Fluidity - NextGen Fluid Dynamics - [ RELEASED ]

I understand that, but emitter's are tied to volumes, so it's easy to calculate radius in world space, whereas simulations are volume independent, so their radius values are in pixel space. I could add an output to the volume editor, showing you the world space size of the emitter etc... But I would like to keep it world space as it's consistent with unity colliders, and other similar components.

As for handles, you should be able to rotate temp emitters to change their direction. I will add an extra transform to represent the simulation direction/position.

Thanks, for reviewing!

Offer's still on the table, if anyone else wants to leave an honest review, I will forward you a token of my appreciation.

 

Most recent changes are almost ready for submission. I am also working on some lighting stuff. GI from fluid, shadows, opacity mapping. Pretty good progress

 

Wow all things on my wish list your doing!

 

I'm not sure how much of it will make it into the next release, but yeah progress is good, I'll post something very soon.

 

The sooner Unity implements this. The sooner we can all have faster fluids

 

Just gave 10 votes!

 

Hey Alex, just coming to the news. Any rough ETA for the next version? I'm asking because the problems I mentioned about a week ago (in my post with the pictures) actually cause a crash within a build. My dirty trick of enabling the volumes in Awake and turning them off in Start works in the editor but leads to a crash in any 64bit build.

So I've had to turn off fluid volumes for now and I do miss them...

 

Hey,

I'm working away at the new features, plus bug fixing. I'm having difficulty reproducing the problem locally however. Could you send me the crash report/log file from the broken build please?

Thanks!

 

Ok, I just sent you the crash report via email. Thanks!

 

I've just had a look at the report, thanks for sending it

The problem is coming from a call to ShaderLab::shaderprops::GetTexEnv. There is some access violation occuring. This is C++ side, and out of my reach unfortunately, maybe you could file a bug report with Unity?

Couple questions though, are you running the latest Unity? and are you running the latest graphics drivers?

I couldn't help but notice in the logs as well that there are some minor problems with your project:

You should probably fix these up as well, so we can get better coverage over what's actually causing the crash.

 

Hey guys, I'm trying to compile a list of features I want to put into Fluidity.

Any suggestions? SERIOUSLY, anything!

 

I did not read the whole thread so i am sorry if that was discused before but it would be nice to emit light ( not some aproximation of point light, but volume light that inject geometry around ) and also shadows ( if smoke volume is big it should block more light for instance ).

 

Yes, you're right, I went ahead and fixed all those warnings. It didn't seem to be related to the problem though. But after spending a couple hours chasing the issue I managed to bypass the crash. Here's what I was doing before: I was disabling the fluid volume at game start and then enabling it when I needed it. Which seemed like a logical thing to do (since you told me that even if the GradientIntensity on the fluid volume is at zero the simulation itself is still running in the background). And that was causing the crash in the build.

So I simply let the fluid volume enabled at all time and that fixed the crash. Hopefully with your new version and independent simulations that shouldn't be a problem anymore.

And since you're asking, here's the workflow I would need for Ghost of a Tale:

In my _Fluidity_Controller_prefab (always present in the scene) I have a bunch of unique simulations defined for very specific purposes (torch fire, magical bonfire, etc...). All the simulations that are not needed would be turned off of course.

What I would like to be able to do is (for very specific purposes) to "clone" one of those simulations, turn it on and assign it to a particular fluid volume. I would then animate the parameters of the simulation to my needs and then, when I don't need it anymore, de-activate the fluid volume and destroy the cloned animation.

Do you think that would be possible?

 

You should still file a bug report, because if it works in the editor but crashes in a build, then it's a problem for Unity, I just don't think I could help you much with it. I can juggle some stuff around on my end, but it'd only be speculative. I would love to see a stripped down version of the project that demonstrates the problem, maybe that would be of more use?

Yeah, I don't see why it shouldn't be possible, what's stopping you from doing this just now? I would recommend using a pool of simulations over inline construction/destruction, for obvious performance reasons.

 

I'd say let's not worry about that for now. The crash is gone so it's good enough for me!

