In a smoke simulation, what is the difference between using the Smoke High Resolution checkbox, as opposed to simply increasing the resolution? I understand that for some mysterious reason, the limit of the resolution is 512, but if I'm doing a simulation that won't need that many divisions, is there really a difference? I've noticed that if I bake a fire simulation without Smoke High Resolution enabled, it will usually only use one CPU at 100% at a time, whereas if I do enable it, all my CPUs are at 100% (and it takes a whole lot longer to bake).

  • $\begingroup$ @cegaton Okay, I think I understand that, but what does that link have to do with it? $\endgroup$ – Anson Savage Mar 25 '16 at 21:17
  • 1
    $\begingroup$ oops... something else on the clipboard... $\endgroup$ – user1853 Mar 25 '16 at 21:30
  • $\begingroup$ blender.org/manual/physics/smoke/baking.html?highlight=bake $\endgroup$ – user1853 Mar 25 '16 at 21:31
  • $\begingroup$ @cegaton Oh, okay $\endgroup$ – Anson Savage Mar 25 '16 at 21:31
  • 1
    $\begingroup$ There's a bit info in the manual too. $\endgroup$ – Mr Zak Mar 26 '16 at 12:18

Turning up the base resolution increases the actual number of voxels used to calculate the sim. The high-res divisions are a separate, secondary effect. They are applied after the sim itself calculates for that frame. High-res subdivides the voxel and adds additional noise based on the behavior of the low-resolution base plume. It's essentially a post-process displacement effect, just a more intelligent one than, say, globally mapped noise.

The advantage to this you can take a blocky, low-res sim that is fast to calculate, and smooth out or fluff it up as though it was higher-resolution. The strength value controls the shape, lower values smooth the plume, higher values fluff it up. This way, you only have to raise your base resolution enough to get the basic movement right, you don't have to keep cranking it to get the detail and smoothness you want. Since you apply that as a post-process when using the high-res function, which means faster bakes. And it's much easier to iterate the design of the base sim, because you don't need to sim at full resolution all the time.

It also helps out with an annoying behavior a lot of gas simulators have where increasing the base resolution can radically alter the behavior the plume. So you get a simulation at a lower, easily testable res, then you crank up base-res for the final bake....and 2 hours later, the result behaves nothing at all like your test bakes. High-res doesn't have this issue, because it doesn't affect the underlying motion, just smooths and adds detail.

If you want more information, you can read the paper that all these fluid-upres techniques are based on: https://www.cs.cornell.edu/~tedkim/wturb/wavelet_turbulence.pdf

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.