So for acoustic simulation, geometric complexity means that the process can take a long time. By reducing geometric complexity, the perceptual effect can be minimal, whilst reducing precomputation time. I want to automate this process, and I believe ML can be used for this. I believe I can use a loss function that includes simulation accuracy (still deciding the metric for this) and computation time, and based on the hyperparameters for these terms, the different factors will be prioritised accordingly. However, I am struggling with understanding how to input a 3D model into the system; I know from a previous answer (How to input a 3d model into ML algorithm?) that you can simply arrange it similarly to a 2d task, but how can I formulate a model that outputs an edited version of that same 3D model?
Asked
Active
Viewed 26 times
