pub fn render_parallel<A: Attractor + Send + Sync + 'static, T: ColorTransform>( renderer: &mut ParallelRenderer<A, T>, config: Config<A, T>, jobs_per_thread: usize ) -> FinalImage
jobs_per_thread. If you get uneven images with low iteration counts, try
At relatively low rations between iterations and pixels (<50), this isn’t much faster.
Because strange attractors are inherently chaotic, we can render multiply images and then add
Internally, this uses three textures,
count for the number of visits to the pixel,
for the velocity at the closest visitation, and the
zbuf which can give information about
which pixel is the closest. The
zbuf also gives us the depth texture.
When combining the rendered image, we have to combine all of these.
When rendering through
render without resetting the
Runtime, this is what naturally
happens. When we use multiple threads however, we have to explicitly combine the runtimes
consistently to what the
render method implicitly does.