Mitchell D. Harper
Motion FX Artist
TheMill x SCAD
SCAD Students from a variety of disciplines within the School of Digital Media work as a team to research, conceptualize and develop creative solutions for current design challenges presented by The Mill, NY. Our Team: Team Ocelot
Lorena Da Silva Diana Casteilvi Yanni Fong Elizabeth Delos Reyes Mitch Harper
Our Ideas
Week 2 has been busy with shooting our background plates and doing R&D for the portal FX. During the early stages of R&D it became clear that I would need multiple Pyro and particle systems to get the look I am aiming for. Week 2
Our References
Picking up where we left off last week, I had decided to use a volume based emitter and custom velocity fields. Unfortunately, this was just not giving me the look I was aiming for so I started experimenting with using two separate pyro sims, driven by the volume based emitter and the other driven by a particle based emitter. Week 3
Here is a recap of the Volume emitter and velocity field. For the volume emitter, I start with a simple torus and rough it up with a mountain node. Then it is converted into a volume and scaled up as the portal opens. The volume VOP adds some turbulent noise to the volume then we add a temperature attribute with the name SOP.
This is the basic pyro setup for the volume emitter. Just a few things to point out. A static groundplane was added and I created a proxy car geometry to use as a RBD collider. This should allow the car to interact with the smoke simulation.
Volume emitter flipbook
As stated earlier, I wanted to add a pop network to the portal to add some randomness to the shot. I created the particle emitter using a circle and scattering/jittering points onto it. I used a point wrangle to modify velocity over time. The circle is then scaled as the portal opens.
Particle emitter flipbook
Combined emitter flipbook
Week 4's major focus is to get the portal to "fit" into its surroundings. This will be done by using ground planes and adding other simulations such as leaves, dust and a little random debris/garbage. During week 3 I added a collider to the car which was interacting with the portal, however, the effect was minimal and needs to be emphasized more.
Week 4 PreViz Week 4
Car Collider I created a low-poly car to act as a collider for the portal and particles.
The car collider does not really seem to be having the effect that I was aiming for so this week is going to be troubleshooting that. Here is my DOP network for my particles. (click to enlarge)
Clearly the particles are reacting to the car and ground but the effect just does not translate once the pyro is added.
UPDATE: After looking through all of the collision networks I discovered that I had the RBD solver turned off in one of the collision networks. It has become pretty clear to me that I need to start simplifying this.
For the ground debris, I used a few particle systems with some basic geometry attached to get a feel for the overall look.
Here is the first iteration. It feels like everything is moving at once and the timing is a bit off.
Here, I have the 2 leaf systems moving with different settings causing more random movement.
Week 5 After studying several portal references from the Terminator I started to do some rough experiments. The "old school" portal seemed to revolve around a chrome ball and the newer ones look more like a spherical volume with some noise inside of it. Both of them have a LOT of electricity zapping around. Test 1: Chrome ball with pseudo electro-network effect
Here is a flipbook of the classic chrome ball terminator portal opening. I feel that the pseudo electricity effect is unsuccessful though
Test 2: Chrome ball with better developed lightning effect
Another flipbook, this time with a better designed electricity effect. To me this shot is still pretty slow.
Test 3: Noise volume with better developed lightning effect
Another flipbook, this time with a better designed electricity effect. I do like the concept of a noisy volume center but it needs a lot more work.
To achieve proper reflections, illumination and shadows from the portal we decided to export the Houdini pyro sim as a VDB to be used in Maya/Arnold. First we take the DOP I/O from our pyro sim and feed it into a Convert VDB. Make sure we preserve the density and temperature attributes! Then connect a File Cache node to export the data as a VDB.
In Maya, create an Arnold Volume and open the VDB as your volume. Once a shader is applied you can now use the Houdini Pyro VDB in Maya to create light, reflections and shadows.
Week 6
I have really been having issues with importing the VDB portal into Maya. When I create an aiVolume to import the VDB into it is created at 0,0,0. This is not where the portal goes so much of my time has been spent trying to get it lined up properly. Unfortunately, since we have been having problems with our rig and our animation, this has become a weekly process. Shading the portal has also been a problem. The default aiVolume shader does a pretty good job but it lacks a lot of the control I had with the Mantra shaders. I am sure a custom shader could be built, but I am concerned about the time constraints. Here are some rough renders of the Houdini portal vs the Maya portal:
These are the same portal at frame 72 (just before the car comes through). I actually just noticed that they are not even in the same place, clearly I have more work to do on lining these up properly...
I was able to get some work in on the portal in Houdini. The following video is actually a lot more successful than it appears. I forgot to increase the dissipation and decrease the density before running the flipbook. My goal was to get the smoke from the portal to follow the car as it goes through.
Week 7 This week I was finally able to get a copy of the Alembic that was being used to do the final car renders. The file also contained stand-in geometry for the windows of the theatre, which I can use to reflect the portal in. This is wonderful progress, and as long as the animation does not change again, its full speed ahead!
Shot 2
Shot 3
One of the main issues we are having with comping the portal is getting its color right. I decided to do some junk renders of a few color options.
The first one is similar in color to the lightning in shot 0 Week 8
Since there is so much red and orange in our background, I wanted to try a blackbody color option.
The blackbody option was actually not bad at all, it sits nicely with the background plate. With this in mind, I decided to try a green one. I think I should have stopped while I was ahead...
Scene Integration The cut from shot 1 to shot 2 does not feel right yet. Our group has discussed having a bright flash at the end of shot one and open shot 2 with a similar flash. Here is a rough render of that idea for shot 2.
Week 9 One issue that I have found with Mantra is that it is NOT accurate when using higher black body temperatures. I wanted to match the portal to the lightning in shot 0 which is about 30,000 kelvins. Mantra threw up a badly clamped red, pink and yellow. As an alternative, I attempted to match the colors myself. Here is a shot of the shader.
As for rendering, we determined that we would need a beauty pass of the portal, a layer with emission, a shadow pass and a layer of reflections off of the car and glass theatre windows. Here is a breakdown of the passes.
Week 10 It has been an amazing challenge to work on this project for the last 10 weeks. I would personally like to thank the team at The Mill - New York for their mentoring, without you, this project never would have happened. I would also like to thank the project professors, Deborah Fowler and Bridget Gaynor, you are simply the best that SCAD has to offer.
Final Breakdowns Render Breakdown
Ok, lets run through the pyro_source network! We start by dropping down a circle and resampling it. Then we transform its scale to create an open/close effect.
Next, we create the outer ring by using polyframe to calculate the bitangent, and then add some noise to the velocity field. After that, we use an attribute wrangle to dampen the velocity over time. Next, we scatter some points to our ring, jitter them and run them into 2 separate pop networks. (click on the pop network nodes to see whats inside!)
On to our "inner ring" of particles. We start with a timeshift node to clamp the start time to 24 frames (this will be used later in the pop network. Then we take a polywire node and use an isooffset to create a volume torrid. Scatter some points into the volume and we are ready to go into the pop network.
Density Field Particles
Moving on to our Temperature field, we start with a polyframe node to obtain the bitangent for velocity. Next we randomize the velocity field. We scatter and jitter some points on our circle and then run them into the pop networks. (dont forget, you can click on the pop net nodes to see whats inside)
Temperature Field Particles