Subject going through a sheet of stretchy cling film, any idea how? + dForce experiment

BlueFingers

I'm trying to get the effect of a character trying to break through a sheet of cling film type of material. So far I have done some experiments with dForce, and though it's been interesting I haven't got the results I wanted. As an indication of what I've been trying to do here are some illustrations.

Here is the starting situation:

First tries looked like this:

So I added subd modifier, a smoothing modifier and made the surface simulations settings even more strechy, less stiff and "light" with less friction and then I got this:

Now the character breaks the fabric and it doesn't look like a character trying to breakthrough cling film (I want the material to stay intact). Though I'm not totally happy with the result, I'll take what I can get, does anyone have suggestions what I would need to do dForce-wise to get the effect I want, of perhaps have an alternative solution to get the cling film effect? A solution in post would be interesting as well.

Paintbox

Maybe magnets at the corners? You can make pins at corners so it doesn't drape so much and might even stretch. Just a thought.

BlueFingers

Paintbox said:

Maybe magnets at the corners? You can make pins at corners so it doesn't drape so much and might even stretch. Just a thought.

That is a very good idea :-)))  I definitely going to try that out! Why didn't I think of that, thank you so much!

BlueFingers

So I experimented a bit more and at this moment and these are my best results up to now. I'm quite happy with the progress as it looks a lot better than my earlier experriment, but at a certain point the subject just breaks through the mesh of the "cling film". I already minimized the stretch stiffness, and experimented with all sorts of settings on the contraction-expension ratio, but I can't keep the subject of breakingthrough the mesh. Does anyone have any suggestions to prevent that from happening?

As you can see the mesh itself stays intact:

hansolocambo

A few things to do to make a dForce simulation more precise :

1/ slow down simulation and improve collisions :

** Simulate at a lower frame rate than the animation : Simulation Settings Tab > Simulation > Frames Per Second (FPS) Multiplier: higher values up to 10 will make the simulation more accurate. 5 should give better results.

** Increase the number of times collision is attempted per frame : Simulation Settings Tab > Simulation > Subframes: higer values up to 16 will improve collisions.

2/ do not use the Genesis with clothes and hair (!) as is :

The BIGGEST issue with Genesis is the objectively terribly nested wireframe around the nail area. It's a glorious modeling mess in there. Problem : simulation struggles against nails because of that bad geometry... until it fails. The same happens when importing a Genesis in the Rolls Royce of soft-body simulation : Marvelous Designer. Garments stay stuck in Genesis nails...

You could use a third app (Instant Mesh, Blender, ZBrush, etc) to remesh Genesis using a voxel-based volume (ZBrush Dynamesh, Instant Mesh, Blender Voxel Remesh, etc). This way you end up with a simplified mesh that still has the overall shape of your Genesis but with a much ! simpler and very clean geometry. Which means a tremendously faster simulation and no more collision problems because of nested polygons. You could try to export the Genesis to a free quad-based autoretopology tool such as the useful wjakob/instant-meshes to do that (although I doubt it'd be able to compute hairs).

Simplified collision objects for physics simulation is as old as the idea of simulation itself. In any game, any movie, physics are calculated on proxies, lower polycount versions of the originals.

3/ You could also smooth (Blender, ZBrush,etc) polygons in the nail area to get rid of the nested polygons. Which would avoid that collision issue.

4/ you could choose some gloves if you have that in your clothes bundles. And put gloves on the Genesis (if you don't know any third party app to simplify those nails). This way dForce will simulate the contact on gloves rather than nails.

nonesuch00

I think maybe null nodes parented on the fingertips & after you position the clingfilm sheet, 5 null nodes there, then parent the clingfilm nullnodes to the fingertip null nodes & the simulate. The clingfilm needs to be configured not to go through the figure or clothing mesh & I'm not sure how. I think you make the character sim layer 1, clothing & hair, layer 2, and then clingfilm, layer 3.

rosseliani

Try to use a much more dense mesh for your plane (sheet film), I think you reach the dForce limits due to the extreme deformation you're looking for. Also try to add a dForce Modifier Static Surface to your character and reduce the Friction in the character's Simulation Surfaces Parameters.

 

BlueFingers

@hansolocambo : Wow what an elaborate answer and friendly offer, that was more than I was expecting. At this moment the scene I'm working with is purely experimental and not meant for a final render, once I understand how to make it work correctly I'll start constructing the final scene.

1. Great tip, I wasn't aware of this, and will give it a try right away.

2. Okay, this will require some learning on multiple levels from my side but I'm pretty keen to get in to it. Though I have to admit I feel a bit out of my depth here, I'm still only making pretty donuts with Blender. - This was actually way simpler than I thought, I made a low poly version and am going to test it -

3/4. Interesting, wouldn't a geoshell do the trick? I'm going to ty that out. - Geoshell keeps complexity of the mesh and doesn't work it seems. -

PS: Wow that is so friendly, when I export the scene as .OBJ, would the Blender measurements work for you?

