The Squid wants to know:

Have you been successful in creating realistic looking fire with Houdini? If so, can you show us a single frame, or better yet, lead us to a URL with a sample animation?

I'm working on my own procedural solution for fire in Houdini right now. but I'm interested in seeing other peoples attempts.

I'll be sharing my results as well. Lets see who can make the best fire!

-The Squid
:idea:

Since nobody has responded to this post I'll explain what I'm attempting to set up. Idealy I want a complete fire system, with combustable's (geo's that emit flames) and igniters (bounding areas that supply the flames “fuel.”)

Since there is no fluid solver currently available within Houdini, I've been experimenting with a cheat that has achieved good results but not GREAT results, GREAT results are my goal.

Currently I am taking a geo, lets say a sphere in this case, and isolating a group of points using my bounding area a.k.a “igniter.” I group the points, delete everything but those points, remove shared edges on the surviving polys and copy the new closed curve up in y space. I add a y noise to the new raised copy and skin between the two. I spline the columns so I have about 4 points per line. I group the base points with the ignition bound and use those as “fixed points” in a spring. I animate the poly as a softbody with very low mass and no drag. I texture the deforming poly with a VEX krinkle shader and light it from the base. Alpha para and perp are set to 0, so only the specular hits render with values over 0.

The lights are attached to points at the base of the mesh, so they appear to flicker as the mesh deforms. The effect is rendered with “deformation blur” in mantra which really helps sell the look of flames. I'll attach a still of these flames once I get home from work and have access to my file.

I've achieved this effect nearly flawlessly in the past by first rendering out an animated flame texture that looks like you're traveling through a tunnel of flames, basically sprites emitted from 0,0,0 traveling outward in all directions. I took that texture and wrapped in onto my fire so that the center of the image was at the base of the flame and the outer edges of the image stretched to the tip of the flame. With the flame deforming and the texture constantly moving outward, I was able to simulate some nice looking fire.

Unfortunately, the texture method is not entirely procedural and requires the flame to be a certain shape in order for the projection to work well. I'm trying to achieve the same effect using the spec hits from my VEX krinkle shader. I have an animated offset in Y but since the mesh is constantly deforming the area's of spec hits change from frame to frame. Animated offset or not, the end result still looks completely random.

I'm going to keep going along this path but I'm interested in hearing other ideas.

Some folks at Side Effects attempted a fire test but the results were absolutely terrible. Infact, the reason I'm doing this is to prove that Houdini can create realistic flames so the company I work for will purchase a license. They wanted to purchase Houdini in the past, and asked for an example of how houdini could create realistic looking fire, and what they got was so far from good it left a terrible taste in their mouths. I'd like to prove that houdini CAN create nice fire effects… I've already proven that I can make better fire in 2 days then the test they received from Side Effects (which took 3-4 days), but nothing remarkable enough to justify the $23,000.00 purchase. Maya fluids still looks better unfortunately. Any ideas?

The reason you probably haven'y gotten a response is because it's non-trivial. Houdini is a general-purpose FX tool but as you know contains no fluid dynamics(yet). Doing something flame-like(and good) using Pops is definitely possible, but as mentioned - non-trivial. The devil would be in the details. Comparing that to a system that already has a plug and play fluid simulator is apples and oranges…one has a plugin solution already built, one doesn't.

Personally it gets my goat when managers use expressions like “can <package x> do fire?”. It just belies a lack of comprehension of the process. Any package can do almost anything - it's a matter of how hard it is to do it. Nowadays the trend seems to be towards the plugin. No question, there's a lot of value in being able to spit out cloth from Syflex, and I personally look forward to a great dynamics simulation in Houdini. Sounds like they're going about it properly, from what I've heard.

Sorry - nothing concrete to add - I'd just be fiddling in POPs and working on shaders if I had to do this - but I'll admit it takes time to get a nice result. Also - there's fire and then there's fire. Fire from a torch? An explosion? A match? They would all be quite different in terms of a shading solution in my books.

/rant off

J.C.

