Realflow / EIAS advice requested

[cjberg]

Correct... that be a daunting task

[arketype]

I have done RF projects at the scale of a glass of water, and it still takes quite a while.
RF is very slow to calculate. I am sure that when RF is used in the movies for ocean effects they have a render farm working for months to generate the fluid solution. This is not cheap.


Have you looked at "Pacifica"?
Pacifica (http://www.paralumino.com/x_cart/product.php?productid=16143&cat=0&page=1)
It shows ocean and wave effects done entirely in EIAS. (with a little AE thrown in).

You could also look into Northern Lights Psunami.
http://www.northernlights3d.com/Products/products.html


Dave

[Veehoy]

You could have a look at Blender fluids. http://www.youtube.com/watch?v=-AiLyQWXjIg

You should be able to import the geometry into EIAS.....

[WHD]

Hi All,

Thank you for the input. My sense it that this would best be done using Psunami as a base layer and then augmenting it with a series of particle fx renders assembled in the composite stage.

I've seen Realflow used in semi confined spaces and my understanding is that generally these sims are run in patches for apparently unconfined areas. That way the splash and interaction of a particular sub event can be accurately simulated.

I've read up on effects like those in "Day After Tomorrow" and others with similar water effects and while simulated it is clear that there is a lot of compositing involved to complete the effect.

Veehoy, I was interested in Blenders fluids about a month ago but found the interface so confusing and unfriendly that I gave up when some real work came in and never got back to it. Probably worth looking into again.

I'm now thinking that I should try to steer the client away from Realflow as a primary solution specially since there is really no close up shots of the interaction. Cost and time is also a consideration here.

Thanks again for the input and I'll check back in the event there are more suggestions or solutions.

Bill

[splitpoint]

Simulation for something like this is a world of pain even on a 64 bit system. Faking it is the way to go, perhaps these will give you some ideas:

http://surfsup.awn.com/?type=article&artID=3

http://www.imageworks.com/ipax/docs/Siggraph2007SurfsUpCourseNotes.pdf

[WHD]

Hi Splitpoint,

Funny you bring that up. I've been reviewing "Surfs Up" since last night. Got the June 2007 CGW right here for reference. Rob Bredow and his team did a great job on that film. Guess it goes to show that the hand is sometimes mightier than the sim.

Thanks for your input.

Bill

[splitpoint]

The hand is absolutely mightier than the sim. The Guardian was evidence of that. Some sim in that one but much was done by hand. Might be a good one to look up as well, I believe they used Lightwave and camera maps for that if memory serves...

[splitpoint]

Not the article that I was looking for but this one talks about some of the cheats that they used in The Guardian:

http://www.cgw.com/Publications/CGW/2006/Volume-29-Issue-11-Nov-2006-/High-Seas.aspx

[Ideas]

Hi All,

I'm in discussions with a client about a potential project. The client among other things wants to simulate a Psunami hitting a village and has asked about using Realflow to do this. I have tested it with a glass of water and that worked but the view for this project is wide so that about 500 feet of coastline and 500 feet inland are visible. My guess is that this is unlikely to work as a realflow simulation due to scale as just a starting point.

Anyone with more experience using Realflow care to comment?

Bill


Hi Bill, I did a job a while ago using Realflow to simulate snow. This was for a stills project.
It took ages just to get the particle effects right - mainly because the render times were so high. I was intrigued because it was the first time I had used it (so, learning curve involved as well), but I must have used up the best part of a week on just that part of the project.
Very slow. Although I would guess that like anything, in the hands of a real pro it would be a lot faster.
Michael.

[WHD]

Hi All,

Well as a bit of a follow up to my original post I figured that I'd see if I can get a bit of feedback on a test that I ran.

This test does not use Realflow or any water simulation. All keyframed and optimized for maximum render speed. It is a multi pass simulation and since it is just a rough test I was a bit sloppy with some of the stuff like the beach projection map but the idea was to see if I could get something reasonable that renders reasonably fast and worry about finessing more later if the job ever actually happens. That does not look too promising at the moment.

This is on my limited ftp so it will probably get pulled when I need the ftp space.

http://digitaldempsey.com/ftp1/Making%20Waves_071_TEST_005.mov

Any feedback on this test will be greatly appreciated since I've been the only one staring at it.

Bill Dempsey

PS here are some notes on the methodology.

Quick rundown of what I did. I approached it as a simple rig that would allow me to create a small section of a wave that could be assembled to build a wave of any size and length needed. This sample has 3 wave segments comped end to end.

The base is a Psunami surface with calm default water.

Layer 2 is a wave front using a long psunami strip that has been deformed with a bezier 2 deformer to arch upward. This is attached to the moving rig.

Next is the water wheel that is a cylinder that has been sculpted in Silo to look watery about 250,000 polys. It is rendered with RT transparency set to refract the sky and has a material I created which is transparent with opacity/color with thinkness. It rolls along like a wheel. I used an alpha cutter on the base and back so only the top 1/4 segment facing the camera is rendered.


Next component is the wave crash that is also based on a cylinder form. I instanced thousands of droplets onto this form. Each droplet was a sphere with about 25 polys. There are about 15 of these used in this pass each . Total polys for each pass is about 3 million. This is attached to the same spinning rig as the water wheel.I also used an alpha cutter on the base and back so only the top 1/4 segment facing the camera is rendered.

