Deforming a plate (SPIF)

Hi all!
For a school project i need to simulate an industrial proces single point incremental forming(picture above). To put it short. It simulates a plate clamped at the 4 sides and use a cilinder-a-like tooltip to push it and deformuing it.


  1. It would be great to know how to simulate it with Blender itself. What i tried is:
  • creating a the 2 mesh
  • subdivide and ctrl+a it
  • set one to static opbject, the other in softbody
  • then use logic to move the cilinder towards the plate

Problem: the plate becomes jelly-a-like. I only can clamp the plate at 4 corners instead of 4 sides.

can anyone help me? i posted the .blend file beneath…

  1. After point 1, i want to use python to generate the proces. So creating meshe en subdividing are done in python. Also i wonder, how can i use Python API to check a collision and read out the coordinates of the vertices of collision? To know which vertex i need to move when cilinder and plates collide?

thx in advance


plate and cilinder.blend (180 KB)TestingLogicPython.blend (759 KB)

Sounds like a displacement map and the displacement modifer may be of help.


hey thx, i have just looked at the 2 links. but it seems like it just modify the textures/surface to look in animation?
Or what is it exactly?

You can use an animated displacement map or the strength of the displacement to simulate the deformation of your plate.


sorry i am a total noob. can you provide a more specific link? or make a demo blend file?
i googled around but only found tuts about textures and the page. is too specific for a newbie

Gave it a try with the SB engine.
So this is pretty much off topic for the game engine.
But since you linked to this thread in the PM, i put it here.
Used a force field to fake the tool … speeds up simulation quite a bit. (i get approx 1 sec per frame)
Note also the special honeycomb style model to have the required structural stability.
That will look very ugly with a subsurf modifier.
plate and cilinder2.blend (156 KB)

okay, i am going to study your blend file and reply if i got any question. Don’t know wot the force field do

first of all: what is a sb engine? is it the same as BGE or am i confusing some things?

ps: u will definitly get a thankword in my paper! :slight_smile:

No the internal soft body engine has nothing to do with the GE. It is part of the animation system like fluids or the cloth modifier.
So it won’t do nothing if you start the game.
You have to run the animation. With the play button on the time line or ALT_A in 3d View.
It has automatic “baking”, that is: the simulation is stored on HD.
The next time it is played back it runs much faster.

Could not help to do another one.
small video

Blech1.blend (254 KB)

But the plate is not permanent deformed. In your first file it seems much better, it has less springback i think. I am going to search for literature and play with the parameters under “physics button”

don’t want to sound like a moron who depends everything on others

btw: i suppose the force field is like an object and when it collides with the plate, it deforms? so basically u didn’t animate the whole thing right? (with animating i mean: drawing the deformation yourself) the way i look at it. it seems pretty that the vertices of the plate are moved by the forcefield. and i just need to increase the accuracy and some clamp/stiffness issues right?

It is permanent deformed.
I did raise the resolution such that the tool can emboss a fine line.
An sure it widely depends on the parameters what goes into vibrations and what goes to deform.
Even timing is important. The tool in the second example moves much faster. This is because i guessed your machine is hammering at high frequency.
But be warned don"t get to greedy on resolution and material stiffness other ways calculation time will rise dramatically.

Yes it is simulated.
Yes the force field takes the place of a spherical object but is much faster to handle in collision code.
Slow Bending
BlechSlow.blend (257 KB)

i see… need some testing more.

thx a bunch! ye i know about the accuracy. It is just to create a simpel model to proove that we can simulate industrial processes. now i am running on an amd x2 6400+, 4gig ram. it’s still smooth

keep u posted

shape keys, i think shape keys, if you r going to use the game engine.


Shape key.blend (149 KB)

ok i will postpone the question. my monitor just died. using an old notebook now. going for my guarantee


Small update to be more like SPIF

BlechSlow2.blend (265 KB)

thx. i just tested some parameters. seems the “use edge” parameter pull affects the springs in the mesh i have read. so i changed the plastic deformation on100% and the pull on 0.150. Now if the forcefield retracts it doesnt look like a jelly (which is unrealistic)

going to finish the chapter softbodies in the book i just found (Bounce Tumble and Splash! Simulating the Physical World with Blender 3D)

PS: i think we don need stiff quads because the atomstructure of metals’do not looks like that. but if i enable it does not deform, but i thought it would becomes less jelly. I will test with the damping and bending stiffness a bit
also i don’t really get the honeycomstructure. logically the plate i made in the blendfile is the same no? just the faces in between got removed…

Sorry Hellooo, havent got time to check your blend yet! i will try later on

If you use bending stiffness there is no need to use stiff quads too.
Bending stiffness inserts springs to every second neighbour including the diagonals of quads.
This is also the reason why the honeycomb structure become stable in highly dynamic situations, may be you don’t need it here and the bending stiffness generated by overlapping springs is enough . As I said I first assumed a kind of high frequent hammering.

In the plate you created there is no inner connection between the upper and the lower plane so that the springs inside are not created and there are no volume preserving forces.

First: the tooltip is rotating (depending on the rpm the plate deformes more/less) but this i will only animate and not bring in account for coz it will be too much encoding.

If i am not wrong i think the basic soft body does only create springs on the connections in and between the quads. So mayb if enabled the stiff quads, the plate would be less jelly?
For the honeycomb strucuture. If i subdivide a mesh. and go to edit mode. i can see what u mean. But how to “draw” lines/connections there? Is it possible like autocad or smth? Coz the structure is good but doesn’t seems realistic huh? Though not a big problem.

I read the force field descriptions. It is in fact just a bunch of vectors with the same beginpoint and pointing outwards (inwards) depending on the radius. So i think this match the principle of spif.
But i need to write something about it. You told me there is no collision. But i need to prove (read: write essay) how the principle of this “half simulation/ half realtime” thing works.

Are you certain, i don’t need to dig in softbody.c? Coz there is one fundamental thing i don’t understand from blender. It is opensource with a lot of plugins encoded. But most of it is written in C. How can i know in which folder i have to look up? I supposed now it’s all in the folder source and the rest is irrelevant?

i found this in the code:

/* the most general (micro physics correct) way to do collision
** (only needs the current particle position)
** it actually checks if the particle intrudes a short range force field generated
** by the faces of the target object and returns a force to drive the particel out
** the strenght of the field grows exponetially if the particle is on the ‘wrong’ side of the face
** ‘wrong’ side : projection to the face normal is negative (all referred to a vertex in the face)
** flaw of this: ‘fast’ particles as well as ‘fast’ colliding faces
** give a ‘tunnel’ effect such that the particle passes through the force field
** without ever ‘seeing’ it
** this is fully compliant to heisenberg: h >= fuzzy(location) * fuzzy(time)
** besides our h is way larger than in QM because forces propagate way slower here
** we have to deal with fuzzy(time) in the range of 1/25 seconds (typical frame rate)
** yup collision targets are not known here any better
** and 1/25 second is looong compared to real collision events

basic blender knowledge …
selecting four/tree/two vertices F_key to create a quad/tri/line

Anyhow this thread has nothing to do with the game engine and I don’t really know where it should be.
But I think you should ask the moderators to move it.

There are forces … who needs collision …
Sorting things:
We want to simulate the effect of a rotating ( what ever made of material ) thing to some other thing. We can’t, cause it is far too complex ! We need to go into detail and THINK!
Think about what we need to consider and what is out of scope.
To me it looks like a fail in defining the task. If you are going to miss this, I am willing to argue with your teacher.
Is there any written description of what you should do in that paper?