Amazon 3D - An 80 minutes movie being produced with Blender

I am very pleased to announce that my team and I have been working on a full length 3D animated movie for some months now and that We finally agreed to use Blender for the whole production.

The title is Amazon 3D, and the final script just got registered at the WGC and HSE.

Synopsis

After arrogant engineer Sam Miller is hired to build the world’s largest power dam over the Brazilian rainforest, he is hurled into the forest on his own, where he will need to overcome his bad nature in order to survive and save his young wife who was taken by local bandits, and perhaps, even become a better person in the process.

Amazon 3D (temp title) is a classical hero’s journey with an environmental twist to it, and is aimed at family entertainment.

The movie will not be an open project per say, but after the initial festival run, all the production files will be made available to everyone under common license.

Our goal with Blender is to push it as far as We can, especially for character animation, fluids and rendering (Cycles).

We found that fluids could be more stable, so We plan on having a programmer having a look at it and hopefully improve on it.

I will start a FB page and a blog for the project, maybe this week-end, and go through some of the material We made and see what I could share with the community for the time being!

I am looking forward in exchanging with the Blender community, and I will share Our workflow and experience whenever possible during the course of the production, also, some of my Teammates expressed a desire to make tutorials as production progresses, no promises here, but We believe that this could be a good way to give to the community!

I will keep You posted.

Cheers!

Paul-André

Hey, sounds nice, I’ve been roaming pointlessly on the web since Mango project is over. A new project to follow is always welcome! The synopsis sounds cool :slight_smile: Do keep us posted indeed!

Daniel

well what can I say… the project is big and I already give you my congratulations just for beginning it! I really hope to see it finished! Good luck!

Sounds awesome :slight_smile: Advantage of such a huge project is that once you come to final renders, Cycles will be much more complete and suitabe for such an undertaking. Best of luck to you guys!

Awesome! Don’t forget to tell us when you get a blog up!

sounds really cool! especially if you get a programmer that can improve the fluid sims while you’re working on it.

Thanks everyone for your support! :slight_smile:

@Dani
No need to wander anymore, You found a home! :wink:

@Sdados
Thank You for your support, and We are dedicated to see this through, so We will! :slight_smile:

@BrilliantApe
Yes, Cycles looks pretty good so far, although We are developing characters with BI at the moment, but everything else is Cycles, so let’s hope Brecht can brew a more finished product in the timeline he planned, but I am sure he will! :slight_smile:

@gregzaal
I will, it will probably be a blogspot blog, nothing too fancy, but peoples will be able to follow some of the development of the movie on it, same on Facebook.

@aermartin
The first idea was to struck a deal with Next Limit and build a Blender plugin for Realflow, but after testing, I doubt that Blender could handle the tens of millions of polygons that a RF simulation generates. Besides, an integrated solution is always best, so We decided to go that route.

Some of the areas We look to improve…

Stability, especially with high resolution simulations and simulations with several obstacles.

Speed, baking takes way too long with high resolution simulation, so We’d like to improve on that.

Better behaviours, like with sticky particles, which often did not look too good in Our tests.

A preset gallery, with preset from whipped cream to gaseous liquids, and few other things.

Fluids are very important for this project since the backdrop is the Amazon forest, so improving on the fluids system is very high on Our to do list.

I may have secured one of Softimage programmers here in Montreal, since the m@r@n$ at Autodesk are moving the Softimage development team to Asia (so pissed at Autodesk these days), I know some top 3D programmers who will probably be looking for a job soon, if not, maybe a programmer from the Blender community, will have to look into that.

This of course will depend on the funds We’ll be able to muster for programming, but I am confident We will be able to make it work much better then it does now!

This week I will go through the tests We made so far, and posts some here and on the upcoming blog, again, thanks for your support!

Cheers

Paul-André

Sounds great, i like a good adventure movie. :slight_smile: Best of luck!

You are going to improve the fluid sim? Wow that would totally rock!

Btw: Maybe a mesher for the fluid particles would be a good start?

Second this.

A prober integration of This Paper mentioned on Blenderstorm would be way more usefull (imo).
There are a couple of Python Script (including mine) which does something similar with Blenders Metaball-System (just way to slow)
Here a Sample from 86’000 Particles. (Sim was more like 1’000’000, just used top “layer” of Particles)
According to their Paper the Time for 217’000 Particles is about 135 Seconds!

Nevertheless a improvement of current Fluid System is very welcome any way. :wink:

Good Luck with this interesting Project.

This is great! 80 minutes is a pretty big challenge, Best of luck to you and the rest of the team.

@quollism @Guss many thanks for your support! :slight_smile:

@GottfriedHofmann, I was thinking the same but for Blender SPH fluids, something similar to Softimage polygonizer for Lagoa.
Here is an intro to Lagoa and it’s polygonizer, Lagoa rocks, and the polygonizer is very fast as You can see!