Next pass was originally a series of Power Particles splash effects but when rendered I had trouble keeping it stable in RR so I exported these as cycles also attached to the rig at the base of where the wave hits to create the explosion splash in the front.

Next I did a Power Particles pass that was just points and this runs along the front edge and floats upward with less gravity and a fan applied. It has about 75,000 points in it. and is comped under the water wheel but over the Psunami effects and blurred so that it looks like mist rising from the wave


Finally I did another Power Particles pass that was just points and this dresses the front edge. It has more gravity and a ground plane to keep it low .It was also set up to start darker get light and then fade. It has about 100,000 points in it.

All these passes were rendered 3 times and added to a beach and sky pass that were assembled in AE with hand key framed roto-masks to control what is seen and when.

None of these 7 passes per wave segment in SD takes longer than 2 minutes per frame to render.

A GI version single frame test of some of the splash elements looked really nice and had a lot of depth to the water effects but took about 3 - 4 minutes to render so that is an option depending on what the final shot lengths will be and how much time I have if any of this happens.

[estudiodesign]

Boah...that looks good...great work !

Only thing I don`t like too much is that "pattering" in the end
when the wave breaks and comes near.
It seems a little bit too grainy or noisy to look real, kind of irritating for my eyes.
May be that comes from your power particle path you described ?
I think they should look softer, more like being a homogen mass of foam.
(may be the different particles should be set up with less colour range ?)

But again... very awesome !

best
Tom

[WHD]

Hey Tom

Thanks for the feedback. I had done a bit of softening to that front section and was worried about that noisy feeling you describe. After staring at it as long as I did it was difficult to judge if I fixed it or not. Clearly it still needs some more work.

Thanks

Bill

[WHD]

Hey All,

Based upon Tom's feedback and a few other folks I've emailed, here is a revised version of the wave.

http://digitaldempsey.com/ftp1/Making%20Waves_071_TEST_007.mov

This now has 2 layers of noise overlayed on top of the wave explosion element in the precomp .Additionally it has also been softened a bit to help break up the uniformly scaled noise patterns that existed in that element.

See what you think.

Bill Dempsey

[estudiodesign]

Hej Bill,

I don`t know why, for me that front part still looks as if a heavy cloudburst
patters directly into the front of the wave...can`t explain it better.
But I don`t know if it is so important and there`s no need to criticize that too much.
I am not sure if I could come half as close to what you did here.
And I am still inspired of your genius idea concerning that rotating cylinder
representing the water wheel. Really cool ;-)

Best
Tom

[WHD]

Hey Tom,

I really appreciate the time you have taken to look at this and think about it. A fresh set of eyes on it has really helped me to focus on what is most important. While not 100% there, it is substantially better than when I posted the first sample thanks to your participation .

Bill

[rtrowbridge]

I agree with the others. The wave itself looks great. But the front edge just isn't quite right. I think a finer mist might help push it over the top. You're almost there.

And for someone only seeing it once, they probably wouldn't notice anything. It's only when you get the chance to watch it over and over that things start jumping out that need to be worked on.

Ross

[ediris]

Just trying to help, have you tried with mr blobby?Or morphing the rolling cylinder to a flat plane?
Edgard

[WHD]

Hi Ross and Edgard,

Thanks for your input.

I agree that the contact point is a bit unrefined. A bit too particle-y. A cloudy mist effect along that edge sweeping up might help to integrate it better with the ground more. Maybe a KMyst or FTurb shader on a form sliding just at the front edge of the wave.

Mr Blobby is something that I have used in the past but I find it a bit limiting since it is slow and tends to balloon up the geometry too fast. It also tends to make everything into a sphere. The intended scale of some of the story boarded shots also puts it in the same problem area that Real flow was in. At least that was my analysis. I had considered it along with particle effects for splashes but once again the scale made it seem a bit impractical. If however someone wants to test 5 to 20 foot tall splash effects I'd be interested to see the results. Hint Hint. :-)

Since this test is supposed to be of a Tsunami wave, A small one but a Tsunami none the less ,its force should not be diminished significantly by the beach since it would continue for a great distance inland. If it were a regular wave hitting the beach some form of scaling would have to occur as you mention as it dissipates. I think most crashing waves just become a lot of froth at a certain stage.

While my originally intended scale was to be about 10 feet tall, my wife had remarked early on that it only looks about 5 or 6 feet tall and that would be about right given the scale of some of the droplets. Once I determine that was the scale I just proceeded at that scale since the components were already there. I figure the larger the wave is the more smokey or foggy the spray would appear. A ten foot wave would likely have a lot more atomized water and a 20 foot wave even more so.

I had no trouble finding Tsunami samples on YouTube for reference but a large Tsunami wave heading right at you and breaking right in front of you is noticeably absent for obvious reasons. I noticed that once it clears the beach it just looks like a boiling white water river except miles wide.

Unfortunately at this moment the potential job that triggered this test looks unlikely to happen but I feel that I'm in a much better position to deal with it in the event it does occur thanks to every ones feedback .


Bill

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.