Does working in the correct physical scale make a difference?

[mightcouldb1]

Just curious if this physically accurate stuff means anything. Lately I've been trying to model everything according to its real world scale. This also allows me to not have to scale the gravity.

Any opinions are welcome and appreciated.

[Marc]

In my experience it makes a huge difference if you're simming volumes (fluids, smoke etc.). Particles not so much since it's easier to cheat the motion.

So generally speaking I'd say it's best to work in real world scale just to make things easier for yourself .

[mightcouldb1]

Thanks for your input marc. I have a curious problem now.

I have a model of a beaker(thin glass) that I am using as a static object in dops. When the object is extremely large (10 meters), the SDF collision volume will calculate over 20 times faster than the correctly sized beaker (15 centimeters).

Any clue as to why this would happen?

[mightcouldb1]

Anyone?

[Netvudu]

I can´t speak much for SDF simulations, and my Houdini fluid simulations are still too few to get strong conclussions, but from past experiences with Realflow simulations I can tell there IS indeed a difference between working on different scales, and not every time is in your best interest to use realworld scale.

For instance with SPH (at least on RF as I said before) if you simulate on very little objects, you frequently get unwanted vibrations that require lots of substeps to avoid. Too big scale simulations requires lots and lots of particles to get a fine mesh surface afterwards. So, there seems to be a sweet spot in-between for many simulations. Of course, if you devoid too much from the actual real world scale, you might get a nice simulation...which looks nothing like reality. there seems to be a compromise there.

If you have enough CPU power (what´s "enough"?, I hear you say ) you might go every time for real world scale, but in my past experience 1 meter tall glasses of water tend to be easier to simulate.

PS: and not to drift away so much from your question...no idea why SDFs would take that much longer when scaled down...

Edited March 30, 2010 by Netvudu

[mightcouldb1]

Yeah, I did a test this morning to see how long the cook time was depending on the scale. The calculation time goes up quite a bit as the object gets smaller, especially for fairly complex geometry. I can post some results if anyone is interested. For now, I am multiplying everything by 10 and my cook times for the SDF are reasonable.

I'd like to know why this is happening if anyone has any clue.

Edited March 30, 2010 by mightcouldb1