Hello,
I know I talked about this before (on odforums), but I always found it strange that houdini doesn't have a gas "swirl" node and instead has a gas disturb. Gas disturb certainly works well under the right conditions, but it seems to have drawbacks. For example, Gas disturb works really well when you have fast moving sims with lots of movement/other forces to break up the shape. On the other hand, if you have a slower smoke sim where the main source of movement or detail comes from the gas disturb (with no other forces to break up the shape), the results don't look very good. You pretty much literally see the block size in the smoke.

In my first photo it is just a cube being sourced into pyro solver with gas disturb. You can see the blockiness created by the gas disturb. Pretty much no parameter on the gas disturb node prevents that shape.

Even with proper sourcing, if your sim slows down you risk getting this "cross" grid artifact look in the smoke. This worsens when you up the substeps. In my second photo you can see a cross like effect in the smoke. Even if you up sim resolution, you still see it. This cross like effect happened in areas with low velocity (or areas where the smoke was pulled into another direction) and lots of substeps.

I have seen better true swirling motion in other solvers like bifrost aero or even the old maya fluids. They have settings to add swirls as well as and randomness like gas distrub. You can get nice swirls and eddies pretty easily instead of the details you get from gas disturb. Houdini has nodes like gas vortex confinement or gas vortex boost but non of them seem to get easily controllable swirling motion. I have had people say to just build one yourself, but this seems like a basic motion you would have to control the smoke. Anyway, any thoughts? I was thinking it would nice to see a node that better controls swirling motion.

after messing around more, you can still get the "cross" grid artifacting even with out distrubance. It seems to happen the most in areas where the smoke is going one way, but then gets pulled in the opposite direction.

I actually found three things that fix the grid artifacting.

1. setting the advection scheme to BFECC

2. changing trace method to the most accurate

3. The best IMO is to set the blend to a small amount. The documentation says this helps get rid artifacting and it works really well!

edit: I paired a bit of blend with the advection scheme too BFECC and that seems to work the best.