36 replies to this topic

[David Gary]

So... You asked for challenges...


For  procedural minds ,
If and only if you have time!


1)Break a piece of geometry in shards ( each shard is not a single primitive but a group of primitives , which is much mooore difficult to achieve)

2)Animate the shards with a particle network ( so it's not just instancing...).  
sorry i didn't post a more complex animation, but ut's just to show the concept
3)Plus ( that's what i did) : the division of the piece must be procedural ( to have control over crack patterns, that is you must define procedurally the groups of primitives).

EVERYTHING MUST BE PROCEDURAL




sorry for the adGif logo. I do not create animated GIFs very often!

[meshsmooth]

Ok crack up a torous then  

[David Gary]

Of Course!

It doesn't depend on the object.



(if animatedGif slow down the display on odforce, i will make it a link).

[David Gary]

Or a head !

[bobster]

Okay Mr FrenchOP, Please disclose brain power.  The ice breaking tutorial is easy enough but for the life of me I can't figure out how to group the primitives, but then I'm fairly new to Houdini.  I'm sure I'll kick myself when I find out.

Thanks

[TheDunadan]

I'd appreciate a hint concerning the following: can a single group SOP output multiple groups or a fuse SOP do multiple seperate fusions ?!
Defining the pattern for the actual groups seems easy enough for me, and I think I know how I could create a script that simply creates a seperate fuse or group SOP for each fragment, but having all these seperate fuse / group OP's doesn't sound like an elegant solution to me / has little procedural beauty .

[anakin78z]

TheDunadan, on Jul 1 2004, 11:27 PM, said:

I'd appreciate a hint concerning the following: can a single group SOP output multiple groups

                    Try the partition sop to do that

[TheDunadan]

I should have studied the online help more carefully, thanks

Anyhow, it's now working more or less. Once I'm more happy with the pattern I'll post my solution.

Jens

[Jostein]

Yo guys!  Did any of you crack this challenge?  I am working on my master's project at the moment - basing much of my particle stuff on the ice breaker (with basher object) - and I would really like to be able to group the primitives into different sized chunks.  

Question: wouldn't it be possible to create a lumamap in COPs (some kind of shattermap) and using the pic()-function, grouping and fusing primitives based on this information?

ANY input on this is extremely welcome!  
  

Cheers, jostein.

[halion]

I tried working on this last night for about 4 hours and didn't get anywhere.    I tried using the facet to make everything unique and then group primitives to form the shapes I want.  Problem is I could figure out how to fuse the primitives together to be one big chunck instead of 5 individual ones.  But this gives you no easy way to control the crack patterns anyways so back to the drawing board.

[Lukich]

Hey, guys!  Did anybody solve this challenge - it remains a big mystery to me, no matter how hard i tried   FrenchOP, please, show us your magic!!!

[TheDunadan]

I wasn't really happy with my solution and I'm going to do a hopefully nice/cleaner setup, but anyhow ...:

I created an extra attribute on the points of the grid and using a VEX SOP I simply applied a slightly modified voronoi kinda noise on this attribute: the voronoi noise has a bit of this cobblestone look. So far so good... now I used the Partition SOP to create the unique groups based on this . I don't know anymore if I created primitive or point groups, but I used something similar to the following roule: "point ("/obj/Ice/IceBr", $PT, "myCustomAttribute", 0)"

Using the trick Sebarrik suggest in his "prim sort otl"  I did a loop, through all the groups and animated these that way. Unfortunally my noise would generate numbers ranging from as low as 00001 up to 99999 (--> ugly group names) no matter if I'd have only 7 groups created via the partition using these values or 1000.

Looping though all those non-existant groups was very inefficient and therefore I created an intermediate attribute and tried several simple hash functions to create somewhat unique numbers ranging only from 0 - 4 * times the number of unique groups the partition SOP would create. Trouble is, for proper hashing you usually need some collision resolving and I don't know how to do so using the expression language in Houdini. Anyhow anyone who's intrested in the hashing theory can get a good introduction on this subject in the good old "Introduction To Algorithms" by Cormen.
In the end using this "hashed" groups some of the initial groups got fused and therefore a fair bit of trial and error was required when playing with the hasing functions.

Well, I guess FrenchOP has a nicer solution for this If I find my old scene file I'll post it.

Jens