Preprocessing of vector fields
Closed this issue · 3 comments
Since we aren't using hardware acceleration for the construction of the vector fields we probably have to fetch them from preprocessed images if we want live demo.
That depends I guess, we can still generate simplex noise, create a vector field from it and apply the curl operator, write to a texture (uploading it to graphics memory) and re-start the simulation. Maybe you are referring to the amount of time it takes to re-generate and that maybe it isn't feasible to do up on stage? Have we measured how long it takes to re-generate the curl-noise vector field? A small hitch (1 s to 2 s) is fine in my eyes (as long as we have a button where we can change several settings at the same time).
Anyway, let's discuss this on the next meeting. But you are right though, if we can't do this in real-time a good bet would be to write them to images and load them. Nevertheless, it could be beneficial to have a this feature even if we can do it real time, in case we would like to modify the vector field directly (e.g. to add other static objects).
Not relevant any more given the restructuring of evaluating the vector field on the GPU for each particle.
Sounds perfect! I'll go ahead and close this then.