View Full Version : FXWars! Car Crash : mJR (not sure yet)

05 May 2005, 04:46 AM
well I missed the last 2, one cause I was working on a short (which I never finished in time for a film festival) and another because of a trip out of town. so this time I'm in, if I can finish what I want to do.

this is something I really enjoy messing around with. I've done a little work like this on a past job an XBOX game called Hunter Reedemer. for all 2 of you that have played it, when you smash something on screen, I smashed it in MAX first... but nothing on this level of detail required of this challenge, most of that was just an unbroken model and broken model with little or no animation inbetween.

I feel a bit behind, all you guys have already started on cool cars and sets, while I'm still in the process of figuring out how to smash and rip metal.

I have a few test examples of what I have been R&Ding.

I've read through the instructions a few times and haven't quite figured out what is considered what? what would be a simulation and what would be hard sience and so on. because off the top of my head I don't know of any sofware, such as LW, MAX, MAYA, or SI that supports metal simulations. such that included settings for metal strength and density, stress points, welds, bolts, rivots, fracturing, bending, warping, friction and heat damage, and anything else that goes along with metals. (I'm no scientist) and I don't have any real programing skills to develop a pluing for MAX that can simulate that. (if only I could get my bro-in-law to code a metal plugin we would be set, but he's busy making a muscle plugin for MAYA. (

so after exhausting every single step process I could think of from particles, to cloth, dynamics, to this and that over the last few days (many sleepless nights) I started to combine a bit of this and a bit of that and worked out a multi level process of simulating weak points in the metal, then the metal skin, then the impact area using dynamics, springs, fractures, hard bodies, cloth, morphs, and then extra dynamic simulation on top of that. trying my best to get as much actual damage to be done with dynamics and not by hand. I call it Brute Force Animation.. old school style

well CC are welcome, I'm going to in the next day attempt to figure out a process for solid metal pieces like a car frame and work on combining multiple setups for things such as a bumper, hood piece, and a part of a fender. then after that, depending on time will determing if I go to an extreme like a high speed crash or do a smaller crash like a fender bender and composiste that into some live action footage.

anyway, there are some avis here so check them out

thanks and good luck to everyone, and most important, just have fun learning and doing.....

05 May 2005, 05:19 AM
hey michael i have just checked ur test that looks awsome.what software do u use? 3ds max?

05 May 2005, 05:22 AM
thanks, yes, 3dsMAX 7

05 May 2005, 05:30 AM
thanks, yes, 3dsMAX 7

very well done test!!:thumbsup:
i must keep my eyes on this thread.:scream:

05 May 2005, 09:29 AM
I think those tests look just amazing. Great job and i really gonna keep my eye on this, and of course other entries too. Too bad i don't have own computer now which i could use to participate to this challenge. So anyway michaeljr, any change you could reveal your secrets how you did those great effects after this competition is done ? :)

05 May 2005, 09:40 AM
I think those tests look just amazing. Great job and i really gonna keep my eye on this, and of course other entries too. Too bad i don't have own computer now which i could use to participate to this challenge. So anyway michaeljr, any change you could reveal your secrets how you did those great effects after this competition is done ? :)

i personaly sent pm to him.but he haven't seen and checked pm i m asking did u done those test with 3ds max7? it will gr8 to know more.:)

05 May 2005, 05:18 PM
I hope it doesn't make me a jerk, but I don't want to give away to much of my, I guess secrets, before the competition is over. If I can get a bit of an edge against those who can model an awesome car, which I have done very little of, I need to hold on to a few thing. After the challenge is over I will have no problem putting together some tutorials on what I'm doing.

for now I will give you a few hints without getting into details.

1st everything I'm doing is all within the ability of anyone that owns software that supports rigidbody dymanics and cloth. there are free cloth systems out there for most programs. I'm not programing or scripting anything, just using tools I have already.

2nd I'll save everyone some time by telling them that softbody and cloth alone will NOT produce crumpled metal because there is no way to tell those things that you wan the object to FREEZE or BECOME SOLID at this momemt of time. A softbody will try to return to it's original shape and a cloth will continue to move with gravity and motion. if the cloth had a STARCH setting, then that would work. if you could say this cloth is solid until hit and then it deforms, but when this much energy is dispersed, this it freeze again. but that will not duplicate stress and most important, a cars pre determined weak points and crumple zones. a cloth will just turn into a piece of cloth instantly all over.

