procedruals) and you will then see the wave hit a building (going along the poly under the fold would be seen too), bounce off the building and the leading edge will bounce back just as physics dictate, with the rest of the wave following the poly untill it hit the 180 fold and 'bounced back to the new direction.

It would probably take longer to read this than to do it. :D

prospector
12-14-2004, 03:32 AM
Just did it to see if I was correct.
I am

Follow pics

prospector
12-14-2004, 03:39 AM
And finally.
Also the object, scenefile and anim
from start to finish took 20 min (including anim render).
If you look closely at the 180 bends you will see that the Waves DO corectly bounce off walls as physics would dictate.
You could build your city as you see fit or just use these as you desire.
Hope that helps

prospector
12-14-2004, 03:45 AM
Also I didn't UV that wave poly as I did it in procedruals instead of pics.
Also I didn't do the knifing between polys.
And as a matter of fact, ya don't even have to subdivide it the way I DID do, you could just do it as 1 long poly then do the cuts and bends.
But then it would be sooo simple that anyone could do it, and the pay for 3D workers would go down the tubes :D :D :rolleyes:

Ya just gotta make it SOUND techie

Darth Mole
12-14-2004, 04:01 AM
There's a couple of plug-ins for After Effects that will do this simply and easily. You can either use radio waves - which propogates concentric circles (or whatever) - or Wave World - which creates 'real' wave patterns based on the movement of a fluid. The thing with Wave World is that it properly reflects from 'objects' in the scene (in this case a greyscale bitmap which is held in another layer).

I'd render your overhead view in LW, and then draw a simple map (I figure solid blocks would suffice to represent each building) and then run Wave world, either over the top or separately and then comp 'em together.

Assuming you have access to After Effects, of course (if not, get in touch - I quite fancy having a go!). I reckon this would take a few minutes to set up and then another few to render. I think by going the full LW route you could tweaking for ages...

prospector
12-14-2004, 05:49 AM
And here is the UV textured one WITH Reciever dish ;)
Following the above along you can backengineer this one easily enough.
Tho ya might wanna slow the wave speed down some.
Which I made as a background avi (just background) as a wave texture with 1 source about 8 meters to the left (-8m I think).
But you can play with it some and get the feel for it.
The last knife cut (at the reciever) was straight across and then I folded the last section over on itself, moved the 'Cut' points to match the dish face, and just sized the end of poly points to fit inside the recieving unit.
As you can see NOW the texture hits the dish....bounces back...and gets smaller (as it should after hitting a parabolic dish) and is absorbed into the recieving head.
Could not do this part without UVing the poly first as the other way just followed the poly straight to the dish and no bounce back.
Also you can see really clearly how thw 'wave' bounces off the walls in a more orderly fasion. (move slider 1 frame at a time).
Anyways...have a go at it if you like or use it if you like.

prospector
12-14-2004, 05:59 AM
Darth..
Will it bounce the waves off the walls in a way they should bounce? Leading edge first following along the wave to the end?
And will it make the wave bounce back from the dish to the head on reciever? And resize as it does?

That scene with the UVing took 30 min to build and render. Both scene and backdrop waves to use as the texture.
Not really a lot of tweaking except for wave speed.
could probably find a better made (4 min for me) dish somewhere on the net.
Maby more time in the building of the structures would look nice, but for just a demo on radio waves, this crude example should suffice.
If it were a paying job then I would take some more time (put arrows and jitter the wave poly some, split it up where it would hit the corner of a building so it deflects both ways, give it some falloff so wave will eventually dissipate.) stuff like that.

Darth Mole
12-14-2004, 09:37 AM
Now I see what you're doing, no it's probably not quite right. All the reflections with Wave World are 'real' - so it propogates out from a central point and are bounced around accordingly. I'm at work at the moment, I'll try it out later...

Bill Nelson
12-14-2004, 10:49 AM
Gentlemen, thank you very much. I applaud you! Very novel and efficient use of the software! As Darth suggests (yes Darth, I've seen the 'Radio Waves' effect... can't recall if that runs on AE 4.1 or not) and that's the omni-direction I'm going. It is an omni transmitter (hand-held, actually) that's attached to a portable video transmitter for use in urban search and rescue. The receivers are positioned at quite a distance, and the multipath signal that results from such reflections cause a problem. My client furnishes 3 antennae for the receiver set up about 60 ft. apart that result in a diversity reception that is quite acceptable.
The waves are really cocentric circles, like a true radio wave. The bouncing around the buildings would be more fluid-esque, I imagine. Would your suggestions serve in such a situation?

Bill Nelson
12-16-2004, 12:47 PM
prospector, others...