Well, the fact that you can't yet turn ON/OFF simulations independently in the current version...?

 

Ah I see, if there are no volumes using a simulation, then it shouldn't be updating. Can you confirm that?

 

Okay, I just ran a series of tests and here are the results:

1) When all the fluid volumes' gameObjects (and/or their fluid component) are disabled the simulation is always running in the background (as confirmed by the presence of Controller.Update in Unity's Profiler). The only way to really stop the simulation is to uncheck "Run Simulations" in the global settings.

2) If a fluid volume component (or its gameObject) is disabled when the game starts, then it will NOT turn on when the component (or its gameObject) is enabled later on. That's a really big problem of course (and one I suspect is at the source of the crash in the build).


As a side note, would it be possible to please name the fluid controller something like "FluidController"? Or "FluidityController"? It's just that the name "Controller" is very generic and could apply to characters or anything in Unity; it just makes it a little more difficult to identify in the Profiler or even among the scripts currently open in a script editor...

 

It's in a namespace so it's ; Fluidity.Controller, it should be showing up in the profiler as such. Doesn't FluidityFluidityController seem a bit redundant?

Okay great, there is a bug to fix for me there then

 

No I think it's great! I'd even suggest "FluidityFluidityFluidityController" for good measure...

Anyway, as you can see it shows up as just "Controller" in my version of Unity (4.3.1):

 

Okay, point proven, I'll rename the script It's easy enough to do anyway.

 

Ok, great! Thanks Alex. The way you have it for the FluidVolume component is very clear. So FluidController would really make sense (I agree with you "FluidityControler" is not the best choice, especially with the namespace)...

 

Overall I really rather like this product, seems like there's a fair few fixes in the next version so I'll hold off putting a review down until then.

I have a problem at the moment that there seems to be some oddities happening in linear color space, these are quite obvious just running the jet demo from a default import (odd color bloom on engine startup, heat haze being white rather than transparent).

Initially as mentioned by others above I didn't quite understand the relationship between the volume size and the simulation size. It made some sense once I'd fiddled around a bit but isn't isn't particularly intuitive to start with. It might be worth having an option just set a "density" and have the simulation derive it's size from volume dimensions (i.e. density 32 to 1, means a 1 size cube is 32,32,32 but if the volume scale was doubled in Y the simulation would alter to 32,64,32 automatically).

Additionally having some cool preset gradients to select from too would be nice, since the demo ones you've got look great but are quite complex.

 

Mostly what I'm looking forward to is the stuff you have already mentioned, especially light-related stuff. But I reckon once the stuff to make it appear more like liquid is finished, that will spawn some further requests.

 

Hey again! I finally have the opportunity to spend some real time with fluidity, so im going to! New project, very enthusiastic about it, very ambitious, but if motivation stays it will be done, as its the kind of thing i may well get a bit obsessive about. I already know circumstances in which i'll be using fuidity, and ive already modelled and animated the model i initially felt would benefit (i made it ages ago and have been aching for an opportunity to use it), so i can play around and see what happens there. Thinking about it there will be quite few situations, in various contexts.

However, there's something central to the idea, which is basically what i'm finding most exciting, and i fear it may be entirely outside the remit or ability of fluidity as it stands, it could mean a fair bit of work, its very heavy on smoke/fume generation, and the scale is enormous. Your pyroclastic work has a clue

And here's the feature request! A volcano is central to the idea, based currently around the imagery of the volcanic events in Eyjafjallajökull, Iceland in 2010, youtube has a good few videos of them. I'm aware i'll probably have to do this myself really, given the work and that it is possibly out of the scope of this shader), although i do have a very clever pal to occasionally help out, but you asked for features so there you go! The actual needs of it are a little uncompromising however, and watching documentary imagery will explain, actually here's a quick vid:

It's very big, with large plumes of detailed propogating fumes rising with the stack, many individually, and it doesnt really dissipate until it has risen tens of thousands of meters and been scattered by the wind higher in the atmosphere. It animates, obviously, much in the way the pyroclastic shaders ive seen do, although just as gas. The whole plume would be lit be a directional lightsource and internal lighting. It would be animated procedurally over the course of an hour although the actual eruption would have happened some short time before the 'game' starts

I don't think the detail in the fumes could be generated realtime but wouldnt know how to replace it, particles?