I'm afraid I haven't done anything quite like this since I was a student but since no one else is replying I thought I'd add my 2 peneth. I would keep going the way you are with the geometry but try to do more with the shader. You can do loads of nice flame effects with noise functions in VOPs or VEX. Since you are using meshes you should be able to control the effect up the length of the geos by using the parametric s and t values. This way you will also get much more procedural control over your simulation but you won't need fluids ( which I imagine can be somewhat unpredictable ). The whole thing should be nicely animatable and of course you can use geometry attributes to override aspects of the shader in specific areas.
If you haven't already I would post this question over at odforce. They love this type of thing, you'll probably get several people giving you examples.

The other question I would ask is can you do more with sprites? I think that's how they ended up doing the dragons flaming breath in LOTR.

here's a dodgy method that works in busy scenes:
rendering motion-blurred particle sprites
(with a radial gradient) with other objects as mattes
and then treating (carefully) in 2d with the old 5D
(now speedsix) glass filter.
this only good if its nowhere near the camera. there is
also a slight offset from the matte after using glass.

still waiting for aura:
http://www.chaosgroup.com/software/AURA/ [chaosgroup.com]


-cpb

“Personally it gets my goat when managers use expressions like ”can <package x> do fire?“. It just belies a lack of comprehension of the process. Any package can do almost anything - it's a matter of how hard it is to do it. Nowadays the trend seems to be towards the plugin. No question, there's a lot of value in being able to spit out cloth from Syflex, and I personally look forward to a great dynamics simulation in Houdini.”

Gets my goat too. But I think it's a reasonable question to ask when you're bidding a project that requires a good fire effect and you're trying to make a decision as to which software package to invest in. Let's see… Houdini costs $24,000 and Maya Unlimited is $7,000. There are thousands of Maya artists and about 50 Houdini artists… Maya comes with built in fire using its fluids solver and Houdini doesnt. However, Houdini offers the artist more flexibility… hmmm… we have two weeks to turn around a good fire element to the client… Okay, as long as Houdini offers a good fluid solver I'm sure a Houdini artist could make their own fire pretty quickly… wait, what? NO FLUID SOLVER IN HOUDINI???

Looks like we'll be going with Maya Unlimited.

But hold on a second… I just watched the Side Effects reel and there are a ton of shots in there from The Day After Tomorrow that seem to use a fluids solver. How is that possible? Wait a minute… those shots were done at Digital Domain and required a team of people working to build new tools into the software to make it possible? So they had to literally build their own toolset to get the job done in a 24,000 dollar package when the 7,000 dollar package comes with it built in??? Hold the phones!

Houdini puts stuff in their reel that isnt even possible with the included tools? Did I miss a disclaimer somewhere???

You see, it's very hard for a Houdini artist at a non-Houdini facility to survive. I know from experience. I've had producers SHOW ME the Houdini reel and then ask me, “Why doesnt your water look like that?” Not everybody is an enlightened Houdini artist. Some people are “running a business” and want to see results so their clients pay them lots of money, which they need in order to pay for $24,000 software packages and $50 an hour artists. And some of them are even naive enough to believe that what they see in the Side Effects reel was actually done with tools built into Side Effects software!

From a money point of view (and lets face it, businesses rely heavily upon real world economics), Houdini is the worst choice of all the packages. After dishing out $24,000 you have to pay an artists several thousand more before you see a good fire simulation? Great idea! Let's reinvent the wheel at the cost of our employer!

Houdini is awesome, dont get me wrong. I WANT to use Houdini, but defending Side Effects is like defending the Dodo… sure it's amazing, unique, and really cool – but if it can't evolve fast enough to compete in this changing world, it's extinct.

How about updating the Downloads>Shader section which has been “currently under construction” for about 2 years??? Come on, get with it!

:evil: :evil: :evil: :evil: :evil: :evil: :evil: :evil: :evil:

So, if Maya's fluids are so good, why does everyone write their own? If Maya's fire is so good, why not use it? Why do DD and R+H (where I work) and so many others write their own? Well, because the out of the box “solutions” are limited, closed and hard to modify.

