Simple scene questions

[Filip Kaczorek]

The scene is simple but has 1000 frames so debugging it is tedious.
So I'll ask a couple of questions here hoping that someone had this problem and solved it

The scene: simple source emitting only smoke (animated strength) and a rather strong side wind balanced with the velocity damping in FFX ( basically a chimney smoke).

1. I noticed that changing the grid resolution changes the simulation. ie.:
- cell size 20: in 100 frames the smoke goes all the way to the other side of the frame
- cell size 10: in the same 100 frames smoke reaches only the half (middle) of the frame width
- cell size 5: reaches about a quarter
It's as if a unit for velocity or strength was a single cell not a real world unit - is that correct?

2. I had the scene animation range 0-200 frames and left the simulation for night on range 0-1000. When I came back today and changed the animation range to 0-1000 to view the simulation - it looked good till frame 200 and then something strange started to happen with the smoke shape and even parts of the smoke seemed to freeze in space, while other just passed them by.
When I resimulated with the animation range 0-1000 the problem did not occur.
Is there a bug with simulating out of animation range. I'm just asking as I was unable to recreate it in a shorter scene

3. The simulation up to frame 200 looks great, good clumpy detail, but afterwards the detail seems to elongate, as if a horizontal direction blur was added. The velocity of the fumes seems constant, so the smoke doesn't seem to speed up there (no animation on wind or velocity damping).
Is it possible that the smoke travelling trough the grid creates (through heat or pressure) some sort of a tunnel that after a while influences the shape of the smoke?

Like I said general questions here as submitting the scene would ask everyone to simulate a 1000 frames sim

[Filip Kaczorek]

Also it seems that the Pause button in the simulation window stopped working

Stop and Cancel do what they did, but after clicking Pause the button changes to resume but the simulation continues

[Filip Kaczorek]

Ok I give up.
I played a lot but all I can do is "move" the moment when the detail disappears
Heres is frame 113:




And here is frame 170:



You can see how the detail just disappears - gets smudged out horizontally.

Here is the scene:
http://www.sendspace.com/file/hjnq3w - Max 2009

Can someone please take a look and give some suggestions - I just want to retain the detail from the first 4 secs of the simulation

[Rif]

do you have difusion ?
I dont have 2009 .. cant open the scene..

Do you use motion blur ?

not sure but you have to expect grid changes affecting smoke behaviors right ?

[Filip Kaczorek]

Sure I have some diffusion and no motion blur.
I'll experiment with diffusion though.

Strange thing is that this change occurs everywhere at once. It's not gradual that at some point the smoke is emmited with less details. At the same moment details dissappear in fresh and old smoke. Like someone did a fade between two simulations???

[Capitanred]

Is it possible that the smoke travelling trough the grid creates (through heat or pressure) some sort of a tunnel that after a while influences the shape of the smoke?

that makes totally sense to me.
I had that problem with explosions when you want to make mushrooms or bubble like fire rolling up. the first one is easy, but if you try to do a second one at the same place, it does not work. I also tought that it's because after the first one there is this tunnel of velocities that drags instead of slowing down the outer parts of the fuel and smoke.

[Filip Kaczorek]

Tried dissipation and difussion, none of them really affects the issue

[sitnisati]

I'd said it's the way how wind is applied.
Wind is adding velocties to the grid and without falloff, smoke would accelerate.
This stretch seems to me that the most stretched part moves at the fastest speed in the grid,
probably because of the wind.
We will try to deal with this and add another way to compute wind.

Kresimir

[Filip Kaczorek]

Yeah but isn't velocity damping supposed to counteract the wind - like we do with normal particles?

It seems to work because the cloud as a whole moves with a constant speed, so maybe it's just on the details level?