@nonesuch00: I haven't worked with null nodes yet, would you have a link where I can learn more about null nodes? The sim layers are a great suggestion, I'm going to try that out.

@rosseliani: Ooh, the plane is already pretty dense as it is and I have applied a SubD modifier and Mesh Smoothing modifier to it, but I'll certainly try adding more polygons. I already applied the dForce Modifier Static Surface with minimal friction to the subject. Thank you so much for the suggestion and thinking along.

I haven't mention it previously but perhaps it's worth knowing that I toggled "Self Collide" off in order to prevent it from exploding, good thing was that it made the simulated plane look a lot more like cling film.

PS: I just found the Daz tutorial on nulls.

BlueFingers

Update:

I made a low-poly version of the G8 figure without nails and the lowest dForce friction modifier,

I increased FPS modifier to 10 and the amount subframes to 16, results remain similar:

I haven't tried increasing amount of polygons on the plane yet will try that next.

 

nonesuch00

BlueFingers said:

@hansolocambo : Wow what an elaborate answer and friendly offer, that was more than I was expecting. At this moment the scene I'm working with is purely experimental and not meant for a final render, once I understand how to make it work correctly I'll start constructing the final scene.

1. Great tip, I wasn't aware of this, and will give it a try right away.

2. Okay, this will require some learning on multiple levels from my side but I'm pretty keen to get in to it. Though I have to admit I feel a bit out of my depth here, I'm still only making pretty donuts with Blender. - This was actually way simpler than I thought, I made a low poly version and am going to test it -

3/4. Interesting, wouldn't a geoshell do the trick? I'm going to ty that out. - Geoshell keeps complexity of the mesh and doesn't work it seems. -

PS: Wow that is so friendly, when I export the scene as .OBJ, would the Blender measurements work for you?

@nonesuch00: I haven't worked with null nodes yet, would you have a link where I can learn more about null nodes? The sim layers are a great suggestion, I'm going to try that out.

@rosseliani: Ooh, the plane is already pretty dense as it is and I have applied a SubD modifier and Mesh Smoothing modifier to it, but I'll certainly try adding more polygons. I already applied the dForce Modifier Static Surface with minimal friction to the subject. Thank you so much for the suggestion and thinking along.

I haven't mention it previously but perhaps it's worth knowing that I toggled "Self Collide" off in order to prevent it from exploding, good thing was that it made the simulated plane look a lot more like cling film.

PS: I just found the Daz tutorial on nulls.

Basically make the nodes "static" not dynamic and the friction should be near 100% at those nulls to keep the cling wrap fixed to the fingertips and then drape all around the the non-null areas as the simulation runs.

a)

b)

BlueFingers

I had to take a bit of a break, one can only endure so many dForce simulations before going nuts, but I'm back running the experiments.

I have now completed the experiments with a much denser mesh for the cling-film plan, over 3 times as dense and these are the results, I again used the low poly figure I made, and kept the FPS modifier at 10 and subframes at16:

The results are only marginally better in terms of stretchiness, but not what I was hoping for. Also the mesh is buckling under it's own dForce weight (see bottom of the plane) though it is the same material interms of dForce variables, and the dimensions are roughly the same.  I tried lowering the Density GSM even lower than 12 but I was getting a "Unknown error executing clFlush" error with very light materials. I have a pretty up-to-date system with plenty of RAM and VRAM, but it was not liking the high poly count of the cling-film plane (+ SubD and Mesh Smoothing iterations) and navigating the scene in wire frame was just short of a slide show. So I'm starting to think I'm just reaching the limits of what dForce can do.

Next I'll start experimenting with the Null Nodes to see if I get a bettter result.

RGcincy

Thanks for alerting me to your thread. How's this? (Uses Geneis 2 male.) Hands need a little work.

 

BlueFingers

@RGcincy: Wow, that looks really good, thanks for sharing this! It's way more extreme than what I was looking for, but that is great as what I'm looking for must be possible. I'm amazed the mesh kept up with the deformation as I was sure I had pushed the deformation to it's limits. I'm also kind of curious how the non-affected area of the plane kept it's form, not that I'm looking for that, but it means I think you used a different method than I did. If my coming null node experiment fails, I was going to try something with Mesh Grabber. Could you shed some light on how you achieved your results?

BTW: You've got some very interesting threads, definitely going to look into 360 and stereo renders.

RGcincy

I'll write up a more complete explanation and post it to my dForce thread but here's the basics:

1. The plane has 250 divisions
2. dForce settings are 0.01 stretch, shear, and bend stiffness. 100% Buckling stiffness and 0% Buckling ratio. Density is 600.  
3. Gravity set to 0.
4. Animate figure so it is close to but not touching the plane at frame 0, then move the figure into the plane to whatever depth you want at frame 15.
5. Run a 0 to frame 15 custom animation.
6. The plane also had a dForce modifier weight node with the outside border set to an influence weight of 0. That keeps the edges fixed. This is optional.
7. Nothing was done to the figure other than setting the pose and moving it.