Question regarding slow down behavior - strange movement

[olika]

Hello phoenix friends,
i have a question regarding the slow down process.
At first there is the regular sim of the dust the bike creates (hopefully attached animation is working)
Is used adaptive grid for covering the long dust trail the bike creates.
I already ended up with about 100 mio cells in the end when the dust really expands.
So therefore increasing size in the resim for wavelet and stuff didn`t work out cause then it was like 600 mio cells with amp res factor= 1.
So the resim i did was: amp res factor =0, because of not increasing grid resolution i choosed the interpolation method and the slow down mehtod in the time bend controls.
For the time stretch i switched from 25 frames in max to 240 frames (maximum i think) so you have a change about factor 0,104.
All good to this point, but:
I you take a look at the rendered slo-mo you`ll see that the smoke somehow follows the bike now in a unusual way.
In the base sim the smoke is created through particles from a small plate underneath the tire of the bike, so they "Stick where they`re created", i mean they don`t move, which is wokring fine in the base sim
but in the slo-mo the whole sim is somehow following the bike, huh?
Don`t have a clue what is causing this, maybe someone encountered the same thing before? Tips?
Thanks in advance!
Btw it`s great fun playing with phoenix!

[Svetlin.Nikolov]

Hey, would it be possible to share your Input rollout settings and how your timeline looks? Do you have any custom Start frame under the Simulation rollout? Not sure how the base sim would look, but is the issue that the resimulated animation is just offset from where it should be?

[olika]

Hi Svetlin, thanks for reply!
See new gif, all i changed was going to interpolate instead of wavelet ( amp is =0 so it shouldn`t make a difference right?) and changed the slow-down method to general purpose,
and hei! it`s not moving with the bike anymore so that looks good to me,
but now it`s skipping somehow from frames to frames, looks like it has it`s problems between the origin keyframes.
Will try to simulate now with interpolate and slow-down method on...

[olika]

[Svetlin.Nikolov]

Hey,

Indeed - with amp=0 the interpolate/wavelet options are ignored since the resolution does not change. However, if you don't have any special starting frame, re-scaling should work. Which version are you using? Can you grab a very latest nightly and try it this way? There were some issues with the general-purpose method in 3.04, which were fixed a few days after that and are available only in the nightlies. However, if you still have issues, would it be possible to send the scene or simplified version of it that is still able to reproduce the issue to Support?

Thanks very much!

[olika]

I use v3.00.02 for the simulation.
?hm could it be i`m too stupid to find the nightly?
when i log in, there`s only nightlies for vray and vray blender?
I?m sorry i don`t see any phoenix nightlies

What do you mean with a special starting frame?
I have that biker driving in realtime (25fps) from frame 0 to 150,
then somewhere in the middle i want to switch classic into the slow motion shot.
So therefore i stretched the timeline from 25 fps to 240 fps which results in that the section i want in slo-mo is between frame 1000 and 1150.
That means the resim of the origin frames for slo-mo is starting at frame 1000,
does this mean we have a special starting frame?
I?m afraid resimulating the whole in slo-mo from beginning couldn`t be the solution right?

[olika]

Just on more thought,
could it be that the nightlies show up account related?
Had that problem with the forum in here too,
cause i bought phoenix later than vray, you guys had to update my account settings and integrate phoenix too so i was able to post in the phoenix forum.
Maybe it`s still not connected to the nightlies? Just a thought...

[flipbook]

There's a sticky thread explaining how to get nightlies :P

[olika]

?hm thanks i know i know but see:

[georgi.zhekov]

Just send a mail to support@chaosgroup.com and they will help.

[olika]

Hi again,
now everything resimulated with the newest nightly 3.04.01.
Seems a bit better for me, now it seems like when you look on the left side away from the biker the smoke behaves fine,
i assume now maybe it has something to do with the particles emitting the smoke?!
Still used the general-purpose method for it, will try a slow down again now with the nightly

[olika]

Hi Svetlin,
just send you the stripped file ove the support adress,
hope that helps
Thanks in advance

Still have no clue why, even with the nightly, when slow-down method is used,
the smoke is pushed with the bike direction as it would be linked to it ;- (

[Svetlin.Nikolov]

Got the scene, thanks! I hope in a couple of days I'll get there and will be back to you with info or eventually fixes

Cheers!

[olika]

Great! Looking forward what you will find.
after a night of sleep i assume it`s maybe the base sim which is already a bad base for the resim?
Cause in realtime everything is quite fast, so the particles have only 1-3 frames to emit.
Maybe thats so short, fast and rough that the smoke must come out so puffy (does this word exist? ;- ))
In the realtime shot a lot of it just dissapear with the speed and motion blur.
In the meantime i find it`s really an interesting task ( haha i thought in the beginning aah c`mon, let`s add add bike with a smoke trail quickly)
not only because of the large gridsize you end up.

[Svetlin.Nikolov]

Hey, back with some info:

First off, I should have noted earlier that a slowdown with a factor of 10 is very likely to produce issues since there is too much extra data to be created between the original frames and this will imminently lead to various artifacts or uneven movement of the fluid. There is no exact lower limit, but usually if you go below play speed of 0.2, it's very likely to have issues. Be sure to check this topic on the docs site: https://docs.chaosgroup.com/display/...-TimeBendResim

The reason for the flickering with the General Purpose method is that it would place an exact frame from the base simulation roughly each 10th frame and figure out the other 9 frames inbetween without interacting with the scene. This way, the moving particles emitters will affect the grid each 10th frame of the resimulation and new smoke will suddenly appear all over at that moment and be carried around the next 9 frames.

The Slow Down method, however, seems to behave strange indeed, which was the original topic of the thread. You should make sure to use just one step per frame in order to get less diffusion and loss of detail. However, it seems that despite that, the smoke lags behind and we'll try to debug it and see what's the reason for this.

Will keep you posted!