Flip "depth" water

[Sophie]

Hi,

I am working on a flip simulation of a rapid for a school project. I noticed my water only builds up a super thin layer of particles, so it doesn't really fill up the river bed. I tried making the particle separation different in the emitter and the dop network, but that didn't work. Someone also said increase the emission rate. To do that, I tried to increase the scale source volume, but that also doesn't do anything...

Any ideas?

[chrisdunham95]

I usually have to run a sim like this for a few hundred frames previous, so the build up takes place like it would in real life. 

OR you could create flip particles from the river bed shape to fill it straight away (flip from object) and then emit flip particles also to get the river flowing correctly. (if you provide your scene file I can provide a quick example inside for you if it helps)

Hope this helps,

C. 

 

 

Edited April 6, 2018 by chrisdunham95

[Sophie]

I did run it a few hundred frames before and then gave it an initial state and let it run again, but it didn't do a lot. I have to admit my particle separation was a bit bigger there. 

I attached the file now as well, thanks  

ProjectBlockC.hipnc

[StepbyStepVFX]

Hello (bonjour ?),

I think they show examples of what you are trying to achieve in those video.

Maybe it can inspires you ?

 (Around 1h20)

 

[StepbyStepVFX]

 

1 hour ago, Sophie said:

I did run it a few hundred frames before and then gave it an initial state and let it run again, but it didn't do a lot. I have to admit my particle separation was a bit bigger there. 

I attached the file now as well, thanks  

ProjectBlockC.hipnc

A newer one, where he explains the method described by Chris above (around 15 minutes)

 

[Sophie]

Thanks for the video's! Now that you mention it, I have seen them before but I didn't remember.

I tried using the flip from object, but for some reason I lose all the volume. I tried decreasing the particle radius scale and change the particle separation. My teacher said I shouldn't have to use the boundary layer. I checked my collision too, and it looks like that is working. Any ideas why this is happening?

 

[StepbyStepVFX]

It’s hard to tell just from the video... Do you know if the fluid is leaking ? It looks like fluid is going below ground... If so then this can be a problem with the collider object (the river bed). How does vary the particle count ? Is is constant or decreasing (in this case, means that particles are killed) ?

For having first class colliders, please look at this video from Jeff Wagner (around 1h-1h30, he talks about Flip sim). I don’t know what you used as collider, but maybe you could try to convert / project your river bed into a Heightfield, seems to be working pretty well. And by the way, in the video, he explains all the attibutes you need on your points (vel), and also what piece of data (points, poly, VDB) are used by the solver to have proper collision. Really interesting video :

And from the previous video, I think I remember there was an option to keep the particle count constant on the emitter, particularly interesting when doing simulation on a domain where fluid both enter and exit. Maybe you should look at that.

Last question : what are the colors of your fluid sim ?

 

Edited April 10, 2018 by StepbyStepVFX

[Sophie]

Thanks again for the video, will definitely watch it! 

About the leaking: I made the fluid container a bit bigger at the bottom and turned of the geometry to see if the particles falls through the ground, but that doesn't happen. I noticed the particle count is decreasing. 

I am not entirely sure what you mean when you say keep the particle count constant in the emitter. Do you mean when using this flip from object? And do you mean the boundary layer then? Because I tried that as well, but that didn't really look nice and the teacher said I shouldn't need it. 

For the color... I cached the fluid in my scene, opened it with another scene, load the file and give it a color with a ramp from attribute. Then I have the values 0  0  0.9 for the first point in the ramp 0  0.9  0.9 for the second point in the ramp and 1 1 1 for the last one. I did this to try and get the blue gradient that you get with looking within the dopnet, but that didn't turn out too well. 

Anyway thanks again for the video! Gonna see if this will help me

[StepbyStepVFX]

Yep, I was thinking about using the boundary layer, with proper padding. If you extend a bit your setup, and  limit the padding (whre particles are reseeded to keep the particles count) outside the renderable zone, I don’t understand why the result wouldn’t be nice compared to a sim withtout the boudnary layer (and the proper VDB specified in the Surface Volume).

And concerning the particle count decreasing : they must be killed, probably when escaping the boundary. Have you enables killed putside volume limit ? If you don’t use boundary layer, maybe you should enable the collision with the boundaries, but then it won’t really a river anymore...

i will look at your file

 

[Sophie]

I added a file that I sent to someone else earlier today, of how my setup was this afternoon. This is the V8 file. The other one, is how they adviced me to try it with both a flip from object and a separate emitter at the top. Ii haven't reached any nice results yet. They also said to try and add more trees at the bottom, so the fluid can't escape as quick. Or otherwise just a box that blocks of part of the end, so it can't flow out as quick as well. 

Rapids_BigV8.hipnc

Rapids_BigV9.hipnc

[StepbyStepVFX]

OK, after having looked at your file, everything is setup correctly, I would just say that your source is way too small given the huge surface the fluid must cover (use the measure SOP to see the surface, times a tiny thickness of fluid, and it will be way bigger than the boolean volume you have used (although it is emitting continuously). So if you want to avoid the solution of the boundary layer and a surface volumes that match the river bed, try to increase the size of your emitter in height. It will make a huge waterfall, but if this is outside the render camera field of view, should be ok :-)

Hope that's help !

By the way, in which school are you studying ? Supinfocom, Gobelins, ArtFX, other ? Just curious ! :-)

[StepbyStepVFX]

38 minutes ago, Sophie said:

I added a file that I sent to someone else earlier today, of how my setup was this afternoon. This is the V8 file. The other one, is how they adviced me to try it with both a flip from object and a separate emitter at the top. Ii haven't reached any nice results yet. They also said to try and add more trees at the bottom, so the fluid can't escape as quick. Or otherwise just a box that blocks of part of the end, so it can't flow out as quick as well. 

Rapids_BigV8.hipnc

Rapids_BigV9.hipnc

Sorry, I didn't saw the last files before sending my previous message : I just opened the v9, but your emitter is still very small. The slope is steep, so the flowrate is high, so your particles are quickly spread out in thin layer (that could accumulate next to the trees, indeed, but that's not sufficient), and then reaching the boundary of your domain.

So I would definitely grow your emitter : either in height if you want to render the flood entering the river, or maybe as said before along the river bed. Or maybe, rotate your river bed so that the slope is less steep, and the fluid has time to build up thickness before reaching the out boundary.

[Sophie]

Okay, thanks! Yeah I thought it was mostly in my riverbed, because if I tried with more "simplified examples, it still decreases, but it looks better. For now I was surfacing another result that I had last weekend, so at least I have something that looks decent, because I also still want to look into white water :/...

I will definitely keep trying with the fluid from object though! I'll try maybe rotating my river bed indeed, or making an obstacle that is not visible in the render (or both). 

 

Edited December 15, 2018 by Sophie
not valid

[StepbyStepVFX]

23 minutes ago, Sophie said:

Okay, thanks! Yeah I thought it was mostly in my riverbed, because if I tried with more "simplified examples, it still decreases, but it looks better. For now I was surfacing another result that I had last weekend, so at least I have something that looks decent, because I also still want to look into white water :/...

I will definitely keep trying with the fluid from object though! I'll try maybe rotating my river bed indeed, or making an obstacle that is not visible in the render (or both). 

I am at NHTV in Breda, the Netherlands, not sure if you've heard of it. I believe our school is also hosting Everything Procedural in Breda this year, end of April.

 

Cool, would have loved to spend few years learning in such a school ! Post your results in Work in progress or Finished work (once finished), i am curious to see what you have achieved !

good luck ! :-)