Building A Fractured Box With Cookies.

[Symbolic]

Hi,

I am trying to build a pre-fractured box using cookies. But I have the following problems.

First of all, when the boolean parts come together, I get bad edges.

you can clearly see the artifacts.

//

Using a facet is kind of fixing the problem.

No more bad edges.

//

My main problem is that I need to have an internal thickness. The cube is empty inside and I need thickness to the chunks.

I tried using PolyExtrude... But it did not work.

There is nothing in between:

//

Thanks for your time and help.

Here is my file:

fracture_test_v1r3.hipnc

[aracid]

hey man

I dont know if you know of Craigs Fracture SOP, as far as i know, you should be able to reverse engineer it, and would probably resolve many issues that you might run into when building a procedural shatter sop.

found in the exchange : http://www.sidefx.com/index.php?option=com...&Itemid=149

i hope this helps

[Symbolic]

hey man

I dont know if you know of Craigs Fracture SOP, as far as i know, you should be able to reverse engineer it, and would probably resolve many issues that you might run into when building a procedural shatter sop.

found in the exchange : http://www.sidefx.com/index.php?option=com...&Itemid=149

i hope this helps

Hi aracid,

Thanks. I know about Craigs Fracture SOP. I have tried it... it works... but in my case, I am trying to make a cube that is empty inside and has some thickness to the walls. I think the Fracture SOP creates volumes of chunks.

[ihab]

How about creating your double walled cube first with a boolean and then run Craig's fracture sop on it.

hope this helps

Hi aracid,

Thanks. I know about Craigs Fracture SOP. I have tried it... it works... but in my case, I am trying to make a cube that is empty inside and has some thickness to the walls. I think the Fracture SOP creates volumes of chunks.

[johner]

Hi aracid,

Thanks. I know about Craigs Fracture SOP. I have tried it... it works... but in my case, I am trying to make a cube that is empty inside and has some thickness to the walls. I think the Fracture SOP creates volumes of chunks.

I played around with your file a bit in 9.0.768. You were awfully close to the right settings; with the Cookie SOP it's often a matter of tweaking, turning on Jitter, etc., especially when dealing with thin shells. I'll attach what I got working. A few of the steps I took if I remember correctly:

1) You don't need to subdivide the cube that you're cutting, only the one you're using Mountain on to get the jagged edge

2) You should probably apply the facet at the end after all the cutting is done. Both of these are results of, the fewer points sent through the Cookie SOP the better, I think.

3) You had Pre-Convex geometry off in the Cookie SOPs and it seemed to work better on for me.

4) I did a Reverse SOP on one part of the cutting pieces just to get all the primitive normals pointing the right way

5) I collected the various parts into groups called piece1 and piece2 so you can send them to a DOPs sim as an RBD Fractured or Glue object if that's what you're thinking next.

6) The Cookie SOP is very sensitive to the thickness of the PolyExtrude with these settings. If you change it and parts start disappearing you may need to turn on Jitter in the Cookie SOPs

Like I said, your settings were very close, but in cases like this the Cookie SOP can be a little flaky, so you might have the right settings, but need to turn Jitter on or something before you see the correct results. Also, the Cookie SOP has had some work on it in 9.1, so I got different results with the newer version. You didn't say which version you were using, so I went with the released version.

Also, if you're about the send the results through a DOPs sim, very thin "plates" like this are hard to represent as collision volumes unless you crank the resolution on the volume way, way up. So you might have to crank that way up, turn off Volume Collisions, or use proxy geometry. The last idea is probably the way I'd go. Do a similar, mostly reference copied SOP chain to what you have, but without the PolyExtrude, meaning you'll be dealing with the solid box version. Much more pleasant geometry for the RBD solver to work with. Then use DopImport to transform the "shelled" box pieces by group name. This is a bit more involved to set up, obviously. Let me know if you're interested in seeing how.

Good luck.

fracture_test_v1r3_modified.hipnc

Edited December 26, 2007 by johner

[Symbolic]

Hi johner,

Thank you very very much for this detailed answer... I am working on your file right now...

Seems to be working nice... I will try to make other tests as well... with a sphere and a tube...

Meanwhile... I am very interested in your DOPs related techniques (proxy, volume calculations etc.) that you mentioned before... about thin objects...

Thanks for your time. It has been a little calm around here due to Christmas, I was just starting to get scared

[Symbolic]

...ok... hi again...

I found one old file from my school days... It has a sphere as an example... So the FACET kind of helps the cube to look clean... But when the object is a sphere or something that cannot be FACETed... how we get rid of the connection artifacts?... or do we just try not to feel too bad about them?

