PDA

View Full Version : Fire -like Max or Maya



erikals
04-03-2005, 05:11 AM
Hola. I guess this is not my number 1 request, but I think these two give pretty good results. Would be cool to see something like that in LW :)

http://zgrachan.com/images/thumbs/tallFire2_thumb.jpg
http://zgrachan.com/pages/portFluids.htm

http://www.chaosgroup.com/images/aurapic.jpg
http://www.chaosgroup.com/software/AURA/

erikals
09-29-2005, 06:47 PM
I'm bumping my own thread here...

Found the two .pdf docs from this page (http://www.cfxweb.net/civax/archives/00000070.htm) to be interesting

Explosion - (pretty cool, ILM -Fedkiw)
http://home.no.net/erikals/cgtemp/Bang-01.jpg
http://graphics.stanford.edu/~fedkiw/papers/stanford2003-02.pdf
Also see,
http://graphics.stanford.edu/~fedkiw/papers/


Smoke
http://grail.cs.washington.edu/projects/smokeControl/images/smoke.gif
http://grail.cs.washington.edu/projects/smokeControl/

(Important note?)
The exellent Lightwave smoke simulation below was based on Fedkiws papers (the guy who hosts the explosion.pdf document above). (If you are a programmer I'm sure you know who Ron Fedkiw is). Sadly though this simulator for LW posted below is in-house only, but it means the same can be done by the programmers at NewTek. (Edit: The developer name is Reevan McKay. He works for Topix Studios @ Canada. His email is http://home.no.net/erikals/cgtemp/reevanmail.jpg)

http://www.topix.com/reevan/RV_smoke.jpg
http://www.spinquad.com/forums/showthread.php?t=6005&page=1&pp=15&highlight=smoke
http://www.topix.com/reevan/nicesmoke.mov
http://www.topix.com/reevan/funkyflow_1000.mov

Gregg "T.Rex"
09-29-2005, 07:11 PM
I'm 200% :agree: with these all come in LW, ASAP !!!!!

erikals
09-29-2005, 07:24 PM
Great!! :beerchug:

http://www.cfxweb.net/civax/pics/flames_pic_tn.jpg
Here's another flame.pdf by the way from PDI Dreamworks (Shrek dragon)
http://www.cs.brown.edu/~tor/sig2002/StructuralFlames02.pdf

But the great thing is that the last smoke simulator shown on this page has already been made, so it should be possible to remake.
http://www.spinquad.com/forums/show...highlight=smoke

Panikos
10-02-2005, 04:10 PM
I am trying very hard not to comment, otherwise I will accused for aggresive attitute and Moderators will run in to settle me :grumpy:

Knight Chat X
10-03-2005, 12:21 AM
I am trying very hard not to comment, otherwise I will accused for aggresive attitute and Moderators will run in to settle me :grumpy:

All work and no play eh Skippy?

Yeah, more special effects would be cool Maya does a good job making that part simpler, so it's easier to implement those kind of elements in an environment.

Exper
10-03-2005, 04:58 AM
Panikos... I feel the same! ;)

erikals
10-03-2005, 06:45 AM
Just another link,

http://www.exocortex.org/ben/gallery05/smokeComparison.png
http://www.exocortex.org/siggraph/2005-2.html

Other links,
www-sal.cs.uiuc.edu/~yyz/publication/octree_sa.pdf
http://www.dgp.toronto.edu/people/stam/reality/Research/pdf/smoke.pdf
http://www.dgp.toronto.edu/people/stam/reality/Research/pdf/GDC03.pdf
http://graphics.ucsd.edu/~henrik/papers/fire/campfire.jpg
http://graphics.ucsd.edu/~henrik/papers/fire/

More interesting links,
http://www.dgp.toronto.edu/people/stam/reality/Research/pub.html
http://www.gamedev.net/community/forums/showfaq.asp?forum_id=12

lardbros
10-03-2005, 01:18 PM
Yep, this would be ace. I'm sure as the LW SDK opens up more and more we are going to see some killer plugins made for free and that are cheap.

These smoke simulations look incredible. It's more of a problem with LW's particle control than HV though i reckon. Although more controls within HV would always be a brilliant idea too. I had a look at the thread on smoke with Lightwave that Reevan McKay was working on and it's simply amazing. I hope he sees sense and makes it available for us lot.

I'm not sure it's a plugin though, more of a simulator outside of LW that can be used with it... if that makes sense?

erikals
10-03-2005, 02:13 PM
Yupp, just not sure how much work it is, it also depends on what NewTek wants to focus on, but I feel this could be a great bonus for LW users.

