Death Eaters Smoke trail

#12

Wow, some interesting stuff here. That partitioning stuff sounds like it could be useful.

I never personally liked the “smoke trail” thing in the Harry Potter films. Being a fan of the series, I can say that this never happens in the books, and in fact, it’s specifically stated that just about no wizard can fly unaided, so when Voldemort shows that he can in a midair chase sequence, it’s supposed to be a big deal. Wouldn’t be as big a deal with me, since Death Eaters are supposed to be all into the dark magic stuff, but then in the OOTP film, the good guys can do it too, with WHITE smoke trails, no less.

I dunno, I never liked it. It created this air of insurmountability when it came to considering what skills were within Harry’s capacity and how much farther it seemed to set the bar up from him, when during the book version of the Department of Mysteries battle, he actually did pretty damn well given he was uno-en-uno with a dozen odd notorious murderers off and on. I’d hoped it was a concept that would remain isolated to one film, much like the third one when they’d appeared to have cut the budget on effects for various spells and made them invisible save for their end-result, but no such luck as we immediately see more Death Eaters pretending to be the “Locke-ness Monster” from Lost in the next film…

#13

Yes. Most of it is fully synthetic (including sky, plane, missile and all of Washington, DC you see there), with Krakatoa being used for the Nanomites eating the plane and the missile trail which was simulated in FumeFX but had to be rendered in Krakatoa for memory limitation reasons.

The Nanomites were rendered in many layers using the Particle rendering mode of Krakatoa. The particle animation was done mostly using Particle Flow and Box #3 Data Operators.

The missile trail though proved to be tricky, because the scene was animated in more or less real-world scale AND in real-world speed. Thus, with the missile moving fast and trying to avoid the chasing plane, the FumeFX simulation grid had to be huge, and only a small portion of it was actually populated voxels. Still, it required around 64GB of RAM to render in Scanline or VRay and we did not have network machines with that much memory to do that at the time. So we tested with getting the data out of FumeFX and rendering it in the newly implemented Voxel mode of Krakatoa. Sitni Sati were very helpful in providing us with the FumeFX SDK so we could make Krakatoa create one particle per voxel directly from the FumeFX simulation and then render the resulting data a bit more efficiently. Since Krakatoa renders one “scanplane” at a time and not the whole 3D volume, it needed less than 600MB for the same data. It might have been slower, but it could be rendered on hundred network machines if necessary as memory wasn’t an issue anymore.
So the missile trail was rendered as direct FumeFX rendering, not PFlow particles driven by FumeFX.
You can watch a video of one of our Siggraph demos last year related to those effects here:
http://area.autodesk.com/player/loader.swf?p=/player/main.swf&f=http://areadownloads.autodesk.com/oc/ibc09/sig09_d2_frantic01.flv

You can also read more about Krakatoa in last year’s movies in my blog:
http://lotsofparticles.blogspot.com/2010_01_01_archive.html

#14

Fantastic information Borislav.
What can I said.
Unfortunately I don’t have all that Ram.
Actually my machine is a:
rocessor Intel Core Quad i5 750 2.66Ghz Sktll56 8Mb Box
Motherboard Asus Sktll56 - P7P55D iP55
Pack 4x 2048Mb DDR3 Corsair XMS3 CL9 PC3-12800 (1600) i5-i7

Do you think that 8 GB Ram will provide me some good results?
I’m reading your blog at the moment.

#15

What RAM do you mean? As I said, we used only 600 Megabytes of RAM for the trail to render in Krakatoa. With 8GB of RAM, you should be able to fit up to 300 million particles in memory, that’s an overkill even for most things we do on movies (except for a couple of shots in the past where we went higher, but using multiple layers).

Your machine should be fine…

#16

You can get some real decent fume sims with 8gigs. You may have to use a couple of grids depending on how huge your trail is going to be but since you mentioned it is a head moving from one side to the other I think you’d be well within reason.

#17

Thanks a lot Mr. Borislav.
Like I said I’ll post my tests as soon as I can.
If I have any more questions or doubts about the “manual” of the Krakatoa I’ll tell you.
It’s a shame there aren’t a lot of tutorials out there.
Well just have to study a little bit more and test and test until I get the result I’m after.
Cheers

#18

Hello everyone.
I’ve finished my final animation project but unfortunately because of my university I can’t place it online but if someone is interested I can send them a link from yousendit to download, but don’t tell no one lol.

Just wanted to let you all know that In the next days i’ll start to put some tests online.
Thanks for all the references until now.
If someone knows extra tutorials please share them with me.
Thanks for all

#19

I’m following some tutorials but it’s not easy.
The commands are a lot different.
For example I made FumeFx emitter and created with Particle Flow a FumeFx birth operator.
I’ve opened Krakatoa and rendered the particles but they appear flat black.
I now that if I had some lights I’ll loose that flat look but whats with that black thing?
I can’t find that Override particle colors to change that so I started to mess with everything and I’m using the Override Emission (Use) to change the color. Is this right? Why isn’t using the default color that is in the Particle Flow?
It is not easy to follow the basic tutorials that are online.
I can’t find that Particle Color Controls.