Here is the file. Despite lots of FUSEs... it still has the connection artifacts...

fractured_sphere_end.hipnc

[Netvudu]

Just for my own curiosity...what

Edited December 26, 2007 by Netvudu

[Symbolic]

Yeah... ok... I guess you are right... I just wanted to know if this is how it is... or are there any solutions for the problem...

Edited December 26, 2007 by Symbolic

[johner]

Hi johner,

Thank you very very much for this detailed answer... I am working on your file right now...

Seems to be working nice... I will try to make other tests as well... with a sphere and a tube...

Meanwhile... I am very interested in your DOPs related techniques (proxy, volume calculations etc.) that you mentioned before... about thin objects...

Thanks for your time. It has been a little calm around here due to Christmas, I was just starting to get scared

Sure thing. First of all I agree that most of the time you'll be swapping the fractured geo in right before the actual break occurs, so don't get too hung up on the render appearance of it, artifacts, etc.

I'm attaching a file with the proxy idea. Basically there's a parallel network within the geo3 object that cuts out two pieces from a box set up so that the outside edges of the geometry look the same. Based on a Switch SOP, the proxy geo is either solid or just a thicker shell. Either way will work to get a decent collision volume. Strictly speaking, a thicker shell is probably more correct since the Center of Mass will be more similar to your render geo and the basic behavior should look better. Either one should work from a collision perspective so long as nothing is supposed to be colliding with the "inside" of the shell.

The thicker shell version looks just like the render geo, the shell is just thicker to make collision volumes more robust. Lots of the nodes are Reference Copies so the proxy will update correctly if you change the render geo. Also look at the "Thickness Multiple" on the "thicker_extrude" node. So the proxy geo has two groups called "piece1" and "piece2" that look just like the render geo. Then you set up the DOP network to use the proxy geo, and use a DOPImport SOP set to then transform the render geo based on group names. If you go into the "dopnet1", you'll see the proxy geo, and you can view the collision volumes there to verify they're good. Notice that if you set the "Thickness Multiple" to 1, or just point the RBDFracturedObject at the render geometry, you'll have a very tough time getting a decent collision volume.

Hopefully it's clear from the file, but I can answer any questions you have. By the way, if you're doing anything remotely serious with DOPs, you owe it to yourself to get Craig's Dynamics DVD, from:

http://www.thegnomonworkshop.com/dvds/cze01.html

I went through it again recently it's really thorough.

If you want something cheaper and downloadable, he has a tutorial at:

http://gnomonology.com/prod/21

that goes into proxy vs. render geometry for DOPs.

Good luck.

fracture_test_proxy.hipnc

[Symbolic]

Wow!

Thanks. I will go through your example... thank you very much. Sounds like a very valuable workflow...

//

I have been trying to learn and practice with DOPs for a while... but have always been demoralized by strange issues.

I know it is all about problem solving, but sometimes it is quite annoying. Things like geometries sinking into the ground plane... pieces of fractured geometry that suddenly start to move around even the fact that they must stay still.

I have the two Houdini DVDs... Houdini 101 and DOPs... I have to say they are great. For the DOPs I have seen all chapters from 1 to 11. It helped alot. But there is not much about some crazy things that usually go wrong...

So to put it in another way... Resources on DOPs, examples on the forums etc. are great... But It is still hard... at least for me. But what is not!? I guess it all depends on practice...

[aracid]

Wow!

Thanks. I will go through your example... thank you very much. Sounds like a very valuable workflow...

//

I have been trying to learn and practice with DOPs for a while... but have always been demoralized by strange issues.

just remember man, u'll be surprised with what u can get away with if the object or RBD sim just feels right, i.e. timing, weight and choreography.

if you're building an effect and dont be disheartened by not getting 100% accurate collisions etc AND almost all the work ive ever seen comes from layering different simulations on top of each other.

so if u are going to practice, try cover the whole spectrum of what u wish to achieve (ie big rocks, mid range lighter rocks, small rocks, dust and sand), often a saving grace is that these issues can be hidden with dust or secondary effects, that often add alot more realism and is essential but often is neglected.

[Symbolic]

Hi aracid,

I will keep your advice in mind. Thanks.

[Symbolic]

Hi Everybody,

Still looking for a good tool that will help the "fracturing" process...

