Fluid testing help

I got a chance to do some serious blender fluid test. I barely do any full blown test on blender’s fluid. It’s a city flooded by a tsunami. So here goes.

The noised plane should be an obstacles and the boxes. But why does the fluid floating like that?

Also once I got two layers of collision without any collision object on it. Any idea what’s gone wrong?


Increase res and scale the scene up a bit, will increase baking times but it’s worth it!

Well, in realflow the workflow is like this : work on lowest res as possible that is representative then work slowly to increase the resolutions.

In this resolution even the collision has gone significantly wrong. I don’t need higher res to see that it is wrong already. I made both preview and final resolution the final resolution is 100 and both got this kind of collision. It collides with the boxes without any problem. But with the base it collide far from it.

So any idea?

For each object (domain, fluid, obstacles etc), in object mode try Ctrl+A. See if that helps.


nope. doesn’t help much. still got the same problem

I’ll try test it on another kind of scene… I need a better understanding of how the parameters would work.

But just in case you want to take a look on what’s going on I’ve attached the .blend


citytest.blend (430 KB)

As I said earlier Ctrl+A (Scale and Rotation to ObData) for all objects.

Your simulation world-size is just 3cm, you need to increase more to the 10m maximum. Also reduce the simulation time to much less than 100secs in the domain settings if you want a more realistic simulation. You need to fake a city simulation like this.


Ctrl + A doesn’t change anything much. I know it’s for freezing/applying transformation.

I think I got it wrong at simulation time. Now it collides perfectly. So it’s more of like substeps in realflow right? Meaning it calculates each frame n times.

In that case, how should I stretch/squeeze the simulation animation to match the scene?