#20

I’ve started with some tests with FumeFx, ParticleFlow and Krakatoa and start playing with Magma Flow. It is not an easy tool, without the tutorials it’s very complicated to follow but it’s impressive the type of images we can get, really great.
Black and white cloud
http://forums.cgsociety.org/attachment.php?attachmentid=153166&stc=1
Color cloud
http://forums.cgsociety.org/attachment.php?attachmentid=153167&stc=1
Cheers

#21

Without what tutorials exactly? There ARE tutorials about changing color or density over age.
I would love to see a list of things you feel are difficult and require more tutorials, or existing tutorials that are hard to follow.

In pretty much every post you say “it is not easy”. I beg to differ, but what IS easy in 3D? :wink:
Have you ever used a node-based editor (say, Eyeon Fusion, PFlow Tools Box #3, Thinking Particles, or Softimage ICE)? Were these easier to use and how exactly? What would you like to see changed in MagmaFlow to make it easier for you (besides better documentation which I admit could use an update?)

Thanks in advance.

#22

I really love node based.
I used to work with Fusion, now with Nuke.
The thing about Magma are the concepts, what exact parameter does and what connections are best. The tutorials from the website are really good but lack a bit the explanation on what parameter can do, perhaps I didn’t reach that part of the tutorial.
At the moment I can tell what you could improve just need a little more practice to get used to the system. The lack of tutorials like the ones from Allan Mckay for example, there aren’t tutorials like these don’t know why. CGAcaddemy have some very good tutorials on PFlow.
The tutorial I was talking was that example from the GIJoe production. I’m in that one at the moment. I say mostly it’s not easy because I’m a maya user for seveal years and a lot of the process are different.
I start with Max about 2 months ago.
I’ll keep you posted about my doubts.
Really appreciate all your help.

#23

Fair enough, and quite helpful feedback, thank you!
We are in the process of rewriting portions of the documentation for the upcoming 1.6.0 release (you might notice pages changing daily) and I will try to provide some more in-depth explanations.

That G.I.Joe “tutorial” was just a draft for my presentation at the Autodesk booth last Siggraph so I had to keep it short - I could go into more detail why certain things are done for sure.

Unfortunately, I cannot develop a MagmaFlow DVD for CGAcademy because of conflict of interests. As co-developer of Krakatoa, I am being paid by Prime Focus to provide documentation to all our current and prospective customers. Developing such content for a 3rd party publisher and caching in on my internal knowledge would be amoral.
So I will have to find time to do some recordings and publish them for free sooner or later.

Once again, thanks for the feedback and keep it coming.

#24

So I will have to find time to do some recordings and publish them for free sooner or later.

That would be awesome.
I’ll keep you posted of my progress.
Cheers

#25

I’ve started for some time with the Krakatoa tests.
But there’s one thing I’m not understanding.
The movement of the particles is defined by particle flow right.
I’ve made an smoke animation in Fume and after that I activated particleFlow (fumeFx follow and birth) an then Krakatoa. Well when I made the partitioning in Krakatoa the simulation was completely ruined.
I’ve noticed now that the velocity channel wasn’t activated on FumeFx, is this an important subject?
Thanks
Cheers

#26

The velocity channel is required for the particles to actually follow the simulation… sooo yeah it’s important. :thumbsup:

#27

One thing I’ve noticed that sometimes they blow out the FumeFx simulation.
Why does this happens?
Should I increase the number of particles?

#28

They’ll blow out because the influence is probably a little too high, and/or because your sim just gets close to the edge of the grid, if a particle gets a velocity from the grid that’s high enough, it’ll escape the grid… the simplest solution is to turn down the influences, or just turn on the option that kills the particles when they exit the grid. You could also do some fun stuff with scripting (or box3 preferrably) to tell the particles to head back into the grid if they are getting too close to the boundry.

#29

Scripting at the moment it’s not my field.
But I tend to activate that option that says to kill the particles that go outside the container.
I’ll check that influence parameter.
thanks

#30

I’ve created smoke in FumeFx.
I really liked the look but I wanted to see how it behaved in krakatoa.
This isn’t the first time an animation I make looks very bad when I try to render it with Krakatoa. The rate in ParticleFlow is 200000. I think this is high enough to get detail.
In the part of the channels in Particle Flow I only have the smoke activated and the max value is 40 and the min is 1.
I have a spotlight in the scene but it doesn’t seems to have any affect on the particles.

#31

You need a lot more particles in your krakatoa render, try starting out with 2 million and work up from there. The idea is to achieve a really dense cloud that is more representative of smoke, rather than dust.

Give this a read: :slight_smile:
http://lotsofparticles.blogspot.com/2009_11_01_archive.html