And for extra fun, it would be visible from 200 miles away initially, with the middle section of the 'game' involving approaching it rapidly (the world composer thread explains how i figure id do that)

I'm pretty sure all that is way outside the scope of this, but you did ask! And I do need some way to create this, and it be fully 3D, as eventually the fumecloud and its scattering 'head' will be towering above you as you fly past it

Haha, sorry, i will continue to look, somewhat obsessively for solutions myself, but the way the flame rises right at the beginning post gave me the idea to be like 'lets ask him, why not'. I expect you to say 'stop being silly', but may as well try.

 

Purchased!!

It's looks on my copper pod engine

 

yes! I want to use this for volumetric coronas around stars.. can it? this would simulate the effect of CME's (coronal mass ejections)

what about usage as volumetric clouds? hehe thinking outside the box on this one

obviously water, but more importantly, crashing waves/shoreline.. is there a way to add noise akin to an FFT ocean simulation?

 

The crashing waves thing, do you mean vapour as the wave crashes and the foam edge comes out?

Cause that would be sweet, and totally going in my ocean after bigger issues are boomed

 

that, and along coastlines. think of surf.

with fluid dynamics, you can more realistically achieve that than any vertex/FFT solution.. I've just been waiting for the tech to mature enough to be able to do it. even now i'm not sure that you could use fluid dynamics to simulate large areas like oceans, but man that'd be cool.

another thing i'd like to do is use it for weather simulations, or clouds.

the trick is to be able to add some perlin or other noise to give it texture. simply changing the 'viscosity' for the solver doesn't change how it 'feels', if you get my meaning

 

You definitely couldnt use this for oceans I dont think but surf effects sound good, i pimp this a lot and all i did was put the pretty bits into a whole but it shows how you can get your surface foam looking great

http://www.youtube.com/watch?v=vmwdP4OFgtI

Having a bunch of very small simulations going on for bubbling surf.. well, id leave it to you to figure that out then copy it, hah

I was having a bit of a dilemma with clouds cause its hard to do a game set so much outside as my planned project is without decent clouds, i only really enjoy the volumetric effects and Nuaj is very pretty but too expensive for me usually (although i might plump up anyways) - however Fluidity's 5thdemo scene is simply a volumetric object in a room, im assuming shaped by the 3D texture assigned to it. There's turbulence library in the asset store that im strongly compelled to acqure anyways but a friend in the industry said for a cheaper option if you're after clouds just go have a look at iq's famous raymarching demos at shadertoy.com

Either way, in his generosity the Fluidity fellow gives you all the hints you need to get making clouds in one simple demo scene. Just wish I knew how to illuminate and shadow them now

 

nice water waves
needs more spray along the hard shore/cliffs though, and that's where you might be able to attach some fluidity emitters to handle the spray/wave crashes? not sure, I don't have the asset ;(

the fact that it's possible to do clouds with this is excellent
this one asset could solve my 'plasma/volumetric corona' type thing as well as some of my water issues and clouds

 

+1 for clouds, maybe something like this? http://reset-game.net/?p=284

 

Quite pretty blueivy but they don't appear to be fully volumetric clouds, when as stated you're given everything you need to create a volumetric cloud in fluidity except a 3d texture source. As i said theres an asset that seems perfect for such things. Again though i'm hoping some clue is given on how to illuminate and shadow the volume,as i keep going on about, i'll be simulating heavy ashy smoke, nothing like a flame so the shading would have to be distinctive

@Marionette - I've got a few choices although some migt require patience regarding coast effects, going the actual 'arty' implementation with suimono would mean changing the type of grid to a square rather than radial (im a little clueless here so ill enjoy some education while learnng) and then still enjoy suimonos teselation after the grid was displaced by the fft calculations. As far as i understand it, it would let me use suimono's method to map coasts so i can add foam and suchlike. That ocean is i suppose obviously gpu based and getting height information from the gpu to the cpu can be a performance knock, ive not tried anything with it really but i'd have to be careful how many times i actually try get the heights of a surface