I think there is a reason that the hard stuff (fire etc) is done at large facilities Because it's hard and needs a large team of people to create it. I don't see a lot of flying logos being done at R+H for that reason, because they are (usually) easy and can be done by anyone, anywhere (with creative talent of course).

Instead of bemoaning Houdini's lack of a fluid solver, why not work somewhere like DD or R+H or Sony where you'll be encouraged to use Houdini for what it's good at, with lots of cool extra tools )

Having said that, as someone who also does projects at home, I share your pain…

Cheers,

Peter B

Hey man, I agree 100%. I personally dont love the out of box solutions either. Maya's Fluids get you good results fast, so long as what you want can be achieved with the limited options made available to you. A lot of times I've been able to whip an element out in Houdini a hell of a lot faster than I would if I went into Maya and started messing around. Houdini's environment gives you huge flexibility by giving you tons of tools to be creative with. I love Houdini – don't get me wrong. But after working at a couple of small houses I can see the other guys perspective.

I've been in situations where I'm sitting right next to a “fluids artist,” working my tail off building tools in Houdinii to achieve a certain effect. Maya boy clicks a couple of times, hits *render,* and BAM! Instant gratification for the superviser. All of a sudden the guy who can tweak preset parameters gets put in charge of the entire effects team – while I, an experienced Prisms/Houdini artist am left in the dust, answering questions like, “Why have you been working on this volumetric smoke for 5 days? Maya Fluids has smoke that looks better than this in 5 minutes.”

I want a fluid solver in Houdini. I want a real solution for rendering volumetric smoke, none of this limited i3D crap. I want to be able to make a body rigid, and have it react properly. Is that too much to ask for in a $24,000 software package? I dont think it should be.

As for working at a company like DD or R&H (Personally I'd much rather work at R&H) or even Sony, I already have. I like to take breaks from the big boys and work at the smaller houses. When I do, I like to bring Houdini with me. But the cheaper competition out there (software-wise) has made my arguement to purchase Houdini less and less appealing.

And again I should emphasize how misleading it is for Side Effects to use work done at Sony and DD and R&H as an example of what can be done with Side Effects software. Since when is it okay to sell software based off of images created by somebody elses proprietary tools? Atleast have a disclaimer that reads;

“The work you are about to see was created within the Houdini environment but could not have been possible without years of research and development by some of the industries leading artists who, as a result of industry demands, were challenged to create new, better tools to allow for fluid solving, cloth dynamics, rigid bodys, and voxel rendering. Houdini does not come with these custom tools, and none are available to purchase. You will have to develop them for yourselves, while competing with companies that already have these incredible technologies in place. Good luck!”

:shock:

From a money point of view (and lets face it, businesses rely heavily upon real world economics), Houdini is the worst choice of all the packages.

Before even making any kind of defense to this, you have to think about it relatively to each person/company.. does it make sense for all examples in your arguments of every small shop out there to invest in houdini.. probably not speaking from a money stand point as you've stated.. but to generalize it as a whole as being the worst choice of all, i would say you're mistaken.. numerous times it's been mentioned and talked about here, on the mail forums, and odforce about how well houdini has integrated itself in to the pipeline, allowing for many companies to not shell out tons of dough on support and mel wizards.. also realize that not every machine needs a master liscence.. maybe only td's and such. so after comparing a few good houdini guys with an expensive package or a good number of expensive mel experts with a cheap copy of maya doesn't really follow your money point of view after all..

And again I should emphasize how misleading it is for Side Effects to use work done at Sony and DD and R&H as an example of what can be done with Side Effects software. Since when is it okay to sell software based off of images created by somebody elses proprietary tools? Atleast have a disclaimer that reads;

ok, and this is different from any other package boasting on how their product was used in high-end production? so the fact that gollum was placed all over maya's site for how long is now supposidly leading customers to buy maya because they now can make their own personal gollum all of the sudden? no, only a fool would buy in to that (and im sure there's a lot of fools out there.. that's what the advertising is counting on).. it's up to the individual person to do some research on how things are produced with these products before shelling out 24 grand on something, not SESI's or any other company's.. of course they're gonna show off how people are pushing their package to the limit.. that's good business for them.. could you really expect anything less?

