OK I understand how hard it must be to develop a fluid simulator (or probably don’t), but I am just sick of watching Realflow videos (http://www.youtube.com/watch?v=8Pmm9UKqc5I) from 2006 that completely blow the blender stuff out of the water, so to speak.
Granted this might be my fault, but why is it so F%$&@^ hard to make a realistic giant tsunami-like wave that actually crests and breaks?
I encounter two problems - the fluid doesn’t allow forces to act over distance realistically, i.e. I have a “wave generator” at the back of a puddle of standing fluid. if the generator (a giant paddle) is moving too fast, fluid simply breaks off near the paddle and goes flying diagonally, while the rest of the puddle is undisturbed. If it moves slowly, I get something of a cresting animation (I have a pretty good sloping “beach” underneath) but the wave doesn’t crest, it simply moves like oily goop and sometimes breaks into strands that separate unrealistically, or it forms a “wall” with the bottom part moving faster than the top, thus no crest. There are no eddies or anything else insinuating forces across any kind of distance, it is all localized, and the fluid volume also seems inconsistent.
Now, is there something I’m doing wrong? I made the fluid domain the max size (10 metres! is that right? why can’t it go up to 100?) and the fluid resolution is something like 250. I have a fast machine but it never looks right. I am tearing my hair out. I am considering extreme acts of violence, or worse yet, (gasp) piracy. I think I am done with the fluid sim and I will stick to making other cool stuff. Any thoughts from people on this? I had such high hopes with the “improved” fluid of 2.43
or are you, yourself, in your spare time, for free, on your home pc, trying to achieve the effects seen in your linked video, that a team,working for a hollywood company, full-time, using $7000 per seat software, on high-end workstations with 1000-PC render farms, achieved?
yes the inspiration was definitely from Mpan3, but the desired effect was more of a giant, single unbroken wave with a clearly defined crest and tube. I mean obviously I wasn’t saying I wanted it to be on par with that shark jumping out at the helicopter bit, but I just think it could be a lot better. I know that video was ridiculous, and they had a render farm to do it, but I have so much faith in blender this was the first time I’ve really been let down.
Yes, I think you can do something similar to that with Blender’s simulator. Buy yourself a super quantum computer from five years from now and you can do it easily. But you first need the time machine.
Seriously, though, Blender needs to support distributed simulating.
realflow’s sim is particle based(and then ‘metaballed’, so to speak), while blender uses a free surface lattice something method. i am sure both are different in many ways.
Plus, i am sure to get a good cinematic fluid flow in realflow, like say the wall-of-water in poseidon, the artists has to setup a ‘guide’ for the particles to follow, you can’t just simulate a 2million square miles ocean and then blow wind on it hoping the simulator would come out with a massive wave.
I believe I am reaching the limits of the Blender fluid sim as well. This is a vortex I did that took 3 hours to bake at a resolution of 70 and world size of 1.5 meters. .avi 2 MB Xvid. Impressive, yes? Not so much when you bump the resolution or world size much higher. My computer just stays on frame -1 for close to 8 hours, so I sought out some help to see if it was my machine. I sent mpan3 a PM, and he’s having trouble getting it to work as well.
something funny is going on here, I first did an extremely quick test-bake at res=50, and it came out great, I can see a vortex forming due to the blade (top figure). Althought the quality is horrible as expected. But when I increase the resolution to 250, it took 8 hours to bake the first 80 frames but I get something completely different: the water has turned into a hollow bow beginning at frame 2! (bottom figure)
I have no idea why this is happening, though I am sure i havn’t changed any settings. I have a feeling you are hitting the limit of the fluid engine and soemhow it has gone all unstable. Maybe this is why your computer is having trouble it in the first place.
I’ll do more test and let you know asap.
(Hope you don’t mind me posting this, mpan3…)
LotRJ, lol not at all, but as you may have noticed I deleted the files already, sorry about that.
While you are here, let me tell you that I’ve done another project that involves blender fluid(can’t disclose more detail than that, sorry.) and i ran into the same problem: part of the fluid disappears at high resolution, usually the top. So it’s not just your file. Somehow blender fluid just isn’t stable anymore at high res.
I’m curious to see what Colin Lister will have in the new Blender book. He has produced some great results with textures.
I’d love to learn more about when and where to use “real fluid” simulation (physical properties etc …) vs “fake” (particles / textures …etc). I realize it’s not a simple subject by any means, and it’s why people specialize in the field.
mike, to my knowledge, no media companies use pure ‘real fluid’ for their work, simply because they don’t offer enough control for the artists.
That’s probably why the blender fluid sim is so resource-intensive and hard to use, because it is more ‘real’ than any guided particle approach.
They used lattice deformations to get the basic shape of the huge wave in the process of Intermedia logo creation. I think that it could be used also in blender to get the basics done… after that it would require some shader work for water texture and maybe some particles to get the spays of water.
I’ve decided what we really have in blender is a great “oil” simulator. No seriously. I think a massive part of the fluid sim problem is the lack of account for air bubbles, which really influence the look and behavior of water. MPAN - I definitely noticed the 250+ res problem - in trying to make my tsunami simulation when I cranked it above 250 or set the fluid to render as smooth shaded, there were serious render problems where the top texfaces were cut off! not sure if this was happening to you, but on the whole 2.43 actually seems LESS stable than the fluid in 2.42
and when I said “Realflow” I actually meant “Scanline flowline.” Realflow is included with maya, whereas scanline flowline costs AS MUCH as maya. loooks like I won’t be playing with that any time soon