So yes, a lot of the spray, edge foam, surf and suchlike still has to be done, nevermind figuring out when and how to employ unferwater effects. The alternative is to watch scrawks progress on his blog and see what does anddoes fit

Sorry for the the derail anyways, i hope to have a few questions answered so i hope the fellow gets back shortly

 

they cheat quite a bit with the lighting and they aren't volumetric clouds, however they definitely pull it off

it's definitely a believable cheat

 

- Presets: a pulldown menu or set of prefabs with various awesome setups... flamethrower, burning fire, heavy smoke, dry ice, torch smoke or fire...
- Burst control: the ability to create various fire and smoke bursts. Ive done it using animation frames, moving up and down quickly.. creates a beautiful mushroom type fireball. Maybe the translation can be done in the script with value or slider?
- attaching fluidity to dynamic objects can be amazing, but needs a large area to pull off well. can the attaching of the zone to a dunamic object be improved? can there be an easier way as well?
- Fade from camera distance setting (fade off at a distance and/or close distance to camera)
- fade from contact/near tag/layer set object
- global max number setting with fade off of oldest (like footprint decals, so we cant blow up the scene)
- Playmaker scripts

Thanks!

 

Not sure if this is already implemented in this, but I know the projects we would use this for could really benefit from this, collision planes, so you can have multiple planes for the fluid to collide with (box out a room, or a ceiling, so smoke would say drift up stairs into the room above realistically)

 

While on the topic of ocean scale navier stokes, I'd like to share a demo I made a couple months ago. I developed this technology for as a freelancer for a company called CANDELLA Software (basically, don't ever accept any jobs from them as they didn't pay me in the end and ignored all of my contact, after I delivered).

Here's a link, forgive the crude nature of the demo, I was hired to make a navier stokes river, not model the world.

::River Demo

 

That river is great! It is a little rough but the point is communicated perfectly well, is there any way of packaging that, or something based around that up, as an asset store thing i guess, i dont know how generous you'd be feeling, but yes, would be very useful to me

I dont know if you are pressed for time or just got bored of me waffaling on but regarding some of the things i was discussing earlier in the thread, do you have any thoughts? Its quite a big deal to me as its a significant part of the idea im putting into practice and a can see only fluidity in terms of unity assets as being able to simulate a lot of the stuff going on in a volcano like i had an example video of, on all scales. Although i do have friends in the computer graphics profession they dont have the time to look at fluidity and have a clear idea how it might be implemented, so obviously your familiarity with your product is valuable!

I suppose theres a lot of questions to be answered in this thread though.

Also I don't know if you saw this http://forum.unity3d.com/threads/117569-FumeFX-amp-Unity but i kind of put 2 and 2 together regarding how good something like fluidity could be with fumefx simulation data, not sure how immediately id need something like that but it sure would be interesting to see

 

Fortunately for you the size of the volume isn't necessarily relative to the rendering cost, it's mainly the amount of screen pixels covered by the volume and resolution of the grid that will matter for you.

This is an interesting one, I think a volcano is a great idea for a demo (*bumped*), and here is my thoughts;
- You will definitely need lighting, that's what's going to bring out all the intricate detail for you. I'm currently working on this.
- You could combine pyropuffs and fluidity here, but at a heavy cost, I've seen people do this kind of thing before and in my opinion, the visual quality isn't worth it, so I wont be working to include this in Fluidity any time soon. Instead I would recommend using vorticity confinement.
- The sort of resolution you are going to need might be on the scale of 200^3 +, which is obviously quite expensive once you add in all the pressure steps and you're definitely going to need vorticity confinement, I guess it could be quite handy to dynamically change the resolution of the grid... And I think it's possible.

