A couple of Workflow Questions

[ben_hamburg]

I actually just switched out your ship for our dummy ship and recalcluated with pretty much the same settings and then rendered with your Light setup, I might have changed the Image Sampler settings, that gave me about 13 Minute per Frame and lloked good.
Then I recreated the Sim setup in our final shot, I might have changed a couple of settings (because your scene was 1 Unit=20 cm and ours is in 1 Unit=1m i mulipltied a couple of values), which might have given me more foam/splashes.
Without foam and splashes the scne still renders pretty fast, even with Domelight+HDRI.
I´ll try again with your light setup and render settings, so I´ll know if it´s a problem with light/render settings or a problem with the sim.
We´re actually not using light cache throughout the whole project, i figured it´s faster to skip and still looks decent with only IRmap.

But can you please till answer my question on how to properly render the foam/splash pass, because I´ll probably have to use different Light setup for it and I couldn´t figure it out yet.

[ben_hamburg]

Ok, did a quick test render, it´s definitely our light setup...disabled the domelight and GI and now it renders really fast...so if you could help me out on how to setup the scene so I can render the scene with our HDRI light setup and then extra pass with only foam/splashes and optimized light setup for only foam and splashes...

[Ivaylo Katev]

i'm not a big compositing specialist, but would suggest the following:
i think you can render the entire scene with simple lighting, placing the hdri as environment map. the water actually does not need lighting, it uses reflection and refraction rays. the ship will be visible in the reflections, but it is the only part of the scene that needs good lighting. so, render it alone, with GI and dome light, and export the z pass. render the water only exporting the z pass. mix the ship and the rest of the scene using the z pass of the ocean and the ship.

[ben_hamburg]

I think I already figured it out. I can simply render the scene like you said, then render foam/splashes with simple Light and just use vray properties to make everything else matte objects. I was just confused, because the Vray atmospheric element renders completely black, I thought it was supposed to give me a pass of only foam/splashes.
And I´m not a compositing specialist either...

On a sidenote: Are you considering selling "Simulation Node licenses"? For this project it would actually be great to have two ore more extra nodes just for simulation, so I could work on one sim, send it to simulate and then start working on the next sim...
Even if Phoenix is way cheaper (and as far as I can say just as good) than its competitors, buying several complete licenses for just one project kind of seems overkill...

[Ivaylo Katev]

really the atmosphere element must give you the foam/splash.
but if you will render them separately, you can use directly the output. however, if you render the foam without the water, you will lose the foam/water coloring

[ben_hamburg]

Still couldn´t get the foam/splashes in the render element, renders only black, no idea what I´m doing wrong. Am I supposed to just add a vray atmospheric render element and be done with it? Can´t find a way to pick the phoenix atmospheric effect anywhere...
But due to the limited time we have for rendering, I think wel´ll have to go the route with rendering foam/splashes separately and do the coloring in post.
That way we can also lower sampling settings for foam/splashes pass and get even faster renderings.
I noticed that the frame I was trying to render was already at 9M foam particles, so that might have been a bit overkill...
Too bad I can´t have more time for R&D to get a feeling for the different settings and how they affect the particles, i do realize that there is probably no proper way to predict the particle count in advance...phoeni would have to be able to predict it´s own simulation future...

But maybe you could shed some light on how to get longer foam trails...?
Do I have to up the Foam Half life value (it´s at 0.5 in the demo ship scene)?
Do I have to increase the "outside of grid" life of foam or splashes?
Or do I have to increase birth rate or threshold?
Or do I just have to let the simulation run longer (right now I have a "preroll" of 150 frames before the actual shot begins) do see longer trails?
Like you said: I could get away with fewer particles, if I just shade them as bigger sizes for shots that aren´t too close up...i just need the particles to stay alive longer...