However, when 1 guy can do this by himself (as written above), then maybe they could concider to hire a guy just to do fluid dynamics. http://vbulletin.newtek.com/images/smilies/brians/phone_call.gif

Btw. Here's some more links I stole from Splinegod at Spinquad...

Smoke Fire and Goo (www.siggraph.org/s2004/conference/papers/papers8.php)

----------------------

Target-Driven Smoke Animation (www.cs.huji.ac.il/labs/cglab/research/tdsmoke/)

Raanan Fattal (www.cs.huji.ac.il/~raananf/)
Dani Lischinski (www.cs.huji.ac.il/~danix/)
The Hebrew University of Jerusalem (www.cs.huji.ac.il/)

----------------------

Fluid Control Using the Adjoint Method (http://grail.cs.washington.edu/projects/control)

Antoine McNamara (www.cs.washington.edu/homes/antoine/)
Adrien Treuille (www.cs.washington.edu/homes/treuille/)
Zoran Popović (www.cs.washington.edu/homes/zoran/)
University of Washington (http://grail.cs.washington.edu/)
Jos Stam (www.dgp.toronto.edu/people/stam/reality/index.html)
Alias Systems (www.alias.com/)

----------------------

Simulating Water and Smoke with an Octree Data Structure (http://graphics.stanford.edu/~fedkiw/)

Frank Losasso (http://graphics.stanford.edu/~losasso/)
Frederic Gibou (http://math.stanford.edu/~fgibou/)
Stanford University (http://www.stanford.edu/)
Ronald Fedkiw (http://graphics.stanford.edu/~fedkiw/)
Stanford University (http://graphics.stanford.edu/)
Industrial Light + Magic (http://www.ilm.com/)

----------------------

A Method for Animating Viscoelastic Fluids (www.cs.berkeley.edu/b-cam/Papers/Goktekin-2004-AMF/index.html)

Tolga Goktekin, (no link)
Adam W. Bargteil (www.cs.berkeley.edu/~adamb/)
James F. O'Brien (www.cs.berkeley.edu/~job/)
University of California, Berkeley (www.cs.berkeley.edu/)
.
.

aurora
10-03-2005, 03:22 PM
Keep those links coming guys! Most of them I already had found and printed off but some are new and great! Some of you are aware of some of the current plugin projects I'm working working on. But this is what I really want to write but the details for a truly good, fast, flexible fluids solution are still escaping me. So the more info I and or others can get the closer we'll be to being able to create one kick-a plugin which we all know we need in :lwicon:

aurora
10-04-2005, 08:50 AM
These smoke simulations look incredible. It's more of a problem with LW's particle control than HV though i reckon. Although more controls within HV would always be a brilliant idea too. I had a look at the thread on smoke with Lightwave that Reevan McKay was working on and it's simply amazing. I hope he sees sense and makes it available for us lot.

Well theres part of the issue. You would not use LW particles you need a custom volumteric render in LW, in fact I beleive Maya's fluid solution is a seperate solution then their pFX solution (can't remember for sure so don't take this as gospel) On the brightside its easy enough to create that custom volumetric renderer. Its figuring out the fluid solution thats the difficult part and alot of that is first deciding on the method you want to use to solve for the flows because theres all kinds of different methods each with there own pro's and con's. Thats why all these links are great. It helps in building a chart of these methods and figuring out which one(s) would provide the best bang for the solution.

Exper
10-04-2005, 09:41 AM
I'm sure you'll figure how to solve the thing. :thumbsup:

If you'll need a place for testers... I'll be happy to setup the proper stuff (you know my friend). ;)

lardbros
10-04-2005, 10:48 AM
Well theres part of the issue. You would not use LW particles you need a custom volumteric render in LW, in fact I beleive Maya's fluid solution is a seperate solution then their pFX solution (can't remember for sure so don't take this as gospel) On the brightside its easy enough to create that custom volumetric renderer. Its figuring out the fluid solution thats the difficult part and alot of that is first deciding on the method you want to use to solve for the flows because theres all kinds of different methods each with there own pro's and con's. Thats why all these links are great. It helps in building a chart of these methods and figuring out which one(s) would provide the best bang for the solution.