I shouldnt have generalized so much there. It's true that my own position is going to differ with somebody else depending on what it is they're doing, where they're working, etc. My perspective is unique. I guess what bugs me the most is having such an expensive package without a good fluids solver, a good system for rigid body dynamics and cloth/softbody simulations, and no real good way to render volumetric smoke. i3D can work for clouds and such, but it's motion blur is just a cheat which stretches the smoke texture based on velocity, so you can say goodbye to nice fluid-like results.

Your analogy with Maya and Gollum and MEL programmers is a good one. I know how expensive a good crew of Maya artists can be. I've used Houdini to some advantage at small houses, but the wheels have always stopped when it came time to attempt a rigid body simulation, or a really nice, fluid-like fire/smoke simulation.

yep, what Peter wrote.
for fluids use maya, aura or code (or cut ‘n’ paste) your own.
for cloth use simcloth (fast, free).
for ridgid-body dynamics use reactor (odd, but quite fast).
these are just the ones I've seen.
so the tools are out there, just be a mercenary
and swap & change according to circumstance.
user loyalty to a particular app may retard its
development as its continued patronage is not a
reflection of its relative capabilities.
but complaining is good, especially if you can go
to the trouble of spelling out exactly how to
fix/implement what you're after.
But I also would love to use Houdini for production,
but in its current state (given alternatives) and
with my limited experience there's no way. So I
try to make time for experiments and complain on the
forums about specific problems. so far the responses
have been fantastic so I'd suggest a similar slowly,
pedantic approach.


-cpb

squid:

hope i didn't come off too strong back there.. was rereading and noticed that sometimes when i'm making a point, i come across as an ass.. :wink: but yea, your points are very good ones and i know i've been itching like crazy to get my hands on some better dynamics (rigid, soft, fluids, etc) within houdini.. (and so many others as well, which is why it seems like SESI is really pushing it for 7 gold release.. can't wait..)

cpb:

or fluids use maya, aura or code (or cut ‘n’ paste) your own.

:shock::shock::shock::shock::shock::shock::shock::shock::shock::shock:

i can't believe you've forgotten the wonderful world of realflow.. forshame. poor poor mark and his beautiful hard work…

realflow… never tried it. I'll have a look next time.
there's also glu3D but haven't tried that either.
I've heard that both require a bit of work to avoid
the slo-mo-custard look.

-cpb

Hey Dave,

deecue
i know i've been itching like crazy to get my hands on some better dynamics (rigid, soft, fluids, etc) within houdini.. (and so many others as well, which is why it seems like SESI is really pushing it for 7 gold release.. can't wait..)

You better not wait because dynamics won't be making it for Houdini 7.0, AFAIK.

Cheers!
steven

First off, let me say that I agree with you on the whole lack of sim tools in houdini. It's a bit of a sore point…

And secondly …

the_squid
But hold on a second… I just watched the Side Effects reel and there are a ton of shots in there from The Day After Tomorrow that seem to use a fluids solver. How is that possible? Wait a minute… those shots were done at Digital Domain and required a team of people working to build new tools into the software to make it possible? So they had to literally build their own toolset to get the job done in a 24,000 dollar package when the 7,000 dollar package comes with it built in??? Hold the phones!

Good luck trying to do Day fluids in Maya. DD has Maya, why don't you think they used it for those? hmm.. I wonder, maybe ‘cause maya fluids suck when trying to do anything other than a preset?
I worked on that project, and did my fair share of the fluid sims for those shots and let me tell you that there is no ’off the shelf' package that will do that for you. And if you think it will then you're having a nice trip to lala land.

But seriously, if you're trying to justify to someone to buy houdini to do a couple of small fire shots, then I doubt you're going to win. However, if its a large project that involves an ass-load of fire and the power to modify it at will throughout the entire pipeline without going back to the drawing board, then you'll have fighting chance.

Cheers
Marc

P.S. What's the bet that the Day water shots are on the maya/xsi/lightwave reels too? That lot has never been exactly above board with their choice of shots to put on their reels.

