Final Destination 5 - Bridge Collapse
Introduction:
The reference video is an excellent showcase of extremely accurate control of suspension bridge destruction since it displayed the whole progress of the bridge structure collapse. This project combined many different techniques, including Vellum, RBD and Crowd simulations. It may have been the most challenging project I have ever worked on because it involved the use of the crowds and advanced rigid body destruction. A unique method was used to create the bridge's swing movement.
Reference
Render Output
Evaluation:
Upon examining the reference, we can observe that the bridge is fractured into two portions. The rear section is driven by a lattice that is distorted at the SOP level by means of sine and cosine waves using VEX, with further twisting motion added subsequently.
Custom Fracture Techniques:
When starting with the road's concrete fracture, one might immediately select the default RBD Material fracture node, which, when adjusted appropriately, occasionally produces a good result. However, for this project, we needed a very specific method of fracturing, so I chose a different approach.
Next, I will talk about clustering. After closely examining the reference, we discovered that there are two kinds of clusters that form when a bridge collapses: large chunks that break apart completely and small clusters that break at random points, resulting in random cracks and breakage. To address these issues, I designed a unique custom clustering HDA that would produce large cluster chunks and used the default RBD Cluster node for the smaller chunks.
Big Clusters
Small Clusters
To strengthen the effect that is demonstrated in the reference, new constraints were created and initialized utilizing the cluster information.
The following step was to break the metal bars beneath the bridge, which required the use of many Boolean approaches since metal bending needs very equalized fracturing as it is utilized to change the original shape.
The sine waves that caused the bridge to sway were masked at the margins so they had no effect there.
The lattice deformer could only alter the pieces' transformation, not their orientation, because the bridge was deformed after the RBD Configure packed all of the fractured pieces. In order to fix this, I had to extract the orient attribute from the lattice geo and transfer it to the packed pieces.
We noticed that the road was breaking into smaller pieces at the edges of big chunks separation after running the simulation's first pass. To address this, I took the original pieces that were at the breakage points and further fractured them into smaller pieces (a process known as RBD upres). I then fed those new pieces into a new bullet solver, setting the new pieces as active and the remaining bridge as animated static objects so that they would collide with all the geometry.