Ahem... yeah, i'll try not to sound like i know what i'm on about next time :D Trying to sound clever isn't my strong point.

aurora
10-04-2005, 11:19 AM
lardbros - I was not sayng anything about you or your comment. Just something about the needed nature of the solution. Sorry for the way I worded it, my bad.

lardbros
10-04-2005, 01:13 PM
Aarrrrgggh god****it!! I would have gotten away with it? Now i look stupid for confessing to no knowledge on the subject. :D

No worries, i'm not very easily offended (and no that's doesn't mean everyone can give it their best shot to actually offend me). I should really stop taking up posts with tripe! :D I'll get kicked out.

Basically i was just using my limited knowledge of a scientific subject in relation to LW and what i have observed.

Looking forward to what some of you extremely clever programming types can come up with. It's so beyond me it's not funny... even Lscript and expressions have passed me by so far. ****, even counting my fingers is beyond my scope. :)

erikals
10-04-2005, 01:53 PM
...even counting my fingers is beyond my scope. :)

I know, I feel the same. That's the reason for posting this here hehe ;D ... math is evil! :D

Aurora, you saw these right?
http://graphics.stanford.edu/~fedkiw/papers/

Edit:
Found these also.
http://www.maths.qmul.ac.uk/~hve/MAS209/
http://dragonfly.tam.cornell.edu/cfd.html
www.iop.org/EJ/article/1367-2630/3/1/309/nj1109.pdf
http://www.cc.gatech.edu/~carlson/
http://amath.colorado.edu/faculty/desterck/websiteW/publications/Publications.html
ctr.stanford.edu/Summer00/farge2.pdf
users.aber.ac.uk/sxc/WORK/konstanz_c30107.pdf

Again, might be double posts here.

erikals
10-04-2005, 02:16 PM
(I figured most of these articles are saved in .pdf format. Try too google for e.g. "Fluid dynamics" and ".pdf")

best,
Erik

Knight Chat X
10-04-2005, 04:00 PM
Actually, those documents and methods make quiet alot of sense.

It all boils down to the various formula's designed to handle different parts of the effects, so the equations give the full range of each desired effect.

The documents seem to fully cover the seperate areas/aspects, very organized.

Putting that into actual code is a bit different story, if you understand the math that supports the principles shown and develop this kind of software you can do it.

erikals
10-05-2005, 05:22 PM
One thing about flames though, it would be nice if the math was made so that it was easier to upgrade in the future, so one could have even more realistic flames. I'm not sure how hard that is though, I guess maybe it is too hard... Here's a reference for flames, you can see that the examples on 3D flames from Max/Maya posted earlier isn't quite realistic, though close, in some cases. As for smoke I feel that the smoke-simulations are a bit more realistic.

http://erikalstad.com/cgtemp/3xFlames.jpg

Edit:
Hmm, think I start to understand why most simulations have been made to look like the flames in example number (1) above. Example (1) is mainly the flame used for torches/lighters and stuff, and small objects that are cought on fire. I guess that's why developers have started with these kind of flames, as it often is moved around like e.g. a tourch. The flame examples in (2) (3) (4) are usually more static and therefore stockfootage of flames have been used instead. An exception was "Shrek" that use flames that look more like example (3) and (4). This was probably because of two factors, -that it was hard to use stock-footage, and -that they wanted a 3D look for the flames. Same goes for the nuclear bomb shown on the first page, it's quite hard getting a good stock footage of a nuclear bomb, especially when it's viewed from above. (See Terminator 3: Rise of the Machines)

So my guess is that the developers next step will be to make flames like in examples (2) (3) (4).

I haven't mentioned water and fluids too much in this thread, but I think that is more straight forward as far as the "graphic/visual look". One thing that I've wondered about though is when we will see great wave simulations. (No, not Realwave, which is ok, but not anything near realistic IMO). Well, for simple simulations Realwave is great, but it can't e.g. simulate how waves splash on the beach strandline etc. I think I've read somewhere that Fedkiw was looking at/ was interested in that problem though.

So, what's next?, :D grumble-grumble... :stumped:

Panikos
10-05-2005, 05:26 PM
Polas Fire&Smoke Generator is a very good tool in terms of simulation.
LW Hypervoxels were written for general use, they are slow and often crusial parameters are not shared between the simulator and the renderer.