3rd think out side of the box. since there is no METAL button, you have to think of ways to simulate what happens to the metal. think of the metal like a character. at least that's what I"m doing. so I'm rigging the character (metal skin) and giving it limits on what it can and cannot do, then simulating within those limits to move it around.

and 4th since there is no one button solution, sometimes it may take more than one object to smash something to look like one object caused it. think about those cars in Jurassic Park that smash themselves and make it look like the T-Rex is hitting it. well there is no T-Rex, it's CG, but the car smashes it self by using wires and things to pull the car into itself.

some good movies to look at and get info would be the 4 disc Pearl Harbor, Star Wars EP1 making of disc, oh there is a movie by Stallone (forgot the name) where he races F1 cars, crappy movie, awesome wreck scenes done in MAX. and another movie, can't think of the name right now, where the kids beat death and death hunts them down, the second movie of the series has a huge massive crash. Matrox 2 crash scene, but that's hyper real slow mo.

hope this helps a bit. maybe I should post this in the tips tricks area also

oh and START SMALL, sure it may not be as cool as smashing up your nice, cool, high res car model, but if you just start right off, you are going to be pulling your hair out in 5 minutes when it looks like crap. take a bumper off and figure that out first, then the hood, that your time, start small, think small, write things down, do it like the big boys do, start with simple shapes and objects, get the process worked out first before blowing time on the high res stuff. it's a whole lot faster and easier to recalculate dynamics on a dozen simple box objects than it is a 10,000 poly car.

05 May 2005, 06:09 PM
Heh of course it doesnt make you a jerk. Atleast in my opinion. That's why i asked if you could reveal something after this competition :)
Thanks for those hints, gotta write them up and start testing soon as i get new hardware.
Good luck for competition! :)

05 May 2005, 09:04 PM
Hi there Michal.. I saw your animations test and look great...impresive

Is my first time making real stuf with reactor .. I dont know If I can make something but I have learned a lot with this contest ... but you look like you have manage this for a time ...

Well with your tips I will make some new test and wait what happend

05 May 2005, 09:51 PM
thanks, ya, the ticket with Reactor is to start small and use the lowest poly objects that get the job done, then build up and up, but remember to TURN OFF simulations that are already done, if you don't, it will recalculate all your Reactors.. thus probably messing everything up.

I've been messing with built in dynamics since they first came out in MAX. but back then, before Reactor, it was a major strugle. it was hard just to get a box, to stay on top of another box, no matter how hard you tried, most of the time it would eventually work it's way through each other and keep on falling!!!

I just wish I knew math and scripting, then could just write some custom extensions to use with Reactor.

05 May 2005, 07:22 PM
for kicks and giggles I was getting bored with the TOY CAR and Wheel constraint in Reactor so I attempted, crapoly, to build a simple suspension.

it doesn't handle torque very well, goes all sorts of crazy when power is applied, so I need to stick my head under my car today and get a little better idea how things are setup, probably need some more mass to the parts and tie rods might help. right now the ground is moving up and down, so it's not supporting the weight of the car, but I"m going to try to figure out if it's possible to copy and dublicate these setups, cause it's a pain to setup over and over and over.

but it's time to mow the rest of the yard.

05 May 2005, 10:13 PM
i have send you a PM .

06 June 2005, 04:21 AM
ok well I got a little PM today from bRuNuShky wondering if I was sick or just dead :) cause I haven't made any updates, thanks for caring, but no I am still here. have been very busy in the real world. I teach a church related Boy Scout type group and we had a regional camp out with some 500 kids and that sucked up a whole weekend plus all the prep time. then since the weather turned nice I have been re landscaping the yard a lot. tons of bricks (and I mean a ton of bricks) and loads of dirt and mulch with some sweet flowers. here is a pic of the lights I just put in last night, those freaking blood sucking insects turned my back into a pin cushion.

but here and there I have been working on some stuff and watching closely everyone else progress. I'm really impressed with some of the models, sets, and renderings of near photo real stuff. so I have decided, since I just don't have time to model and light a "real" scene, that I'm sticking to doing more physics in my wreck and not worring about the other stuff. I'm hoping to have a near 90% physics based wreck.

