I really have to excuse but I have to ask one of the
which-programm-is-more-usefull-questions.

I´m MOSTLY interested in destruction effects:
especially fracture(!!!),collapse and explosions.

I took a peek at 3ds Max and
I took a peek at Houdini.
For both applications I would evaluate my knowledge
between absolute beginners and intermediate level.

I know that 3ds Max has great plugins for destruction effects
like volumebreaker or rayfire but I did not used them jet.

lerning two 3d applications at the same time is a bit much for me so
I think I have to decide: max or houdini.

Of course: This is a side Effects forum but:

Does anybody have experiences with destruction effects in
hudini, max, volumebreaker or/and rayfire and can help me a bit with my decision???

Greetings Gerd

Not that I'm a complete expert. But, I have Rayfire. I end up using rayfire mostly just because It's pretty simple to work with in max. (and most of my projects are done in max). But If your talking about some seriously massive destruction sims. I would go with Houdini. Which seems to work with complex simulations waaayyy faster than rayfire would. If your destroying a building: rayfire.

If your destroying multiple buildings that's collapsing on a bridge full of cars while the earth is being torn apart between them: houdini :!: :mrgreen:

Overall, learn both. Its useful. I did it (to an extent), and i'm 16.

Plus, 3ds max doesn't have reactor anymore, which would make car destruction very difficult

Stereoscopo
If your destroying multiple buildings that's collapsing on a bridge full of cars while the earth is being torn apart between them: houdini :!: :mrgreen:

Go big or go home

If you want to do anything more complex than simple voronoi fracturing you should go with Houdini.

Volumetric fracturing is very powerful and can be set up with bullet and bullet glue constraints. You can achieve quite complex fracture patterns with a lot of control.
This will take time to learn and set up as you will be building custom setups and not pressing a button.

By understanding custom setups you can design your own fracture patterns, not just in the surface, but in the volume as well. You can create different representations of your rbd pieces, from polygonal, to volumetric, to points. Each representation can be used for different goals.

polygons -> bullet sim
volumes -> intersecting volumes can be sources for dust an debris.
points -> efficient way to store your rbd sim. – instance low detail or high detail polygonal pieces onto them at rendertime.

Now think about deforming rbd sims - like metal shards denting, tearing and rolling around. At this point Houdini is about the only program that can handle that. Most other softwares stop working properly with changing point counts.

Have fun learning!

I mainly work with Houdini these days, but for destruction I still am a big fan of Thinking Particles. Prebreaking can be done in a lot of ways but in terms of performance, TP is still the best solution out there to simulate tons of geometry and you don't have to be concerned about concave/convex stuff …

just saying, please don't boo me Peter

cheers,

Manu

Well, I have not used thinking particles. So I don't know what kind of fracturing possibilities it offers. Please elaborate (wouldn't want to boo you uninformed ). Any tests - or links you can share? And yes, the concavity can be annoying, but I think that is only a matter of time before it gets ironed out.

The only other big alternative I see for advanced fracturing which is currently not part of houdini is DMM.

I think one of the most important things is to have control over the fracture pattern. Whether it is wood, crystals, glass, metal, (re-enforced) concrete, rocks, sand… And this side has less to do with the fracture/rbd engine and more to do with procedural modeling.
– I don't really know how DMM actually deals with this. I'm guessing it is defined in pre-baked patterns/functions that influence the tetrahedron bonds. - Or perhaps users can define the breaking patterns themselves from… textures? volumes? functions?

Then I find that the secondary effects that are based on/triggered by the primary sim benefit from all the analyzation tools houdini has to offer.
trail -> velocity
chops -> acceleration
dops_data -> impacts/velocity
transformation matrices on points -> efficient storage and fast scrubbing as all your transformations can be applied on points, multithreaded through vops.

– also, Gerd, consider the industry you want to work in/use this for. Games and commercials will be more likely to use a system like TP (and max in general) - whereas large scale fracturing for feature will most likely require more custom, streamlined setups - which you can set up with Houdini, but it will take some rnd time.

Also, slightly off-topic, but here are a few reasons that you may want to consider learning houdini more in general. (This was posted in regards to Maya, but is completely applicable to 3dsmax as well – I used 3dsmax 10 years ago and it has not changed much since imo ).

This was the guys question:
http://forums.cgsociety.org/showthread.php?f=86&t=1034527&page=1 [forums.cgsociety.org]

This was my reply to help inform them:
http://forums.cgsociety.org/showpost.php?p=7241540&postcount=17 [forums.cgsociety.org]

pclaes
Well, I have not used thinking particles. So I don't know what kind of fracturing possibilities it offers. Please elaborate (wouldn't want to boo you uninformed ). Any tests - or links you can share?