:i_agree:

Mylenium
10-05-2005, 11:34 PM
One thing about flames though, it would be nice if the math was made so that it was easier to upgrade in the future, so one could have even more realistic flames. I'm not sure how hard that is though, I guess maybe it is too hard... Here's a reference for flames, you can see that the examples on 3D flames from Max/Maya posted earlier isn't quite realistic, though close, in some cases. As for smoke I feel that the smoke-simulations are a bit more realistic.

http://erikalstad.com/cgtemp/3xFlames.jpg

Edit:
Hmm, think I start to understand why most simulations have been made to look like the flames in example number (1) above. Example (1) is mainly the flame used for torches/lighters and stuff, and small objects that are cought on fire. I guess that's why developers have started with these kind of flames, as it often is moved around like e.g. a tourch. The flame examples in (2) (3) (4) are usually more static and therefore stockfootage of flames have been used instead. An exception was "Shrek" that use flames that look more like example (3) and (4). This was probably because of two factors, -that it was hard to use stock-footage, and -that they wanted a 3D look for the flames. Same goes for the nuclear bomb shown on the first page, it's quite hard getting a good stock footage of a nuclear bomb, especially when it's viewed from above. (See Terminator 3: Rise of the Machines)

So my guess is that the developers next step will be to make flames like in examples (2) (3) (4).


I dunno if you have read any of the papers you posted links to, but if you have, you should remember words such as volume coherence, micro vorticity and turbulence modulation all of which can be things that can make simulations extremely slow and hard to code (from what I gather). You are also completely forgetting that your other examples 2-4 look different because they are influenced already by collisions and external turbulences and thus look less synthetic, whereas your example 1 looks like it was done in a studio without external forces/ influences. If you would go through such lengths as to model those phenomenons or describe them with formulas, there is a good chance that your fire will look more realistic but at the cost of even more slowdown. I have played a bit with Maya's fluid effects a bit and with proper attention it is possible to come up with quite realistic stuff.

Mylenium

erikals
10-06-2005, 04:48 AM
No, unfortunatly I do not read the papers.

The thread was just ment to provide links to papers that could possibly help +hints about flames.

I see what you mean though, and I've been nagging myself abit about not reading the papers, but time is too short I'm afraid. IMO it was better to post ideas/advice than to leave it out. I realize that can be somewhat messy though, as all one really finds on this page are many links which are unsorted, and maybe even duplicated. +me throwing out ideas.

Could you post some examples of the realistic flames you made?

Knight Chat X
10-06-2005, 06:02 AM
Whatever works, personally I think if you can get the job done with math and you understand the basics to achieve desired effect do it, if there's a way to make things faster by using a simpler method in combination or stand alone solution then so be it.

The easier things are to develop and/or the better our tools are, the higher the quality the production and the more you can focus on other things.

Give you example of what I mean, let's say a person that's not a rocket scientist and doesn't have time to re-invent the wheel just wants to achieve a standard result which has been in use for quiet some time now and an effect others use, that person may just wanna click Add Effects/Flames/Torch, Add Effects/Flames/Explosion, Add Effects/Flames/Camp Fire, Add Effects/Flame/Ceiling Fire, Add Effects/Flames/Fliud Fire, etc...

How bout water effects, like Add Effects/Water/Faucet, Add Effects/Water/Lake, Add Effects/Water/Rain, Add Effects/Water/Storm, Add Effects/Water/Ocean, Add Effects/Water/Underwater, Add Effects/Water/Moisture, etc...

Or this, Add Effects/Smoke/Thick, Add Effects/Smoke/Thin, Add Effects/Smoke/Explosion, Add Effects/Smoke/Fire, etc...

Have ability to change simple settings, keeping it as simple yet flexible as possible and you got not only something that's easier to work with, but useful, quick, and it's easier to get that desired effect if you narrow things down a bit.

Give a person a feature that has a thousand options, and if the feature settings are not easy enough to understand to use right away and require the user to read multiple books with 1000 pages each just to get the desired effect through other sources or go to a university to take several classes of both math and physics that's just too much, not to knock the education system or anything but if a professional programmer with a PHD in computer science get's all worked up he just went through 4 years and learned how to popup a simple message box something is seriously wrong there, lol.

Anyways, keep those links coming if you got more because all that's interesting for things such as game engine development as well, it's hard to find information like that when people don't talk about it.

Mylenium
10-06-2005, 10:27 AM
Could you post some examples of the realistic flames you made?

Sorry, I can't. I don't have any footage of them and I'm just too lazy re-installing Maya (we dropped it, so I don't use it anymore) to get the few files I have working again (would have to re-run the entire calculation, anyway, which takes rather long). I wouldn't also call them realistic. I think I once tried to combine a 3D fluid container with a particle system where each of the particles acted as a Turbulence field and then had a simple expression that changed the color of the voxels based on their local velocity or something. Really nothing fancy, though it moves towards things like in your picture 3.

Mylenium