I have contacted Thiago Costa, the maker of Lagoa, in the hopes of getting more information, maybe a paper or two!

Blender also has some basic tools for stuff like mist and maybe foam, which would also be great!

@Bashi, awesome work man, and thanks for the links, on my side I have a few other papers and I am awaiting one from the guys who made this simulation with custom 3D code, they say its pretty fast!

EDIT: Bashi, just read that Blenderstorm post, pretty much what I had in mind, thanks for the link but the link to the paper seems to be down!

PA3D: Was also talking about a mesher for SPH fluid particles :slight_smile:

Someone has already started working on a basic mesher for Blender particles, there is already a patch for 2.60:

http://pfs.planetblender.org/?p=35

And here is a demo:

Maybe you can start from there…

@GottfriedHofmann That looks pretty good, I downloaded the patch and will compile it for Linux64 as soon as I have time, thanks a bunch, as it looks like a promising start!

I am having a meeting with a 3D programmer next week, will try to gather as much code and info’s till then, thanks again Sir!

@MarkJoel60, sorry did not see You there! :slight_smile:

Yeah, my English is many things but perfect and I cant seem to be able to change the title, but thanks for the heads up anyhow!

OK, been real busy here and I did not have time to go through the material We did recently, so I asked for a screengrab of a character being modeled, so here it is…
Comments are more then welcome!

I have queued some of the fluids tests We did, hoping to be able to render them in a fashionable time, I will post these as soon as they are done!

Also, I meet with a 3D programmer on Monday, reserved 2 hours to get a good start on the fluids optimization project, so, if anyone has suggestions regarding the ways fluids could be made better in Blender, please share your thoughts on the subject and I will discuss them with the programmer.

Thanks

Bit depending on what way you go, but for Me best Blender Fluid would Be:

-Particles based (may be even opengl/gpu, later. see http://andrewfsu.blogspot.ch)

-resonable Fast Mesh/Skin generation… (whatever that means :wink:

-Motion Vector generation

-Texture generation or similar (foam and so on)

-works together with all other sims ( :wink: (i know, i have to wait another 5 years for that :wink:

-Multiresolution? Like Calculate Lowres Scene, and calculate Highres for scene or certain areas? (just a thought)

thanks

Hello again, still very busy here, but I have compiled a list of what should be discussed on Monday’s meeting with the programmer.

A way to create a fast and efficient meshing solution for the particle fluids for rendering, that includes the fluids and splashes.

A way to create foam and mist particles with texture support.

A system for open large body of oceans with foam and splashes, probably based on the Tessendorf algorithm.

A system for small scale, very precise simulations, probably based on Blender SPH particles system, with foam, mist, splashes and so on, with better interaction with objects.

Grid based simulation, for large scale simulations, like a flood or detailed oceans, with splashes, mist and foam.

I have gathered several other papers on fluids simulation, I will go through them during the week-end and try to sort the good, or relevant ones, to keep things as simple as possible for Monday morning!

@bashi, thanks for the link, I will look into it as well!

As for interaction with other simulations, could You elaborate a bit?

When You say multiresolution, do You mean of a way to generate less polygons in areas that are further away and more for the areas that are closer or do You mean a way to optimize the mesh according to details as in adding more polygons in detailed areas and less in less detailed areas?

Some examples of fluids

An example of splashes and foam (Realflow)

Shore simulation (Realflow)

Shore animation (Naiad)

Object interaction (Realflow)

Waterfall (Naiad)

These are high quality simulation made in two of the most used fluids simulation commercial software, and although the goal here is not to make Blender as good as these as far as fluid simulation goes, it would take too long and cost too much, the goal is to make the fluids in Blender much better, more stable, faster and more usable then they are now!

EDIT: Second link was same as first, I fixed it and added the right link!

are they going to be committed to blender once the movie is done?

The plan is to have the community tests them as they are developed, and will be made available as soon as the project is complete and hopefully become a full part of Blender, this is going to be a community project!

@PA3D, i actually had the Fluid Simulation itself in mind: Calculate for example a ocean surface with a boat at Grid Resolution 120 or so, and then “refine” a defined area, in this case around the boat at Grid Res 400 or so. This way you would have a low Res Simulation for the Ocean and High Res around Boat for Splashes and more detail.
But i guess you might like to keep development on Mesh creation?
Your ideas both sounds interesting too. probably even necessary, at least second one, for this multi-resolution thing i’m taking about.

PART II:
Since the Simulation is Particle based, i’m not really sure how i think this multires could work. An idea is kind of make Low Res Particle Cache and use this Cache to Simulate additional Particles or even change Cache. But probably not so trivial…