TP comes shipped with Volume Breaker which is a procedural breaking tool based on Voronoi similar to Houdini's voronoi fracture SOP. It's been mostly used with a combination of dynamic fracturing and a pre-breaking process with either custom tools (which I prefer) or Rayfire, which definitely allows complex high detailed fracturing.
Overall the advantage of using this package is simply enough performance, TP's solver compared to Houdini's RBD solver is waaaay better and faster, (once you're used to it you can't believe how slow houdini is), plus there is a great connection to fumeFX, I don't even want to get started comparing Fume's speed to Houdini's pyro
Well that being said, you got to test it to see for yourself Peter, doing dynamics with TP is simply mindblowing!

Here's some links:
http://www.youtube.com/watch?v=2VMNGEe9OAc [youtube.com]
http://www.yeatvfx.com/works/surface–collapse–system.html [yeatvfx.com]
http://www.youtube.com/watch?v=AD2l10pJhss&feature=player_embedded [youtube.com]
https://vimeo.com/groups/56252/videos/34067106 [vimeo.com]
https://vimeo.com/groups/56252/videos/17890633 [vimeo.com]

Awesome information on this thread, thanks for starting it and sharing.

pclaes
The only other big alternative I see for advanced fracturing which is currently not part of houdini is DMM.

May i ask whats DMM?

eitht.

Hey Manu, thanks for those links. It does look speedy and fast. but it's still all voronoi based and how is it to control and set up things like constraints?

I really like the ability to cut my constraints procedurally inside of a sopsolver based on whatever trigger patterns I design. I find I rather give up some speed if it means more control - most of the time I work on patterns and smaller sims during the day and the big sims go overnight.


DMM stands for Digital Molecular Matter:
http://en.wikipedia.org/wiki/Digital_Molecular_Matter [en.wikipedia.org]
http://www.pixelux.com/dmmPlugin.html [pixelux.com]

But actually that is an implementation of it. What I really meant is FEA, which means Finite Element Analysis.
Read this for a lot more info on destruction:
http://www.fxguide.com/featured/art-of-destruction-or-art-of-blowing-crap-up/ [fxguide.com]

Here is another great pdf full of good info, this is more on volumes.
http://magnuswrenninge.com/content/pubs/ProductionVolumeRenderingSystems2011.pdf [magnuswrenninge.com]

The section I'm talking about in regards to volumetric fracturing is on page 70:
Chapter 5 - Cutting up models

That really is only the tip of the iceberg. There is so much more you can do with that.

Interesting conversation going on. All my comments are my own opinion.

Having used various tools for destruction in different facilities, both proprietary and out of the box.

I'd say if you're going out of the box, 3dsMax+RayFire+TP hands down beat Houdini when it comes to heavy RBD work. I did various tests for production with the DOPs RBDs and found them to be ok for small debris type work, at very slow speed. But anything involving large scale it is basically useless.

Most facilities have had to introduce a Bullet solver interface built inside of SOPs as a sort of particle based system with geometry transformation . A good deal of information can be found here [bulletphysics.org] from Nafees and a mix of DD and Dreamworks TDs.

This type of workflow is the most robust and useful I've seen so far. And most importantly it is very fast, you can get iterations out fast and in the end it also allows you to easily clean up your simulation by hand and switch out to higher resolution geometry if needed.

I know Houdini 12 has Bullet built inside of DOPs, but it is still limited by the nature of DOPs, which for RBDs imo makes it pretty useless for anything remotely medium to large scale. I will definitely be testing it more and maybe change my opinion on things.

-Rick

Pclaes : TP does have constraints. You can create them, place them, and break them as needed. There are quite a few different types of ‘joints’: Fixed,Spherical,Cylindrical, Ball,Hinge,Spring,Wobble,Slider
Setting them up is extremely easy and flexible. You can do it procedurally based on the particles or you can even place them by hand in a rig, and import the rig into TP.

We used TP for 2012 's LA Limo Sequence: https://vimeo.com/17598843 [vimeo.com] and it was also recently used on Final Destination 5: http://www.fxguide.com/fxpodcasts/fxpodcast-final-destination-5/ [fxguide.com] Manu up there actually worked on FD5, I only did some initial RnD for it. Joints were heavily used in both, although they were still a little ‘young’ to be used to their fullest potential on 2012.

TP does only have voronoi shattering built in, but you can use the cutting tools in max, or rayfire (which offers good scripted functions that use the build in cutting tools) to generate non-voronoi shapes to start with. Usually we would start from the custom cut shapes, and then do voronoi for any subsequent breaking. You could also just start with completely broken object, and just use the ‘fragment’ operator to break it apart, or use joints to hold it together.

Of course, Houdini has some really great advantages when dealing with things that change point counts, or when working with volumes. You just don't get that in other packages, and it's awesome to have, but in terms of ease-of-use and speed I'd still pick TP for large scale rbd work. With enough, well thought-out otls, Houdini and bullet might be offer some good competition, but not out of the box for most users. Like Rick said, a lot of places seem to roll out their own bullet implementation anyway.

Quick edit: in terms of pretty much everything else I'd say go with Houdini. There are a lot more jobs for Houdini in film, and it's an all around kick-ass piece of software.

pclaes
If you want to do anything more complex than simple voronoi fracturing you should go with Houdini.