[Ivaylo Katev]

i found it, there is an option "filtering" in the render element, it must be enabled, i will investigate why it renders black with it, but for now this is your solution

about the foam - if you render it in separate pass with no water, 9M is pretty good, you can render even 50M for time less than 5M in water. beware of the foam size - it affects the final look very strong. smaller foam size leads to thin string like patterns, the bigger size leads to foam that looks like floating islands. see the image, it shows the difference. for a good simulation the foam particles must be in contact to make possible the b2b process. if they are not in contact, the foam looks like dust over the water. thats why the foam size does matter. the foam size depends on the particle count, the bigger is the count, the lower must be the size for the same appearance. this means if you need to increase/decrease the count, you have to change the size too.
you can also try to use the liquid-like option of the splash. see the help (you can find it at help/additional help menu) and there are image showing the difference between normal splash and liquid-like splash. keep in mind that this option makes the simulation slower, it is similar to the b2b option of the foam and all the rules are the same. i see some flying foam particles in your simulation, that is disturbing. can't say the reason without animation preview.
in general, the foam/splash control is the most complicated part of the simulation, the liquid is easy, in the most cases you need only to enable the liquid checkbox and already have good looking liquid

[ben_hamburg]

oh great, I´ll uncheck the filtering then and see what we get!

I was really wondering about the drastic render time improvement of foam/splashes, if rendered as a separate pass. I guess its the reflections and refractions interaction between foam/splashes and water, that make the render time explode if rendered together...

The demo scene was a very good start to get a basic setup, that only needs to be tweaked for those kind of sims, since the system units of the file was different than ours, it was hard to figure out which settings need to be multiplied and whish settings needed to be reduced in order to give a similar look.
That also explains the flying particles: They are due to the fact the falling spped was set to 25m (I multiplied the value in the demo scene by 5...).

So once again to recap the foam size:

I do need to take care of the foam size during simulation, because of the pattern they are forming , if they are in contact.
What would be a good range for foam and particles at a Unit scale of 1m?
Last sim ran with 0.1m size and already looked ok, I´m trying with 0.05m now.
The hardes part is still predicting the particle count, I was already at 4m foam particles after 20 frames and I´m not sure which parameters to lower first for lower particle count, because Simmulation times would get really slow after only 100 frames i guess...
Any advice on that?
Unfortunately at the time we can only worry about getting everything done and not about how good it looks in the end...

[Ivaylo Katev]

the splash count is controlled by the birth rate, the foam count is controlled by the foam on hit option of the splash. they affect the birth in linear way, if you need to double the splash, double the birth rate

[ben_hamburg]

Ok, great,now i know how to control the particle count.

One last question for today, not sure if it is because of the inertia: At the border the liquid seems to "sink in" a little bit, maybe it is because of the shockwave created at the beginning from the moving grid and will disappear after some frames (haven´t simulated the scene past 100 frames yet).
Just checking if you know abuot the problem and if I just need to wait more frames for the "border wave" to disappear.

[ben_hamburg]

hm, not looking too good after 200 frames...I can still see a visible seam at the grid border.

You can see it below, together with some settings...is there any way to avoid those borders?

[ben_hamburg]

I guess the fall off option in the rendering tab only applies to the grid mesh, and not the particles, right? otherwise we might be able to blur out the borders without having to resim.
On the bright side....we are getting nice long trails now....

[Ivaylo Katev]

seems like your have the self arising waves, the reliable cure is to increase the spf, but you can test to use ft advection, i noticed that it does not suffer from this disease.

[ben_hamburg]

Ah, sorry, now I remember you already answered this question in another thread...I´ll try both options on the next sim then, depending on the framing of some shots we might get away with the sim right now.

[ben_hamburg]

Still no luck...Increasing the SPF Value still produces that initial sink in. Using forward transfer produced even worse results. I noticed, that the initially specified Z size of the grid doesn´t stay the same, if I increase the resolution, but slightly changes, so I was wondering if the initial fillup level might be wrong?
For example: My grids Z size is 3m, I placed it at -1.5 meters, so I thought an initial fill up of 50% should make the ocean level at exactly 0m in Z direction.

Like I said, I think we can get awy with it, phoenix does at pretty good job at blending the borders, the only issue are splash and foam particles, because they are forming straight patterns at the borders this way.

I attached another image, its taken in the front viewport and you can clearly see the surface bend atthe borders.