You can see how to replicate this in Houdini by Examples/fractal.hipnc, although it's super slow and you might have noticed that taichi inits every frame. You will find that the whole project is behind the idea that we combine the flexibility of Houdini and the performance of Taichi, merely with more parameters to set, and optimizations for the sake of performance. If you open the fractal.py, you will find that the only thing we did is to pack an existing taichi demo solver into a ti_data_oriented class, but leaving a parameter for the front-end, in this case, the Houdini, to set. From fractal import fractalClass f = fractalClass( 512)
0 kommentar(er)