Volumetric fracturing is very powerful and can be set up with bullet and bullet glue constraints. You can achieve quite complex fracture patterns with a lot of control.
This will take time to learn and set up as you will be building custom setups and not pressing a button.

Hi Peter, what do you mean by volumetric fracturing ?
Is there something else to fracture objects than voronoi and cookie/boolean ?
Also not sure what's the relation to bullet, which handles the sim and not the fracture.

Sorry I'm a bit confused and would love if you can expand a bit.

Cheers
Vincent

The concept of volumetric fracturing is not new:
http://www.sidefx.com/index.php?option=com_content&task=view&id=1454&Itemid=68 [sidefx.com]

and look in that pdf I mention a couple of posts above.

There is no volumetric fracturing tool available from the shelf, but that does not mean you can't make one.

There are a lot of different ways to fracture volumetrically. The simplest is probably using boolean operations – this would work very similar like a cookie sop, but instead you use 2 volumes and a volume mix (or a volume vop with a volume sample as this is much more efficient when dealing with huge volumes as it only samples the section it needs to sample and it is multithreaded). This generally works better than a cookie sop as the cookie sop is prone to errors. You can use this recursively in a foreach and cut your volumes up in a fancy way.

– at this point as an output you have a bunch of fog volumes, which you can turn into sdfs, which you can directly use with the rbd solver. – but not with the bullet solver as bullet requires polygons. You can volumeconvert your sdf volumes, offset them from the surface (which is really necessary to make sure bullet sims don't explode because of intersecting geometry), polyreduce them further if necessary.

Other more complex volumetric fracturing models could deal with “growing” fracture patterns into a volume and using those to boolean against. This is similar to aggregation techniques. You could look into stress patterns to indicate how the fracture should grown into the volume.

Or if you don't want to go that far, you could use various noise functions. Or you can create some template points to get a base distribution of pieces for your fracture patterns similar to voronoi. But unlike voronoi, you are not restricted to planar “cuts” and you can also assign the same id to a bunch of points to grow from those points, this way you could create curved fracture patterns.

pclaes
… You can volumeconvert your sdf volumes, offset them from the surface (which is really necessary to make sure bullet sims don't explode because of intersecting geometry), polyreduce them further if necessary.

I thought about doing boolean with volumes, but it's always the convert back to polygons part which stopped me. Looks to work nice though, I'll try it, and I'll look at the pdf.

Cheers!

I've been wondering whether the boolean operation could be augmented with a volumetric sanity-check. Would need to be in-built though.

Half of the smarts would of course come from the human checking that the volumetric representation is sane.

gerd

You should ask yourself a few questions:

1) Are you intending to use pre-built solutions or build tools yourself? What are your turn-around times? (i.e. can you afford building tools and how much time do you want to dedicate to that?)
2) If you are going to build tools - what is your resource budget? (man-hours; are you going to create tools alone or with more people?), what is your starting point? Does your facility already have a basic toolset or are you starting from scratch?

Depending on your answers, the solution may be of the the kind that Ratmann suggested, or the complete opposite

Houdini framework allows, no-doubt, realize one's wildest dreams… there is a chance of going bankrupt and dying of hunger in the process though

Good luck either way.

http://www.youtube.com/watch?feature=player_detailpage&v=u_2qKQpxCRw [youtube.com]

This is very powerful & easy

MADjestic
gerd

You should ask yourself a few questions:

1) Are you intending to use pre-built solutions or build tools yourself? What are your turn-around times? (i.e. can you afford building tools and how much time do you want to dedicate to that?)
2) If you are going to build tools - what is your resource budget? (man-hours; are you going to create tools alone or with more people?), what is your starting point? Does your facility already have a basic toolset or are you starting from scratch?

Depending on your answers, the solution may be of the the kind that Ratmann suggested, or the complete opposite

Houdini framework allows, no-doubt, realize one's wildest dreams… there is a chance of going bankrupt and dying of hunger in the process though

Good luck either way.

Totally agree. Side Effects should talk talk and listen average customers as well not only programmers and coders from big studios. Some times I think they purposely keep thing complicated (maybe some1 is trying to keep his job .

Typical example are Thinking particles, showing that it`s possible to have very easy rbd activating, ungluing and fracturing. In Houdini fracturing is very easy and good but to activate pieces or unglue them, you have to write code. Imagine that you try to create event which will transfer particles in another group - code again. Things like that you could achieve just by simply wiring few nodes in TP! Yes we should have one button solution (coders hate that and at the same time ability to code on top of that if you want. I don`t think that any1 will complain on that.

Like pclaes wrote, slightly off topic, on “H is taking over vfx jobs”, “but here are a few reasons that you may want to consider” simplifying Houdini workflow:
http://forums.cgsociety.org/showthread.php?f=59&t=1054227 [forums.cgsociety.org]
It shows how far you could go when you provide solid, easy to use, RBD system.

Don`t get me wrong, but I just don`t understand how possible it is to have situation when external plugin provides that much comparing with serious tool like Houdini.