[Custom Build] Blender Fracture Modifier

Dense meshes are a problem for blender itself, then that gets passed to the FM. There are several tutorials and guides on cleaning up meshes. On the FM docs page I put a guide for text meshes also.

You can try some aggressive mesh cleanup techiques and that might help. Also a popular technique in general for VFX is making a substitution mesh by creating a second mesh that you substitute for the first mesh right at the beginning of the sim so it is not noticeable. You can use retopology techniques for that. Or maybe even separate the mesh being broken into multiple meshes to begin with might help.

This second object technique is very similar to the ghost mesh second object technique where you use a simpler mesh in the same shape as the colliding mesh instead of using a complex mesh to break an object. Both of these are standard VFX techniques in destruction. So looking at it from a believable fake technique may give you the results you want.

Again, dense meshes are difficult for blender first of all, and actually, most 3D programs. So as a result it is best dealt with outside of the FM’s workflow. The reason being if you have to change a parameter in the FM it might again react badly to blender’s handling of the original and problematic dense mesh.

This question comes up regularly so please keep us posted on your progress.

Fracture ON!

I was planning to get camera close to a wall and make the wall break. Because the wall is shown up close, I wanted it to look nice so I used the subdivision surface and displace modifier and applied both.

From what I found important for FM, the wall mesh is clean and has no issues. It’s just a bit dense. The ghost mesh trick, if I understand it well, wouldn’t help here because the breaker object is very simple (basically a box) and hidden anyway. I’m rendering just the wall that seemingly breaks by itself.

Can you explain in more details how would the substitution technique that you mentioned work in this scenario? Or is there maybe some other technique that would be applicable?

One other and unrelated question: if I have FM in the modifier stack of a breakable object and I bake the simulation where the object is braking, what is the better approach if I later wanna add some particles that would interact with the breakable object - should I leave the FM modifier on stack and start adding other objects that emit particles or should I rather convert the breakable object to (many separate) keyframed objects (which applies the FM modifier basically)?

@Marach did you get your workflow figured out? I see your question is over a week old.

As far as substitution method, for detailed information you can use just about any tutorial from any package except maybe Houdini, lol… The techniques are basically the same and transfer well across packages. So just do a search to get the general workflow. Then you can ask specific FM related questions here.

The general idea is to select a frame where you keyframe the visibility of your main object and your VFX object on the same frame or one off. You can also check out the docs for ghost objects that will trigger and interact but not be visible.

Start there and then if you have specific questions those are easier to answer specifically.

Keep us posted.
Fracture ON!

Hello there,
just started experimenting with your modifier and I must say: amazing work you did there. Thank you very much for it and all the countless hours you must’ve put into it!
I found a video about an experimental brickify fracture method from '17 which I’d be interested in trying out. Is it still in development and will be added one day to the branch, or is it just developed for internal purposes? Any info on it would be much appreciated.

Hello,
you can fracture an object to bricks by using the “Grid” point source. You can set the offset to 0.5 in the according direction to get a brick pattern. Additionally, you can set a grid spacing to create a gap between the bricks.

1 Like

Ah, just found this option and was wondering what it’s for, so you already answered that question as well. Thank you very much, I’m having a blast right now rendering one destruction test after another.

So I’m just thinking right now, it should be possible to boolean cut the bricks that are created like that (after conversion to a normal object) from a cube of mortar, fracture that one as well, combine both using the container method and simulate them together, right? Only requirement for that to work would be that constraints can be created between clusters/shards of combined fractured objects so that the mortar clusters/shards would stick to brick clusters/shards. Is that possible somehow?

Also just discovered (after one hour of searching the web and circling back to you) on github that you’re keeping the molecular addon up to date which I just found out about, so thank you as well for that. Simply put: You and the other guys rock!
I hope you don’t mind if I’ll write any questions and ideas down right here that I surely will have while toying with the modifier.

you could also just after boolean cutting out the mortar with the bricks apply the mortar and the bricks, and join both objects. then you can further fracture them and use split shards to Islands.

Alright, will try that. Thanks again.

Just encountered a problem that I can’t find a solution to. Tweaked everything to my taste so that shards and clusters behave believably, but out of nowhere, shards inside of clusters started to ‘wobble’ during simulation. See the top area of the middle cluster in the following test render:

My uninformed guess is that only some vertex constrains break on these shards, but not all, which could be leading to this behavior. I’ve already set physics steps and constraint iterations very high right from the beginning of my tests, so I don’t think that’s the problem here.
I’ve got no clue which parameters to change now to get rid of this without breaking the overall behavior of the simulation which took hours to tweak. Any tips or ideas?

1 Like

hmm, if you have vertex constraints, this usually means you connect adjacent shards with each other. With centroid constraints and a larger search radius, you can connect farther away shards too.
In case you dont wanna connect them more firmly, but let them break before they wobble, you need lower breaking settings like angle or threshold.
Alternatively you could try and use Animated / Triggered on the wall and Trigger on the sphere. That will only activate hit shards. In order to reduce the probability something is stuck in the air, you could:
use “Activate Broken”, clusters and longer search radii, as well as a low cluster breaking angle and maybe a cluster percentage (to force cluster breakage earlier and break all the rest of the connections of a shard in case a percentage of broken connections is exceeded.)
The advantage of triggering is that the sim is faster, only necessary shards get activated, and it may reduce the wobble, the disadvantage is you might need to re-tweak your settings.

