Hello, Our goal is to simulate cloth to perform tasks such as folding and spreading. We have made a deformable cloth simulation, and attached a mocap to one of its corners. We are using this mocap to move one corner of this cloth, to observe how the simulation behaves. Problem: The main concern so far is that cloth does not look very realistic. ( We have attached picture and a video) 1. As one can see in the image attached, the cloth has some unrealistic form. And it is stable in that form. At some places it looks like its hanging without support. 2. When we lift one corner and take it to diagonally opposite corner to fold, the cloth forms lot of ripples. (See attached video) 3. Low damping of 0.001 reduces the stiffness in the cloth, but cloth seems like it has no weight. Increasing damping upto 0.1 makes it look like it has weight (when lifted), but the cloth doesn't fold properly. 4. The cloth seems to have no tautness due to gravity when hanging. It looks rigid at many points. Questions: 1. How do we make the cloth more realistic? Are we missing on any parameters? (we have played with most of the parameters in the documentation) 2. Can we improve the cloth simulation any further? or current simulation is best we can get? 3. Can grid cloth work better for this case?
The cloth object can indeed have unrealistic dynamics, because it is implemented as a kinematic spanning tree with additional constraints. The grid2 object tends to have much more realistic dynamics, so you should definitely try it. The drawback of grid2 is that collisions are limited to spheres at the grid vertices, and so smaller objects can penetrate. The next release will have a new object type combining the physics of grid2 with the more continuous geometry and collisions of cloth. But for now there is a trade-off.