personal projects 2023-2024

As a starter for my submission is my latest reel. It includes most of the work shown below and some project I did last year. 

The helicopter crash was my first master thesis project. By doing this project I wanted to learn more about RBD and the matrix functions for it. I had made the house before, so I could bring it out again. I downloaded the helicopter and other props from the internet to make it more complete.

The matrix technique I talked about earlier was intended to be able to remove the animation from your model, then turn it into a proxy and add the animation back later. You can then very easily switch between animation and active to control exactly when the simulation should start.

I had animated my helicopter in Houdini so I could use the xform attribute that is created in your transform node to remove the animation from my mesh. By first milling the mesh on 1 frame, then saying in a wrangle: matrix m = detail(0,”xform”);

Then you must multiply the position @P by the invert of the matrix, so invert(m); Now your mesh is back in the same place as it was before the animation.

To then add the animation back to your proxy mesh later. Multiply your object by the same matrix m. Now all objects are moving in the right direction, but the orientation of the objects is not correct. To fix that, use the setprimintrisic attribute.

The house was fractured and reattached with glue constraints because it simply had to be destroyed by a collision impact. It was a different story for the helicopter because it is partly made of metal, so it can tear and be crushed. I have done several fractures for this.

I started with the fracture for possible tearing of the metal or propeller. The constraints for this were glue constraints but if the glue broke it would switch to hard constraints. In caps I had a sop solver that only looked at the hard constraints to see whether they passed a certain threshold (in terms of how much they bent or wanted to move away from each other). If they reached that threshold, they completely broke the constraints.

For the second fracture I had a voronoi fracture done. These work together with glue constraints that transitioned to soft constraints when broken. This gave you that effect that bends metal.

This was a first slap comp I made, but I didn't like the camera movement. I also wanted to make the shot a bit more cinematic so I ended up looking at a shot from interstaller and tryed to match it with the lighting and feeling of that image. 

Then you have the final shot with some extra props and a new camera. I did the comp and lighting aswell. 

A breakdown to see the full simulation. 

I have also been working on some RBD tests to really understand all the constraints. I made a catapult for this and a trebuchet that actually works through physics.

For the Gundam project I had worked with a classmate who did the animation. We had the idea of ​​having the Gundam come out of the water. Ultimately, this became the IJ next to Central Station in Amsterdam.

For this project I took all the knowledge I had gained about water simulations into account in previous projects. Although I already knew a lot, I also learned new things through this project. I learned how to create a density attribute about the particles you simulate. I have gained more knowledge about how SDFs work and how you can use the gradient. And finally, I learned about how a custom airfield can add a lot of vibrancy to your white water simulation.

playblast of the actual sim in Houdini

The final result of the Gundam. We bought the Gundam model from the internet, A classmate did the animation. I did the FX work and the integration of the CG in the plate we shot at the central station in Amsterdam. 

The Dune explosion shot ended up being my most complicated project from which I also learned the most. I started with the rbd sim for this project. For this I first made a proxy SIM that you can iterate on a lot. You can then use that proxy SIM to guide a high-resolution SIM. The shield was done with a few growth solvers to get the timing right when the shield gives up. The explosion itself took up most of the time. It was two sims. One that grew into the shield and one that was the main explosion outside. After that I added a lot of secondaries such as debris trail and a shockwave.

The rbd sim I started with

A first slapcomp to see if the shield is working as a layer

A second slapcomp to see if the explosion is working in the comp. Everything was renderd as a deep layer so I could simple replace layers, without rendering the whole image again. I would have more freedom in Nuke to try different things. 

the final result of the Dune explosion. I really had some fun make this effect with all the fx layers and combining them in nuke. 

breakdown of all the layers

For the Wilson project, I found the idea of ​​using grains very interesting for white water, so I created my own concept around it. I scanned Wilson myself, because I have one at home. I cleaned it up in Zbrush, redid the UVs and projected the textures onto the new UVs. The environment is mega scans. The water sim was all done with custom forces, which I had taken from a traffic jam from last year where I was also working on beach waves. So the grain sim was all done in the vellum solver with some techniques to make it float on the water and move with the waves. I still had to create some kind of life attribute in caps to sometimes remove the particles. Otherwise, new grains would only be added and then it would become way too much over time.

the final render + the breakdown as a grey render.