Hello!

I know conventional wisdom is to leave the Unit Length preference at 1m, and scale things as they enter/exit Houdini in the pipeline (but never scale cameras of course, etc.) I'm wondering if, in the modern 17.5 era, is this still true? Particularly for the more recent solvers (Vellum, FLIP/Gas microsolvers).

Has anyone tried again recently to use Houdini in a pipeline at a Unit Length at non-one? (dm or cm in particular)

Thanks!

I find none of the solvers are really physically accurate enough for units to matter. (except for gravity) In one case meters might help, in another it might complicate things. Precision issues are often the overriding problem. It's often helpful to simulate things at the origin, regardless of the units.
Working with meters can be a huge drag depending on the subject matter. If you are making macro shots of small insects for example, all the parameters for noises and point cloud radius and other spatial properties will be so small as to relegate user input to manually typed scientific notation.

I prefer to work in whatever units make life easiest. It's usually not complicated to reverse a scale on export.

Edit:
I want to add that the viewport does not like sizes that are too small or too big. Meters is actually one of the worst units to use unless you are working only with building sized objects. The default minimum home radius is a tenth of a meter, which would make objects that fit into your hand almost too small to work with.

In addition to what jsmack is saying.

When I model for 3d printing, my models are typically around a 10x10x10 cm finished size product.

For such products I set the default hip to mm.

The only reason being is that I need to take real size life references (measurements taken with a caliper).

So when I take those measurements I don't have to do any conversions(keep ‘accidental’ mistakes on measurements to a minimum) while working in Houdini or when setting up my file with the 3d printing vender ( they only see ‘units’, and in my case I just specify mm to correspond to the same as with Houdini).

I've been mostly working in dm units for several years without any issue.

Thanks Jonathan! I agree with you, I would rather work at a unit scale that makes sense for the subject and the models. If it's solely a question of precision, and the solvers will act reliably at any scale, I'm in. I suppose it's mostly a question of making sure all distance-based paramters are adjusted properly to the scale in DOPs.

I would be surprised if all simulation nodes scale properly anyways. I don't think SideFX has done due diligence to ensure that is the case…