In case you didnt already, it might be also worthy to look at the documentation:

https://en.blender.org/index.php/User:Scorpion81/Fracture_Documentation

Generally, without looking at your blend and without possibility to change some parameters it is hard to suggest what might be the “magic switch” here.

1 Like

Thank you for your time!
I started out with centtroid, triggered and activate broken, but I wasn’t able to get a sim behaviour as convincing as after switching to the setup I have now. I’ll have to render a couple of closeups of the destruction, so I can’t get away with hiding sloppy sim behaviour somehow.

I’ve read the complete blender wiki version of the docs before starting with my experiments, also followed all three tuts by blenderphysics (great stuff by the way just in case you read here as well).
Is the docs version you’ve linked on github more up to date? If so, it’s not that clear and maybe it would be a good idea to link there a bit more prominently. Just meant as a suggestion from a beginner POV.

If you don’t mind, I could strip down my blend file and upload it to a place of your choice. I’d never expect that you’d tweak it for me to perfection, but I’m pretty sure your experienced eye will spot some possibly miss-tweaked values right away. I’d appreciate any guidance you’d be willing to give with this.

I can try my very best :slight_smile: you could upload that to Google drive and put a link here or so

1 Like

Hi everyone, I am getting trouble when I put Build FM button. As you can see in the picture It says “Fracture Modifier not available in this Blender version. Visit graphicall.org/1148 for the FM-enabled Blender version.”

I tried both 2.79a and 2.79b versions but did not work. In addition, graphicall.org/1148 is not available.

I fixed thanks to Scorpion81. I download him Blender version.

May I learn that it is possible the animation can be exported fbx file. Not the non-destruction model ? I want to see results in VR. Regards

I’m also currently looking into this for export to UE4 engine. I’ve found Ossim so far that could be worth a try: Ossim - Bake simulations to armature (UE4, Unity) - Now with Blender 2.8 support!
But I haven’t tried it myself yet, so I can’t say if it really works. I guess you’d have to convert the FM sim to keyframed single objects first for it to work correctly. If you happen to try it out, some feedback would be nice.

Stripped it down as far as possible and even compressed it, still has some 16MB. Either blender files got much bigger since I last checked, or I overlooked something. Anyways, here’s the link:

Just had an idea as well regarding this. Since there are no example files for FM easily available (or at least I’m not aware of them) and I’m going to tweak simulations for some common materials in my test file anyways, would you like to take that file when finished and provide it as some kind of learning or reference material? From my POV at least, learning how to use something complex as FM is best done when looking at examples that work already and toy with them, so I think would be helpful for others as well. As said, just an idea. And I’d be happy if I could give something back that way.

As long as you have rigidbody chunks as separate objects it should work fine. I did a lot of tests using fracture addon and ossim.

1 Like

Well, being the creator of Ossim certainly adds to the weight of your statement :smiley: Thanks man!

1 Like

Hmm, the file is still that big because the baked sim is still inside. It runs slower because other modifiers after FM are being executed “live” there, plus the autohide is also a “live” operation. And regarding the file, the only useful suggestion i can add here is trying to use the Automerge Distance with a value of 0.1 and maybe crank up the autohide distance to 0.001. This attempts to hide the bigger crack appearing in the upper part of the middle cluster. Unfortunately the displacement method is not perfect, since the displacements dont “fit together” (in general) and thus you get holes or overlaps. Autohide and merge can barely fix that since it happens “after” the FM. What autohide and merge can do is trying to keep cracks between shards longer visually together.

And yes, providing example files for others to look at and learn is a great idea :slight_smile: Some time ago i had some examples somewhere on dropbox or googledrive, but those likely dont work anymore as in they might not open anymore / crash (and that were simple examples only)

Ah, found the ones on googledrive:

https://drive.google.com/open?id=0Bwkun-jSiZ-RRFEweWE1cTV0Vms

1 Like

Oops, thought I had cleaned that up last time. But on the positive side, that saved you resimulating, so there’s that.
Alright, I’ll fiddle with these settings, thanks for your suggestion. I’m aware that these live modifiers add to the overhead, but in my case that’s not so important since I’m fortunately not bound to a deadline right now. Haven’t found another way of making the shards look more lifelike yet unfortunately.
Okay, so actually you’re saying, the wobble doesn’t come from actual micromovement of shards, but is rather something like a Z fighting problem of overlapping faces then?

Well then, I’ll work on it with that goal in mind as well. And I hope you’ll be open to answer all my stupid questions on the way. There’s a lot that I still can’t get behind, where the docs are not helping either. :slight_smile:

Alright, the render is through, and your suggestion nailed it in even two ways. Automerge didn’t work because it introduced holes in the shell of the wall (like missing surface polygons without any interior). But setting auto hide to 0.1 got rid of the wobble and even introduced nice cracks appearing in certain areas whereas I was already wondering how to achieve that. Thanks once again.