Jump to content

photorealistic water


Recommended Posts

So, I'm working on a scene, that a car hits a fire hydrant and I need to make a photo-realistic high pressure water, either in Houdini or Maya, my first question is which one is better? (I personally prefer Houdini but the quality of the final render is the most important thing). and IF the answer is hounidi, what's the best way to achieve it? I've attached an image that shows exactly what I want.Open%20Fire%20Hydrant.jpg

Link to comment
Share on other sites

You know what they say, 90% of the work takes 10% of the time and 10% of the work takes the remaining 90% of the time. Well, this will get you 90% of the way. Take the particles from the fluid simulation and render them as white spheres. Use depth map shadows, motion blur and render an occlusion pass with them. If it looks too "sphere-ish" then you need more particles in the simulation.

There are other more elegant and realistic ways to get the look of rapid water but if the shot doesn't fill the whole screen and the lighting is straightforward then the above method should work. Beyond that maybe someone else can share their experience.

Edited by lukeiamyourfather
Link to comment
Share on other sites

If it were me I'd skip the whole fluid sim part of the equation and just use particles. With such a chaotic motion fluid sims won't do much for you other than steal your time. And fluid sims can steal a lot of time.

I'd also render the particles as points (render tab on the object->geometry->render as points (mantra)). Just be sure to use an isotropic lighting model in your shader so that the shadows evaluate properly.

M

  • Like 1
Link to comment
Share on other sites

I think Houdini is the right choice, cause it can handle multi-million particle rendering jobs with large data sets. Which is what your going to need for this thing to look realistic. The fluid sim can be used for water collecting on the ground.

Take a look at Peter's tutorial on rendering millions of particles with Houdini.

http://vimeo.com/12092711

Edited by hopbin9
Link to comment
Share on other sites

Hey

Many times when doing fluid effects, people chase after large numbers, when in many cases its just variation's of frequencies that are important.

the only thing I'd avoid is trying to get "too much" of the same thing into your sim, we found that in most water shots, spaces are are just as important as the sim itself, so looking at the you posted, you can see the stratification in sim, which comes from the emission criteria.

One thin I think doing when doing something like this is having the same source of emission(with a little variation) and have different seeds, so you have larger shapes that have small variations.

ie starting with some base sim with low frequency, I dont mean frequency in noise as a force, but rather rate of emission, having "spaces or gaps" in your emission is absolutely crucial. creating a pulsing movement, ie pairing your emission velocity with emission rate.

Id also go with the seeding method, that way you can multiply the numbers of particles by just seeding variations of your sim, this aids with keeping the general structure of your water, but when you want to get the numbers up, you can just wedge/seed your sim and come back after a cup of coffee.

it might be worth your while to focus on adding mist either as fine particles or as smoke, rendered fair faintly, with less gravity, ie advecting you particles though a very course fluid sim and advecting more particles through that to get some cool whirly mist, or just using curl noise.

I agree with Marc that SPH forces arent critical,

a lot of clumping can be created via clever emission criteria - especially if you have a lot of motion blur.

If you can introduce SPH as post force, ie dial in your sim and get it working, then add sph right at the end to form clumping and stingy particles, thats great, by using dops, but be aware that you gonna be loosing a fair amount of time, and I wouldn't run it on all your particles, just for a portion of them, giving you the suggestion surface tension is normally all it takes.

when doing this shot, you can clearly read shapes within the structure.

http://www.fxguide.com/wp-content/uploads/2011/01/weta_armada4.jpg

the key essence is that shape supersedes amount, ie, if I doubled the amount of particles, it would loose structure and not aid the shot.

Now given that this isnt a fire hydrant and our scales are slightly different, however this philosophy seems to be applicable in many places.

in the above picture, i had 30% of my particles going through sph the rest were just normal particles with some cool drag models and mist.

In terms of shaders etc, you'd be surprised at how simple splash shader's can get to get believable results, the problem is normally the fact that you have translucent particles, and establishing normals, so that you get some nice kicks of light coming off the highlights but that dont become too "noisy".

deepshadows seem to be crucial in supporting the cauliflower quality of water, but my trick here is to light with a fair bit of soft back rim lighting, with every light using deep shadow maps, to basically shape the detail.

I totally agree that Ambient occlusion is needed, its great for revealing creating shape's, but its good for contact with other objects, but we play it down a lot because theres a lot of light scattering inside splash and foam, and water in that state is not very diffusive.

Hope this helps

Link to comment
Share on other sites

At the scale of a fire hydrant shooting water out at high pressure though, the chances are that the water is going to basically be a mist which I think is fairly easy to get via noise and adding noise to the initial velocity of your particles. So you visually get clumping without actually having chains of particles traveling together. Even if it's a large scale hydrant like in the video below, the water is moving so fast and with so much motion blur that you tend to not notice individual chains and could probably just cheat it all with noise.

Of course, as with anything though it depends on what it's for and who's going to be looking at it :).

M

oh, P.S. To answer the initial posters question: Use Houdini, it's way more fun.

Link to comment
Share on other sites

Brian makes a good point. A lot of seemingly random stuff is in fact highly structured with details at various frequencies (rocks! they are not random!).

With a noise vopsop on an area attribute you get both temporal and spatial patterns. I played around with this idea and attached a file with noisy spraying water. Perhaps it gives you a starting point or triggers some other ideas.

I also think that you can probably get away with pure particles. Although it is more difficult to get the stringy kind of shapes, you gain a lot of speed. On the other hand, you may be able to do a low-res version of it with flip or sph and then attribute-transfer this on your popnet and thus include the desired watery look.

I think what you'd have to do to make this more realistic is to add some sort of (nonlinear) forces that depend on the fluid "density". That's what I tried to do with the numproxy attribute. The idea being that more particle density means more inertia and a different response to forces.

spray.hip

post-4013-12967160925_thumb.jpg

Edited by Macha
Link to comment
Share on other sites

On second thought, I wonder if it wouldn't be better to estimate the density by using the reciprocal of nearestdist attribute like so: k/$NEARESTDIST. It seems like a neat, cheap way that avoids point clouds and/or numproxy with unknown number of particles because we can always choose the useful 0-1 range and get non-linearity automatically.

Edit: k/(c+$NEARESTDIST^2) seems to work well. It gives a smooth, controllable shape.

Edited by Macha
Link to comment
Share on other sites

  • 3 weeks later...

At the scale of a fire hydrant shooting water out at high pressure though, the chances are that the water is going to basically be a mist which I think is fairly easy to get via noise and adding noise to the initial velocity of your particles. So you visually get clumping without actually having chains of particles traveling together. Even if it's a large scale hydrant like in the video below, the water is moving so fast and with so much motion blur that you tend to not notice individual chains and could probably just cheat it all with noise.

Of course, as with anything though it depends on what it's for and who's going to be looking at it :).

M

oh, P.S. To answer the initial posters question: Use Houdini, it's way more fun.

Hey that's a great comp there man! Looks more real than what I see Weta can do ;)

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...