As many of you might know, there is a tool by Blast Code, called Megaton. So I thought that (in theory)... ...a tool like that might be helpful to create a pre-fractured object.

http://www.blastcode.com/index.php

But after keep reading and looking at the tutorials, I started to think that it is not actually as good as it seems. I could not find any example involving objects with internal volume! Most of the examples are grids with some thickness. So I kind of gave up the idea of using Blast Code to generate fractured geometry in Maya and export it for Houdini. Any ideas about that? Am I right? ...or am I missing a point...

So as far as I know, the best results are those posted at:

http://forums.odforce.net/index.php?showtopic=6257

...using that MaxScript.

I think I will go back to my cookies... it will take ages ...even for the big chunks, secondary chunks as aracid suggested may help... but it is still too much cookies for the main parts.

Any final suggestions?

[Jason]

I think I will go back to my cookies... it will take ages ...even for the big chunks, secondary chunks as aracid suggested may help... but it is still too much cookies for the main parts.

Any final suggestions?

I'm beginning to think I'm saying this a lot lately, but have you tried the Cookie in the H9.1 beta? It's meant to be a lot more robust than before - I don't know, I haven't really tried it.

[johner]

Hi Everybody,

Still looking for a good tool that will help the "fracturing" process...

As many of you might know, there is a tool by Blast Code, called Megaton. So I thought that (in theory)... ...a tool like that might be helpful to create a pre-fractured object.

http://www.blastcode.com/index.php

But after keep reading and looking at the tutorials, I started to think that it is not actually as good as it seems. I could not find any example involving objects with internal volume! Most of the examples are grids with some thickness. So I kind of gave up the idea of using Blast Code to generate fractured geometry in Maya and export it for Houdini. Any ideas about that? Am I right? ...or am I missing a point...

I used BlastCode for some glass breaking effects for a film last year. It's pretty cool as a specialized destruction tool (and if you're stuck in Maya), but, in my opinion, there's nothing in it that couldn't be done at least as well, and more flexibly, in Houdini. You can feed similar crack textures into the Trace SOP, followed by a PolyExtrude SOP, to create "cracked" geometry for things like glass breaks. Or just draw the curves by hand with point snapping, although that takes a bit longer obviously. An animated Magnet SOP can provide any of the surface deformation that BlastCodes "explosions" do.

For fracturing already existing geometry, then I do think that multiple applications of the Cookie SOP is the way to go, either by hand or using something like the Fracture digital asset. It might be worth moving to 9.1, as I think the Cookie SOP has had some work on it. You might also use simpler geometry for your Cookie operations, at least for testing. It doesn't need to be as slow as in your example scene, you're just creating a fair number of polys because your cutting geometry is so dense.

If you remember in Craig's Dynamics DVD that you have, there's a whole case study on a building destruction scene. He says in there a few times, the most difficult, time consuming part of the whole thing was creating the pre-fractured geometry for the simulation. I think that's sort of to be expected. While I think the Fracture SOP could stand to be updated, and with the combination of the ForEach SOP and the updated Cookie SOP it could probably be improved, for ultimate control you'll still want to do a lot of that by hand.

[Symbolic]

I used BlastCode for some glass breaking effects for a film last year. It's pretty cool as a specialized destruction tool (and if you're stuck in Maya), but, in my opinion, there's nothing in it that couldn't be done at least as well, and more flexibly, in Houdini. You can feed similar crack textures into the Trace SOP, followed by a PolyExtrude SOP, to create "cracked" geometry for things like glass breaks. Or just draw the curves by hand with point snapping, although that takes a bit longer obviously. An animated Magnet SOP can provide any of the surface deformation that BlastCodes "explosions" do.

For fracturing already existing geometry, then I do think that multiple applications of the Cookie SOP is the way to go, either by hand or using something like the Fracture digital asset. It might be worth moving to 9.1, as I think the Cookie SOP has had some work on it. You might also use simpler geometry for your Cookie operations, at least for testing. It doesn't need to be as slow as in your example scene, you're just creating a fair number of polys because your cutting geometry is so dense.

If you remember in Craig's Dynamics DVD that you have, there's a whole case study on a building destruction scene. He says in there a few times, the most difficult, time consuming part of the whole thing was creating the pre-fractured geometry for the simulation. I think that's sort of to be expected. While I think the Fracture SOP could stand to be updated, and with the combination of the ForEach SOP and the updated Cookie SOP it could probably be improved, for ultimate control you'll still want to do a lot of that by hand.

Hi johner,

I really do not want to sound like a lazy person... I am ready to do the job with Cookies... and I am really sorry about bringing this topic up again and again...

just wanted to explore the issue a little bit more. And that Max 9 script seems to be working nice... it is not a wise pipeline to go between H9 and Max9... but it is very charming...