A man overboard! Un homme à lamer!

[jstrob]

I sent it to your email. thanks Ivaylo.

[jstrob]

Here is how my second version looks like now. The render was not finished but it gives a good idea. The scale looks like the diver is tiny. The drops looks a bit big.

better quality here:
http://www.phylacterecola.com/strob/...2_outPetit.mp4

[Ivaylo Katev]

still thinking the splash shader needs bigger size multiplier
hadn't time today to see your scene, but hopefully tomorrow will have. will try to adjust the splash rendering too

[jstrob]

I am planning to do a separate pass to add some splash. I didn't want it to hide completely the beautiful water mesher so I ended up diming it to just what it takes to make the mesher droplets more white and I will do a separate pass that I will be able to adjust in comp.

Also I had to use point mode cause the other modes were too long to render. and I think the size multiplier doesn't have any effect with point mode.

[Ivaylo Katev]

the size multiplier affects the point mode, the assumption is that the particles are smaller than a pixel, but this does not mean that their size is not important. the bigger is the size the denser is the result. if you see single pixel, the bigger size produces bigger alpha. actually the exact relation is simple -the alpha is calculated as the visible surface of the particle measured in square pixels. if the surface of the particle exceeds 1, the alpha is saturated to 1.

[jstrob]

oh boy good to know! I will take that info into account for rendering my foam pass.

[Ivaylo Katev]

well, i spent some time to try to optimize the scene. first i disabled all the additional slowing options like mb, adaptive sampler, image filtering. disabled the rendering of the foam and splash, and the fat man too.
before to continue i have to mention something important - the foam and the geometry render times are not additive! you can have situation where the geometry is rendered in 1 minute, the foam in 2 minutes, but the both are rendering for ten minutes. so, each component must be optimized separately, don't be tempted to think that it's render time is negligible.
my first target was the water.
here is the rendering of the water with the original settings

you see the render times - 4 min
this is too much, but not the big resolution is the reason. in the most cases the reason is the material. so, i saw that you have used glossy in both the reflection and refraction parameters.
it's important to know how vray calculates the glossy parameter. if this parameter is below 1, it starts multiple rays with similar directions, the count of the rays is determined by the subdivision parameter. it's VERY performance consuming compared to pure reflection/refraction.
here is the image with glossy=1

as you see the render time is about 30 times lower! however, the result is not the same, the sun reflections are gone. fortunately we can have them without using reflection glossy parameter, but using the highlight glosyness, that is pretty fast.


well, this was about the liquid, now i'm investigating the foam, this means i'm rendering only the foam, with no water.
without shadows it renders very fast 13 sec versus 22 min. still not sure is this normal or not. there are some very slow setups, when the foam is not dense. this sounds weird, but the dense particles are rendered faster, because the light cache works well and because the shadow rays are not crossing the entire volume of the foam. i will continue with the foam tomorrow, if the point shader can't be accelerated, you can try to render the foam as bubbles, this is actually more realistic, just needs bigger image sampler subdivisions, but you have them because the motion blur.

[jstrob]

So that's why I think it's important to be able to render them separately (water foam and splash). I think it will be way faster.

I am trying different size multiplier on my points and the alpha is the same no matter I use 0.1 or 1000 as size multiplier...

I also notice my water was occluding too much of my points inside the water. Since I use shadow, they seem to disapear. And now I am trying to use the water surface as a matte to render only the foam and too much foam just disapeared because it is under the surface. So I creaed a copy of my PHXsimulator to use a copy as matte and the origin;a as liquid in the foam shader. That way I can use a higher isosurface level for my matte and get the point out of the water to se them more.

[jstrob]

Here is 3 render for frame 62. I use 3 different iso surface level (see stamp). The original liquid was left at isl 0.1 like originally. I will render the whole sequence with 0.8. (Don't pay attention to render time cause I used DR with my render farm and sometime my computers were not all on.

[Ivaylo Katev]

yes, you have to disable the water shadows, otherwise the foam is hidden

[jstrob]

Though it can be useful to hide a bit of what is under the water level. In this scene I have too many points appearing under the water. And now my foam pass is fast to render. Maybe 30 min a frame instead of 10 hour/frame. I notice what was the longest to render was the bunch of points inside the water at the fat diver's point of entry where the mesher created a splash. I think there are too many points in this area. But once I hide it by using a copy of the mesher as a matte, it becomes very faster to render.

[TSR_trix]

A good conversation about rendering pipeline with water!
In some point might be worth to document it and include to Help?

[waaazoo]

A good conversation about rendering pipeline with water!
In some point might be worth to document it and include to Help?

absolutely! +1

[jstrob]

I am now trying to resimulate cause for some reason there was a bug in my particles. I think I stopped the sim and restarted it later, but half the sim is different...

And now I am trying to get less foam under the water with my sim. So I try with the depth and min age setting. But as soon as the min age is other than zero, I get no foam at all. how does it work? Is it in seconds or frames? and why I don't get any foam even with the min age set to 100?

[jstrob]

Ok I understand depth and min life a little more now. If I want to get less foam, I must put depth to zero. Increasing depth create a lot more foam. And if I want to create less foam I have to put min life to something like 0.1. Higher values soon remove all foam.

What I don't understand now is "liquid like". If I increase it to 1000 I get very slow resimulation time, but always the same result... 0, 100, 1000, all give the same result but just increase sim time...