Jump to content
TheUsualAlex

crowd simulation

Recommended Posts

Has anyone done much crowd simulation in Houdini for movie fx that could be comparable to LotR?

Ermm... Never mind about this post.

Share this post


Link to post
Share on other sites
Has anyone done much crowd simulation in Houdini for movie fx that could be comparable to LotR?

Ermm... Never mind about this post.

The crowd simulation in LotR was done via a custom program, Massive, that operated via a special artificial intelligence network system (applied to all character entities). I don't think Houdini can handle this, unless you make a similair-like system yourself (perhaps as a standalone program outside Houdini).

Though, you can achieve realistic crowd simulation in different ways. I'd imagine AI-controlled crowds can be pretty unpredictable...

Share this post


Link to post
Share on other sites

I'm pretty sure I saw a presentation where Rythm & Hues talked about animating crowds of mice for cats & dogs using houdini. I think they wrote some custom ops for houdini... Something about the interact sop not being cool enough... their test footage looked really cool, where they had two crowds(well, triangles) run into each other, and they did a nice job avoiding objects and each other. Not quite as much action as LOTR, but they had the whole field of vision thing going, and actions based on certain situations.

I think at least... It's been a while. My mind has since turned to strawberry pudding.

Mmm... pudding.

Jens

Share this post


Link to post
Share on other sites

u'r right...the interact POP is not cool enough...

a basic flocking system should have attributes for cohesion,allignment and seperation behaviour ...combined with field of vision

the interact POP does not have any attributes for allignment behaviour and field of vision...thus its lacking one of the most important features of a basic flocking system.

i think the only way u can get it done is by writing your own SOP or POP

(one of my current ongoing projects...I'll keep u guys posted on the progress)

Share this post


Link to post
Share on other sites

right jens, also, for sum of all fears caleb howard did some impressive AI sims for the stadium scene. correct me if im wrong, but i think he was also responsible for the meeses as well :)

so ya, with some serious know how, you can make crowd simulations!!

Share this post


Link to post
Share on other sites

hi mir

are you using VOPs to make a spiced up interactPOP? does VOPs give you the ability to build these missing features that you mention, ie; allingment behaviour, field of vision etc...?

i'm pretty caught up with things at the moment but slowly getting into POPs when given the chance and haven't had time to even look at POPs with VOPs ( though i have done the VOPs tutorials for shaders which was great )

Share this post


Link to post
Share on other sites

hey betty...

yeah im using VOPs to create a spicedup interactSOP...im making a SOP cuz it offers u more functions than a POP ( for eg. the import function is only applicable in SOPs)...VOPs or VEX does give u the ability to come up with these missing features.

right now i've managed to come up with this flock_SOP that implements the field of vision and allignment behaviour...as soon as i get it down perfect i'll start working on the cohesion and seperation behaviour...and after that object avoidance and avoiding a moving particle (like small fish swimming away from a shark)....so lets see wat happens!!

Share this post


Link to post
Share on other sites

nevermind seeing! it already sounds pretty cool! :)

if it has the "Ong" seal it can only be good :P he he he

thx Steven, one more thing to scratch off the "would-love-to-but-REALLY-haven't got time" list

Share this post


Link to post
Share on other sites
nevermind seeing! it already sounds pretty cool! :)

if it has the "Ong" seal it can only be good :P he he he

thx Steven, one more thing to scratch off the "would-love-to-but-REALLY-haven't got time"  list

Hmmm... when did I give the "Ong" seal? :huh:

Did you post to the correct thread, sirogi? ;)

Share this post


Link to post
Share on other sites
...im making a SOP cuz it offers u more functions than a POP ( for eg. the import function is only applicable in SOPs)...

I've been writing some vex pops recently, and the lack of import or even the ability to have 2 inputs in a vex pop have been really friggin annoying and dissapointing. I guess the solution is to write a sop, but then you're basicly writing the whole particle behaviour, and I was hoping I wouldn't have to, because I'm still using other parts of pops....

Oh well...

I think pops is slowly dying...

Vops is almost better for creating particle behaviour. It gives you the controll you never had in pops.

Zig.

Share this post


Link to post
Share on other sites

u can still use other parts of POPs with your vex SOP...

u can output your pop into the SOP editor , workout your vex_SOP...and then in your pops editor u can use a velocity POP and using the "point" function u can import in the values from the vex_SOP...and from there u can just build up your POP network.

hope this helps

Share this post


Link to post
Share on other sites
I think pops is slowly dying...