True Marc, Maya Fluids does suck if you're trying to do something other than whats there in the presets. (Your shot in DAT looked great by the way) I've had to use Maya Fluids a lot on my current project and its been a double edged sword to put it lightly. You're right, what DD did in Day After Tomorrow wont be available in any package for atleast the next few years. I dont expect any package to give me that kind of result out of the box. My biggest issue however is the lack of even a simplistic fluids solver, and again, a good rigid body solver.

No matter what a package ships with, a clients demands are always going to push the artist to make the software do something it hasnt done before. What I want personally is for Houdini to have rigid bodies that blow Maya's away, a fluid solver which is atleast at the level of Maya's, and a good solution for volumetric smoke (which could be incorporated in the voxel/fluid solver.)

In other words, I want the $24,000 package to be able to do what the $7,000 package boasts it can do. Even though real pro's know Maya Fluids is essentially a toy and setting up a good procedural effects pipeline in Maya is a total joke, it would be nice for those of us making the “sales pitch” for Houdini to our employer to have some of those toys of our own to show off.

stevenong
Hey Dave,
You better not wait because dynamics won't be making it for Houdini 7.0, AFAIK.

Cheers!
steven

thanks for the heads up steven.. i kinda knew that it would be pretty impossible to fulfill that many dynamics solutions, but i guess i was thinking there might be some improved rigid bodies (if not a complete overhaul) and maybe some cloth/soft body work.. maybe it was just rumors around the forums i was getting but for some reason i kept thinking H7 was going to hit home with dynamics big time.. isnt there even a cloth vis sop in H7.1.171 already tho?

either way, i'll wait for as long as i need to since im not about to give up on my weapon of choice any time soon.. 8)

the_squid
My biggest issue however is the lack of even a simplistic fluids solver, and again, a good rigid body solver.

yea, but whenever has houdini done something simply? :wink: and that's prob the reason why so many people love it as much as they do.. i know it's one of my big reasons that when SESI implements something new, they do it right and really well rather than half-assing it.. sure, that 7 grand maya liscence gets you a lot of “labeled” features, but who cares if it's crap. i mean really.. it's fluid solver? expression implementation? maya live? and from what i now hear about it's hair system? you know it's all shot in the dark with those.. atleast when houdini adds a feature, you know it's going to be solid.. id rather wait and get something well dedicated to than something now that is sketchy and unkept over the next whatever number of releases..

Hey Dave,

Everyone is hoping dynamics, specifically RBD & Cloth, will make it for H7.0 but they are not ready yet. During the Users Group meeting, they showed RBD & cloth demos which look really promising & IMHO, definitely worth the wait. There is a thread on the users group meeting & I described the RDB & cloth demo there.

As far as SESI is concerned, DOPs (Dynamics OPerators) has a long way to go.

Cheers!
steven

Mr. Ong

Where is the thread on the RBD and cloth demo?

Personally I cant wait for these tools to become available. It's true that when Maya implements a new feature they do it half ass. Nearly everything with Maya is half ass especially when it comes to effects, and definately when it comes to “support.”

Alias is really good at selling Maya as a tool. The people who purchase the software arent always the same people using it. So they see the paint effects and fluids demo's and think they're the greatest thing since sliced bread. Then they hear about houdini and the first questions are, “Does it have Fluids like Maya?”, etc.

When Side Effects implements a new tool or even a whole new set of tools they usually do it pretty seamlessly and integrate it well within the Houdini pipeline. I understand that takes a lot longer than just abandoning your old tools and building new selling points into your software like Maya does.

I guess we're all just a little anxious to get our hands on these new solvers so we can show the Maya folks what Houdini is made of. :wink:

Hi the_squid,

Here's the thread [sidefx.com] about the demo. The coolest thing about the video is the RBD & cloth sims were solved at the same time. Since you're so close to SESI, you might want to take a walk & get someone to show you the videos.

Yep, everyone wants solvers in Houdini now but everyone also wants it properly done rather than it being “half ass”.

Remember the old idialog words of wisdom?“Patience is a virtue.”

Cheers!
steven