# Help on simulating a hammered bronze panel

I’ve been trying to model a hammered bronze panel, similar to the photo. The pattern is quite complex, and so far I haven’t gotten close:

• Voronoi doesn’t work well because the bronze cells have curved sides and I can’t make the sides of adjacent Voronoi cells bend in multiple direction.
• I tried random variable radius sphere distribution, then solid difference but the spheres need more deformation and I’m not getting the spiderweb borders between the cells (2nd photo).
• I thought maybe simulate the hammering action, tried Kangaroo 2 for the first time, used the pressure, volume, vertex load, solid/point collision with spheres dropping on a mesh, see test below. The resulting meshes are harder to trim/boolean and they have odd peaks (3rd photo).
• Maybe setting up a recursive simulation with Kangaroo & Anemone with repeated sphere drops so that a dent is further deformed by subsequent drops, like hammer strikes? As a newb this is deep waters, so I’d love some help on how to best approach this so I don’t go in the wrong direction.
I appreciate any help.

After dealing with what cells to take (random acc some probability value?) … get a Mesh (per V cell : either using the OEM Polyline or some distorted one) and relax it via K2 using some Unary (pos Z) Force. Control the Vector3d amplitude via some sort of formula (say proportional to V Cell Area). Meaning that the small cells are “inflated” less.

That said the more the Mesh is “uniform” the better the results: a classic way to do that is to create a Mesh using the Polyline Center point (and some rational subdivision value - avoid huge N of Vertices that could yield huge Elapsed time). The better way is to create a Mesh like healing a hole - NOT the way that R does it. On the other hand if the V cell has small edges (the V anathema) … well … you can undrestand what would happen.

Plan Z: you can forget Meshes/K2: there’s the good old cheating way as well (real time fast - I highly recommend it):

Hi @connie