Lastly, sorry I've been away for so long, but it's the run up to Christmas, and it's doubtful I will get anything done until next year

 

Cheers for the response! No worries on doing anything before xmas, everyone should be having fun! I was just hoping you'd have something to say soonish beause I was concerned about not having a particular route if i couldnt figure on fluidity, it seemed perfectly suited. You are quite right about the lighting, im gathering reference images of particularly stunning photos of volcanos, not just from Eyjafjallajökull but also in particular the eruptions at Puyehe, Chile a la:



And i'm slightly a photographer and I understand the power of a long exposure and photoshop but would sure be something. Since the volcano will be seen for only an hour or two in real-time, I think your simple but perfectly explanatory 5th example scene could provide much of the atmosphere, but yes, some very opaque fumes with some shape to them and good lighting would be lovely, my main graphics wizard pal was emphasising the importance of proper shadowing too, if just upon the volume. I think much CAN be dealt with particles (Much of the projectiles and such,so thats what i'd use for that but cheers! You've helped a lot, and you know what, it WOULD be a good idea for a demo haha, cant say I wouldn't capitalise on that

Have a good chrimbo and i'll fool around with fluidity a bit to see if I can get the shapes i'd need but it def does rely on lightig a lot (Im in no rush, mind) - I'm intending to purchase the turbulence noise asset because theres some stuff i know it will be handy for but do you see it having any use in shaping the billowing smoke and even explosive smoke?

 

I finally got around to putting fluidity into my Oculus rift level, I'm seeing a really bad offset when I move my head.
It generally stays in the area it's supposed to be but moves around when I look around. I didn't see this before since I had only looked at it using an emmiter without mesh near it.

update: Fixed! for others... you need to move the volumetric renderer above the OVRcamera for both cameras (in the inspector)

 

I must be doing something wrong. I purchased/imported the package, set project for dx11, and can run the included example scenes without errors, but I do not see any of the fluidity effects in any of the scenes?? I get no errors in the console. Any ideas? (Using Win 8.1 x64, Unity 4.3Pro, Geforce gt 640 ).

 

Nevermind, turns out I just needed to update Nvidia graphics driver to 331.82 and now everything is working awesomely

 

I love fluidity! Using it in my game http://www.headtripgames.com/iomoon
Will have a video for you soon, thanks for the awesome tech! Especially adding oculus rift support

 

Love the video dude! Great work!

If you get those rocks to hit off things and fracture into smaller pieces that would be some serious eye candy!

 

Can the fluids in this simulation interact with each other? We have Fluvio right now but it's too slow for our needs.

 

Hi BobBobson!

Yes, multiple fluids can interact, so long as you are using multiple emitters which belong to the same grid.

 

Hi,

I couldn't resist buying your package. It's funny to play with but there are a series of issues to make it actually usable in our project;

1. Can't start the game without an actual fluid on the scene or the main camera will render black.
2. It’s impossible to drop a fluid prefab in the scene at runtime or the fluid won't render. Funny enough when you move the emitter into another fluid area it will emit.
3. Fluids still stay very bright even when they are in shadows
4. Copy multiple instances of a fluid in the scene, then run it and you will notice that the amount of fluid in each has increased. If you delete the other ones and re-run the scene the amount of fluid is correct.

I suppose most of those shouldn’t be a problem but what about the shadow issue?

 

You're right, most of thoses arent a problem. Just seems you are using it in ways I haven't tested yet (which is a very good thing BTW )

Shadows are a problem though, as the fluid is essentially blended, I would need to perform a forward lookup into the shadow map at each ray march step. Not a problem (well it would be kind of costly, but there are things which can help with that) except I don't have access to the shadow map as rendered by the light camera.

Unless you are using a bespoke shadowing system, I can't help you there. Sorry.

 

Alright, damn I would have loved to use it for a central special effect in one of our game mode

This said we still might use it but not on fully dynamic stuff, more like level decoration for example.
Still hope to get number 1,2 and 4 fixed though ^^

 

Sorry I couldn't help you in this regard, but I can propose an alternative idea. There are assets available on the store which replace the Unity built in shadow systems. If you were to use one of these, I'd be happy to work with you to provide shadowed fluid systems!