I used an explicit FEM method to simulate the shape of porcelain fragments after the fall of ceramics, but the problem of time and computing power make me give up. Luckily I met this program and the results seemed reasonable. I’m totally new in the computer graphics field so I want to know how physically reasonable.
Does Fracture Modifier base on mass-spring method? Could you tell me which method was used?
Where can I read about the theoretic base? I want to know at least what material properties used.
We are using the Bullet library in Blender which is a rigid body simulator that uses constraints with specific breaking thresholds to keep elements connected. This is a discreet element method and thus a stiffness matrix as in FEM is not used, here is a paper about it: http://kostackstudio.de/dl/docs/2012-EG-STAR_Rigid_Body_Dynamics.pdf
We have also developed the BCB add-on for simulating materials more realistically. Other than the FM alone the BCB computes the breaking thresholds for every degree of freedom in every connection using real world material properties. It’s a bit more difficult to use though. You’ll find more information here:
@motion blur topic:
Scorpion81, Kai and I are taking a look at your file and workflow issue @Mark_Spink. Hopefully we’ll have some usable info for you in the next day or so.
Please understand, it is a combination issue of how blender handles modifiers and mblur, how FM does, and the workflow you are using.
Multiple things need to be fixed it appears including blender’s modifier interaction with mblur to have a proper and complete fix.
So as an alternative, one workflow modification is to use nodes and comp techniques to do your post mblur. Also FM likes to be the last modifier in the stack. So always keep that in mind.
I would recommend milestone file save versioning and archives so you can go back to key points like right before applying modifiers. This still preserves somewhat your ability to make changes.
It all depends on if you are in a professional setting or just having fun playing with the FM.
Keep in mind we are still actively developing FM 2.79x versions. We are currently refactoring our 2.79b codebase to a more maintainable and extensible code base. After that we can apply a partial fix in the FM code to your mblur situation.
IMO, the 2.8 codebase of blender still has many “opportunities” for improvement that need to be addressed before a 2.79x feature comparable version of FM can be put into it.
So until then we are recommending a blended workflow using FM to output what you want in destruction then finishing in 2.8. Keep in mind, there are 2.8 professional type and more complex workflows that are still borked. While the simpler and less complex finishing workflows the devs have been patching up to be usable are more stable in 2.8 currently.
Or you can just stay in 2.79x for FM tools. We are further developing some interesting technologies as well as fixing issues with this version. Plus I have an outstanding addon on the way that will give creative control over the FM in a less technical and more artistic way.
Cheers for the prompt reply! I thought it might be a sampled motion blur issue, it ‘felt’ similar to an extant bug I found- nothing to do with simulations whatsoever (T57669 on the bug site), especially the associated increase in render times.
Personally I’ll just be happy to know weather I’m just missing some simple step in my setup or not!
As I said; I’ve only been using the Fracture Modifier for a couple of weeks, so definitely not used it in any paid work as yet- therefore firmly in the ‘fun’ department at the moment.
Lastly (forgot to ask previously) do there exist any instructions / docs for the ‘helper’ addon? I’ve googled it and searched this thread etc. but can’t find anything specific docs wise for the addon.
I shall keep trying to break things & await your further responses… Thanks again!
Hey JTA, sorry for the late response. I ended up baking everything and converting to keyframes and then working on separate keyframed objects to achieve what I wanted in regular non-FM 2.79b. As I explained, none of the recommended techniques seemed applicable to my situation so this is how I managed to get around the problem at hand. While working in the FM build I had a lot of issues and quite a few crashes, also in the last part when I was converting to keyframed objects. If interested, you can see the end result here in low res and with low samples rendered as I did not have enough time to render in higher quality: Rendered video
Hey JTA! I have a question: I wanted to do some testing, so I tried to shatter glass. The problem is that when putting a glass shader on it, you can see all the individual shards before they fall apart. My quess is that utilities->automerge/autohide distance should do the trick and ‘merge’ the shards before they are smacked apart. Except, those two fields don’t do anything here. So how should I approach this?
Hi,
make sure to set an inner material (not the same material datablock as the original material, if you want the same material, copy it and select the copy as inner material) in the Fracture Modifier and refracture; otherwise you may tweak the distances. Furthermore you may try with perform merge on vs off, but note this has issues with fix normals then. Autohide dist must be above 0, so that automerge distance has an effect too, btw.
i tried Bmesh and curve and both give crazy results where everything flies away or crash blender. I used mineways to convert the minecraft castle to an obj and chose the 3d printing preset because it gives me a manifold mesh. here’s the blend file http://pasteall.org/blend/index.php?id=52191 I would really appreciate if anyone can help
Ok, i made a sample fracture + simulation + bake + viewport render
Instructions are in the blend. Additional note : Rectangular alignment should be 1.0 for straight cuts in 90 degree angles.
I have some docs in the work plus some updates to that addon. And I’m writing a new addon that gives additional improvements in combination with the pie menu.
I’ve been slow working on several new docs so give me a little bit more time and I’ll start releasing them. Our 2.79x version will be around for a couple more years so we are still in development of support materials.
@BenMu ya, we suggest you use FM 2.79x for destruction work, then prep it to move to 2.8 official version for finishing work.
That test merge version of FM 2.8 really is buggy and broken since it is not even an alpha version. So it’s not a good idea to use it or support it since we would be sending people into a black hole.
So we recommend you don’t use it and that you use the 2.79x stable version of FM then prep your file for use in 2.80 official or higher. That way you get the best of both worlds.
Is there a way to slow down the simulation while keeping the same behavior? I tried to change the speed in the rigid body world but it dramatically changes the behavior of the sim as you can see here
this is the render I’m trying to do, basically 1 simulation exported as alembic then brought back as a copy to make multiple copies and loop them with an offset, as you can see it worked out pretty well but the only issue is the sim speed
Hi,
the videos dont play for me, for some reason. But… i tested your blend.
If you reduce the sim speed, say to 0.5 and have also an animation and the “Animated” checkbox, you need to double the animation duration.
Like in your blend, move the keyframes for x location and kinematic on your dopesheet from frame 15 to frame 30.
If you want to reduce to 0.25, go to frame 60, at 0.125 to 120 etc… (need to make the cached frame range longer then).
This way the “starting” speed will be reduced as well, and the simulations should look more similar.
I think without having a timescale feature on a cached sim you wont get the exact same shard movements in a new simulation. Too bad blenders alembic importer / mesh cache sequence modifier doesnt have such a frame scale feature… hmm.