since it's so close to the end and I may or may not get done, I figured I can leak a bit more about what I am trying to do since I don't think, not that I have noticed yet, anyone else attempting some of this stuff. I have been doing some research on what is involved in a car wreck and how the car deforms to protect the occupents. So I have done some test with crumple zones controlling the deformation of the cars surface. also weight transfers such of that of the engine and transmission and how it, depending on the weight ratio, controls the spin and flipping of the car. I also have been testing a more realistic suspension system that actually behaves like a real car, not the Toy Car setup of REACTOR in 3dsMAX. this suspension will drop when the car moves to far away from the ground and also will PUSH back after the intial impact of the car. also steering plays a big part in impact control. after viewing some crash footage from under a car I saw how the steering system crumples and the tires actually take a lot of the energy out of the crash.

so I found a pre made VW Golf that I have been cleaning up and modifing to use. I picked that car cause it was already modeled, but also I can easliy find photos and info about it's size, weight, and performance. also this model matches the piece of crash footage I am attempting to duplicate, hopefully 90% of which will all be physics based, no keyframe animation, no speed up or downs in post, etc. the car in the crash footage was about the same size of the VW Golf. this will allow me to reverse engineer the crash to get the exact speed of the car and then the weight of the car, since I do not know how much the doors, or frame, etc. etc. weight in the car. I can only guess with what car knowledge I have, on how much it weights, then calcualte and change until I can get close to what the wreck looks like. then if time permits, I will alter the wreck in some way, placing say an extra object in it's way, or speed up the car, so that the wreck looks different enough from the crash footage so that no one can think I just keyframed over the video footage.

so next will be building and rigging the frame, engine, transmission, suspension, exhaust, doors, truck, skin control, bumpers and anything else that is directly and directly effects the inital crash and energy transfers. then things like glass, plastic parts, dust and debris, which does not effect the wreck will come next.

so that's where I am at.

06 June 2005, 06:45 PM
hey mJR,
gotta say ur tests are looking the closest in this challenge.
there must going a lot of thinking in that mind.
looking forward for the car crash results with ur custom suspensions
and metal responses.
good luck.........

06 June 2005, 07:34 PM
weeeee ha, finally some time to sit down and work. no more church picnics, no more brick work in the yard, no more planting flowers in the 95 degree 90% humitity weather.

first, a round of applause for the deadline, I mean turning of the head extension... (clap, clap, clap) couldn't have come at a better time.

right now I"m still in the process of building a car frame and the main impact controllers. here is the car model I have chosen to use. it's a pre built model I had with a lot of work done to it. probably could have modeled one from scratch with the amount of work I have done taking it from 10,000s of polys and 300 objects down to a managable 15,000 polys and 100 objects.

it has been broken into color coded zones. each zone is a part of the car that is a seperate unit. each zone will have it's own dynamic simulation based off a whole car simulation (if my computer doesn't hose itself). each zone will also have it's own weight, per my best guess of what those parts weight, and will be connected to the zones next to them for weight transfer and deformation. like say if the wheel breaks at the axis and pushs into the wheel well it should deform the fender. or if the bumper rips and pushes into the engine which then pushes into the passenger area.

we shall see, more images to follow hopefull tonight, but I"m exhaust from playing 4 Square and Kickball....

06 June 2005, 07:32 AM
ok well I think I have the frame worked out. one big guessing game on a 3 door VW Golf's frame. also I put in an engine block just for weight transfer and I forgot, but soon to add a gas tank. there is a lot of weight in a gas tank.

next will be to ad the skin controllers and crumple zones. and depending on my craziness frames to the hood, doors, and trunk lid. not sure if that is going to be to many sets of dynamic system for my computer though.

06 June 2005, 05:10 PM
Nice to see you again in the chalenge ... this car "settings" or "rigid" or like you want call it looks good... I cna wait to see this car turn in a mess of metal... :)

07 July 2005, 07:14 AM
ok, well I'm a bit upset at myself for not reading the directions a little better and unfortuantly I have wasted a ton of time in my efforts to do something way beyond what can be done in this short amount of time. so instead of doing as much as 90% with dynamics, I"m going to do a 50 / 50 and make it look as good as possible. I will explain later when I have time what I was attempting to do and why it just wasn't going to work.

so most of my time wasted, I"m busting my butt to get this done in the next 3 days from almost scratch. the last few days have been spent fixing up this model. for a pre made model it only consisted of the skin and basic interior. so everytime I opened a door or look at an angle, something was missing or had a hole in it.

so I threw together a basic engine and engine bay, undercarrage, basic suspension, exhaust, and some door edges. I don't think during the wreck you will be able to see much of anything else. based as close to the original for the 2 off angled photos I have.

here is a pic of the extras I have added. i"m now modeling a low poly control object and I hope to get some animation done tomorrow and the next day after I rig it up. I hope to get it done fast enough to be able to texture and light it more than just gray. but the WRECK is my primary concern.