Vops is almost better for creating particle behaviour. It gives you the controll you never had in pops.

come to think of it, that's true! since POPs don't use the same paradigm that works for the rest of the package....it compresses the whole network tree into one simulation, and takes all POP operators into account at once, instead of processing it linearly like SOPS and CHOPS....

so going the VOPS way makes total sense! except for times where a certain POP operator has substantial coding into it that it's harder to recreate.....without the HDK...

hey! here's an idea! let VOPS cannibalize POPs! all you need is for some POP-specific VOPS to be added to the list of VOP tiles...

Cheers!

Share this post


Link to post
Share on other sites

right, well, it get's more complicated if you want to do collisions, and all the other handy pops, but having them as vops would indeed rock the house. B)

Maybe I was a little down on pops this morning, I think they're still useful, but I can't tell you how many people I've talked to that absolutely hate using pops. It's nothing at all like any other part of houdini, cause you really don't ever know what the heck is going on, unless you put some serious time into looking up the math behind particles.

And once you do that, pops can be kinda enjoyable, I think, although I have to say that using vops to controll particle behaviour is really cool, cause it gives you ultimate controll... EXCEPT for those functions that for some odd, bizzare, stange reason are not available to you, such as the import function.

u can output your pop into the SOP editor , workout your vex_SOP...and then in your pops editor u can use a velocity POP and using the "point" function u can import in the values from the vex_SOP...and from there u can just build up your POP network.

True, that can be done, but I really wonder why I have to... and it doesn't help with keeping things speedy eather, especially workflow wise.

Anyhoo, after having discovered angular velocity (how could I have missed it???) I'm all happy about pops again. :rolleyes:

Cheers,

Jens

Share this post


Link to post
Share on other sites

ok...i got the basic flocking system working...(through vex)

one problem though...if i try and use the collision POP or the attractor POP to navigate around object , it gets screwed up....probably something to do with the fact that houdini processes the whole particle network as one rather than take it linearly...

so how should i go about adding the collision or attractor POP...or will i have to write my own attractor VOP???

Share this post


Link to post
Share on other sites

hey mir

great news that it's coming along nicely. can you let me know, did you use a POP or SOP operator with VEX? a lot of this effects stuff does my head in but it's what i would really like to learn, so a nice clue would be of great help.

thanks

betty

Share this post


Link to post
Share on other sites

ok..

go to http://www.red3d.com/cwr/steer/gdc99/

this guy Craig Reylonds wrote this paper on steering behaviour for flocking systems...these r the concepts that i've implemented in my flocking system.

these r the basic concepts that u'll find everywhere...

i've created a VEX SOP which gives u control over the three major attributes of a flocking system (separation, cohesion and allignment)...

and since i cannot use the attractor POP options for object avoidance , im thinking about writing my own object avoidance code (in the same flock code)...

Mir

Share this post


Link to post
Share on other sites
nevermind seeing! it already sounds pretty cool! :)

if it has the "Ong" seal it can only be good :P he he he

thx Steven, one more thing to scratch off the "would-love-to-but-REALLY-haven't got time" list

Hmmm... when did I give the "Ong" seal? :huh:

Did you post to the correct thread, sirogi? ;)

aaaahh!!! I though pepe was YOUR avatar!!!

so much for speed reading!!!! youch!

ok one more time: "if it has the MIR seal....<etc etc>"

there...the planets are in realignment again! :)

cheers everyone!

Share this post


Link to post
Share on other sites
and since i cannot use the attractor POP options for object avoidance , im thinking about writing my own object avoidance code (in the same flock code)...

just a guess...but it's impossible to use the attractor POP because it would totally override the very same parameters that make your particles behave the way you want them to.....

you might get some mileage with the collision POP though, by projecting the position of a particle ahead of time, and determining if a collision occurs, and act in consequence...(a particle with its a "stunt" double...like

Pstunt(t)=Porigin(t)+(Vel(t+1)*dt))

just a crazy thought, I don't have implementation ideas...come to think of it, that probably won't work with the collision POP, but surely would allow you to make your own code where you check to see if the particle is "in" the collision volume or not (dot product the velocity and the normal of the nearest point of the surface after you calculate the above...makes sense?)

good luck with the implementation,I'm just good delivering such erants! ;)

man I wish you could have attended that R&H demo they made for cats and dogs...they used a lattice and slapped vectors on top to direct the flow of particles in addition to their flocking sysem AI, so the resulting motion would take both into account! (or soemthing like that ;) ) a REALLY cool system....wonder if they mind sharing the info they gave during the presentation, you might try your luck on the houdini mailing list, and see if there are some merciful people :)

Share this post


Link to post
Share on other sites

duh! jsut saw the collision avoidance in the paper you posted (great link BTW, thanks!)...kinda the same way I thought about!...if that guy say so...the you're all set! :D

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×