I'm so pround to take part in the SDGM 560 course and work with The Mill. In the next 10 weeks, I'm going to work with my teammate Adesh Kumar, Jonathan Sugijono, and Ryan to create a professional standard commercial for The Mill!
Week 1
Working with my teammate, we made our team's name as 4 corners. We decided to work on branding the new ID. buzz from VW.
The overall idea is to create an fx transition from the old T1 bus to new ID. Buzz. Here is our previs.
Based on the previs, we found reference about fracturing, pyro dust, and growing plants.
And we also got all the bus models, we are ready to work! The old T1 bus comes from 3d66.com and the new ID Buzz comes from CGTrader.com.
Week2
In this week, I found a specific reference for our transition effect. I would use particles to simulate the peeling fx.
I gave some test on our bus, and it worked fine!
In the next week, I'm going to adjust the fx and add smoke beyond the particles.
Week 3
Based on the advice from my professor and mentor, I decided to use vellum for the transition effect. However, I didn't fully abandon the particle method, I used the particle evolution animation to drive the stopped attribute so I could control the peeling range.
I used SOP level vellum solver, and applied. During the simulation, I found that peeling parts would stop, because the non-evolutive particles would refresh the stopped attribute once they fly across the particle evolution range.
To solve this problem, I add two wrangles node in a SOP solver inside the vellum solver to check the stopped attribute so I could exclude the influence of particles on the peeled parts.
Week4
In this week, I tweaked the peeling FX, applying the peeling to the whole bus.
When I added more models to the peeling node tree, I found that Houdini will crash when applying the edge refracturing node. I don’t know why it happens, but I solved it by separating the models into part and remeshing them before putting into simulation.
I also started to test the Pyro workflow this week and would probably focusing on it next week.
Next week, I will focus on the Pyro.
Week 5
This week i had a lot of progress.
I worked on the particles. The idea is separating polygon of each peel. To achieve that, I use attribute wrangle to get the center of each primitive and get the point cloud.
int mypoints[] = primpoints(0,@primnum);
if(@ptnum != mypoints[0])removepoint(0,@ptnum);
@P = v@C;
Then, I used sort node to match the point ID in the point cloud with the primitive ID. After that, I put the point cloud into pop network and simulate the particle separation animation.
I also add debris particles.
Two thing need noticed.
1. Vellum Post-Pocess node cannot pass string attributes when extruding the thickness for some reason, (maybe it is a bug) you have to dive into the node and transfer the ID of the peel (in my case, it is “piece”) to float before extrude.
2. Debris source node only recognizes “name” attribute to tell the ID of peel.
I also started to work on the Pyro effects. I used SOP Pyro solver, it is much easier than the traditional DOP solver. I used debris particles as my Pyro source.
Next week, I am going to focus on optimizing the pipeline so I can tweak faster.
Week 6
This week I am focusing on the tire and trail smoke. First, I soloed the tires, then I used scatter node to add points for the tires. After that, I rotated the tires and used trail node to get the velocity.
Then I import the points into popnet node and sourcing particles with these points.
Finally, I used SOP Pyro solver to simulated smoke from these particles.
Here's the flipbook of the smoke.
Week 7
This week I mainly worked on tweaking the trail smoke.
To reduce the size of Pyro, I used the vdb clip node to cut all the necessary smoke outside the camera view.
Here is the rendered vdb sequence.
Week 8
On the Monday, mentors from The Mill suggested that I should use deformation for my Pyro source so I could get a more realist tire and trail smoke. I took the advice. However, working with moving source is harder than I expected.
In order to prevent the simulation oversizing, I had to use car's transformation to drive the center of the container. But due to the low angle shot, the scale of the moving contain needs a lot of testing for avoiding a shape edge on the smoke inside the camera view.
For some reason, the tire smoke has low-res artifacts in the first 100 frames, so I ran the simulation 100 frames earlier.
Here is the new smoke flipbook.
Week9
This week, our team has finished rendering and compositing.
Another important task for this week is to create breakdown. For saving more space, I deleted all the cache previously, so I have to cache them again. I set up a TOP network, so I can cache all solver in one sequence.
And here is my fx breakdown.
Week10
So happy to have such a good experience of working with my talent teammates and The Mill's mentors, here is the final version.
Comments