07 July 2005, 01:34 PM
First I would like to thank Samuel Kvaalen for coding this voting page and hosting it.

Go >>HERE<< ( and post your infromation if you want your animation to be part of the voting on Monday night.


07 July 2005, 10:07 PM
oh crap, delima time, what to do. well thanks I have been busting my butt to get it done, but the storm is interupting the internet, so I'm not sure to push my luck and finish up a bit, then render, or just post a test I have done. I'm going to push it and try to fix some of the things, no fx and just post what I have done.

07 July 2005, 06:04 AM
well I got it finished, well enough that is what it is and posted it right at midnight, talk about a close call. thankgoodness the internet came back on. I just uploaded a cleaner version with my logo in it. I'm also rendering another version to post here with some more camera angles. I ran out of time to render it with motion blur or better lighting.

I can only blame myself for not reading the directions closer and realizing that it wasn't required for a "realistic" crash to be all done with dynamics. my attempts and R&D wasted a lot of valuable time while everyone else was making cool models and sets. I only came to the conclusion that the built in dynamics engines are only good when smashing against a solid wall. when it comes to compressable objects like that of dirt or things that break up, they just don't cut it.

So I hope my entry will be taken for the animation and not a finished shot.

Anyway, all was done in 3dsMAX 7. The basic model was a one I had from a CD somewhere. It was heavly modified, interior parts, engine, undercarrage, suspension parts, brakes, basically anything that showed through when the car was flipping end over end. did the extra modeling in a day. The primary wreck was done by hand based off old car crash footage I found. That car was an older 1970s steel bumper car, so I changed it a bit and let it go by letting the doors come unlocked. little more drama and more plastic.

the reason it was done by hand is that Reactor and other basic dynamics systems can't do compressable object. by that I mean something like say sand, you step in sand, you compact it down, it stays down, but it's still the same amount of sand. in my wreck the car digs into the ground, which is why it flips it's self around. I couldn't find a way to recreate that in dynamics other than tie dozen of small boxes together with 100s of springs and shocks and then smash the car into them and hope that they don't come back together. so needless to say, I wasted a ton of time, a ton of time only to get to the point I couldn't even move the screen around. if I knew programming or had a fluid dynamics simulator then maybe I could have figured it out.

all the animation was done over the last 4 days or so (I wasted a month on R&D and yard work :( ). secondary door animations where done with dynamics. dents and rips were done by a mixed of hand and cloth, but not done on the model while in motion. I take a copy of the object to a area with pre setup boxes and shapes, setup the basic motion of the impact area, then drop or push into the metal (cloth) to get some deformation into the shape I want/needed. then use those as morph targets and go back into the animation and morph them at the time of impact. doing it on the model just isn't controlled enough, cloth wants to move in one direction, but in a controlled area it's easy to "mold" the shape you want. with more time I want to get it to look more like dents and edges of crumpled metal and not something like waded up tin foil.

car frame animation was done by hand vertex animation and then I would put some other animation on the body panels to stay close to the frame. third layer of animation was done with springs, flex, and hand animation. things like the exhaust and engine moving, metal edges moving, etc. also more dynamics on the shocks and springs (not working ones) representative ones.

I was hoping to have a "real" suspension I did R&D on that would work much better than the basic Wheel Solver. something that PUSHES the axle down when the car comes off the ground and not let it come back up till weight is put on it. I had to hand fix those issues with the Wheel Solver. so I may put that back in there.

the next layer of animation was going to be solid fractures, things like hard plastic parts scattering on the ground, headlights and taillights, licenses plates, or trim coming off. but the calculations of dozens and dozens of objects was taking to much time so I cut that to finish.

next layer would then be the ground chunks and smaller framents, particles, such as dust and shattered glass.

then some shaders and HDRI rendering, little sound effects, that will finish it up.

anyway, I look forward to the next one and this time I will most definately read the directions better.

good luck and if anyone has any questions, I will be glad to share now that the competition is over. maybe if I get some time I will do some tutorials....

time for a shower and some sleep.......

CGTalk Moderation
07 July 2005, 06:04 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.