Introduction Physics-based techniques enable more accurate simulations to be produced [1]. We present an approach for easy automatic model creation and efficient GPU-based finite element (FE) simulation of gross and fine-scale soft-tissue behaviour, including skin wrinkles. Focussing on the forehead region of the face, we simulate both large areas of soft tissue, and detail such as skin layers (necessary to produce wrinkles [2]), unlike current approaches [3]. Our approach could be used in a variety of fields, from animation to surgical simulation, and it can be used to simulate any multi-layered soft body. Simulation Approach Our simulation approach involves three stages: creating the multi-surface mesh, creating a suitable simulation model, and simulating the model over time. Any surface mesh can be used. The model creation system then automatically constructs simulation-ready non-conforming (voxel-based) hexahedral FE simulation models with bound surface meshes [4]. This includes automatic computation of skin layers and material properties, muscle properties (such as fibre directions) from NURBS volumes, and boundary conditions. Our GPU-based non-linear total Lagrangian explicit dynamic (TLED) FE system includes procedures for muscle contraction, and to enable the non-conforming elements to slide along a surface, such as the skull or deep tissue boundary (a phenomenon often neglected [5]), facilitated using a GPU-based semi-brute-force broad-phase collision detection algorithm. The solver has also been optimised for use with our non-conforming models [4]. Figures 1 and 2 show some example models and simulations. Conclusions Examples have demonstrated the complexity of models that our approach can generate, and their ability to simulate gross and fine-scale soft-tissue behaviour. Improvements could be made, for example, by using shell elements to better model the thickness of the epidermis. Future work will focus on simulating different-aged skin, and using different material models.