Any further suggestions on this model? the expanding rings need to realistically bounce from bldg to bldg, from the transmitter to the receiver. Of the ideas mentioned above, which would be most effective? Or is there a better way? Thanks in advance, and thank you much for your detailed responses above!

exit6
12-17-2004, 07:34 AM
Make sure you post the results -- I'd love to se how this comes out.

Darth Mole
12-17-2004, 09:34 AM
Hmph. well I tried Wave World and it was rubbish. You get strange reflections from EVERYTHING right from the start (even if they're not moving, weirdly). It's much more attuned to creating fluids, caustics, stuff like that.

You could use the Radio Wave plug-in to create the waves, but making them bounce would need to be hand animated... I think. Hmmm... Although you could then use the buildings as sources...

I'll have another go.

prospector
12-17-2004, 01:19 PM
Here is with particles

Now a word about the pic
Point 1...this shows slight delay in secondary signal which causes double echo effect.

Point 2...Multiple signals, multiple times, will cause mostly unrecognisable audio or video signals (I lived in a place like this and no matter how I adjusted those stupid rabbit ear antennas, alls I got was a slight pic in lots of snow)

Point 3... good signal but delayed overall because of signal bounce.

Point 4... Good show of 'RADAR' effect..wave hits bldg and bounces straight back, time delay gives distance.

Point 5...Again slight delay, will cause triple echo effect.

Point 6...Totally clear signal, has sports score before guy at #3 and can then bet final outcome and win money :D Well only a few milliseconds advance notice, but with a quick dial phone......... :rolleyes:

Oh yea..anim is in WMF as avi was 5 megs :mad:

Bill Nelson
12-17-2004, 01:28 PM
prospector, looks cool! If you'd like to zap me the wvm file directly, fire it off to

[email protected]

I've got this beefy commercial cable modem setup, won't take a moment to download. If your upload speed is problematic, I'll understand... this is very similar to what I've been thinking, I think. The animation would cement that. Can the lines themselves be tightend up?

Bill Nelson
12-17-2004, 01:30 PM
never mind the re-send. just opened the small AVI, and that's what I was talking about! cool job and a great visualization of what we're talking about! Best of the Christmas season to you...

exit6
12-17-2004, 06:03 PM
Nicely done, prospector.

prospector
12-17-2004, 08:11 PM
Strange thing I discovered, (probably everyone already knew but me)

when particles fire off at 1 frame (like the rings firing), there is a slight ramping(?) in the firing order and they don't go off in the same frame as you would expect.
if you look at the example scene and move frame by frame (with layer 2 selected on bottom), you will see that 1 poly shoots particle first and it continues around to the last poly.
And this leaves a slight offset ring lineup

I can't figure how to stop it from doing that, as I have it set to fire at a particular frame.
Tho with the right settings it looks like a spirograph. And supposedly they all fly at the same frame.


Hmmm
Any theories? conspiracies? Unsubstantiated half truths? Remedies?

exit6
12-17-2004, 09:08 PM
I can't help you with the ramp thing, and I don't know about the dead wood thing either. 4 man job, three men? Seems like you're still ahead to me . . . although an idiot like myself would take the job and do it myself, and end up pulling three consecutive all nighters and just barely make the deadline, then sleep for 14 hours straight. But that's something else. Actually that was yesterday.
Anyway, it looks like you've got that animation good enough for prime time notwithstanding the ramp anomaly, but how would you surface those particles? I suspect the client would want a nice, smooth 2D looking circle (although you could wow them with a cool hvox thing). Also, how would you get those particles to look like contiguous rings? Shouldn't each reflection point look like a new source of cocentric rings?

toby
12-17-2004, 10:21 PM
Particles aren't very cooperative - I once tried to get 1 particle per 3 frames and it was impossible - until I found out about the 'Key Effect' setting. Set it to 'key', then put a keyframe where you want it to fire...

prospector
12-18-2004, 01:59 AM
exit6...yea HVs would be good but I used Sprites as they render faster for this test.
To get a more solid ring you'd have to fire off more particles which means setting your ring to more than 500 polys (which is what there is in this example), a 1000 poly ring would give you a more solid setting. and if you fire off particles for 3 consecutive 3 frame incraments (say frames 10,12,14) then set frames 10 and 14 at 50% dissolve on the envelope, you would get a ring that started as a dissolve in to the second ring at 0% dissolve and back to a dissolve out ring and make particles in second firing large enough to reach the 1st and 3rd ring.

And every simulation or piece of film of any bouncing of waves or such on the sides of boats or walls or any stationary object, only reflect or bounces back what actually hits it. But it does so in an orderly way (supposing the object hit is perfectly smooth). The best example is #4 in the above pic. As ring hits, it is reflected back just as it was hit, so you end up with a semi circle going just the opposite way.
A round building (like the Capitol Records Bldg in LA) wiuld give you a semi circular return bounce with each particle bouncing off in the exact angle it was hit.


Toby..yep, that's what I used.