This project uses houdini bullet destruction, and takes advantage of the efficiency of the packed primitive workflow.  The system allows for largescale destruction with a relatively small memory footprint becuase the structures are created with the same set of packed geometry for the base of the structure. When combined with the effiency of a custom smoke solver and rendered with Redshift, I was able to produce the desired result on a standard workstation with a good videocard.  This system will be expanded to include much higher resolution pieces and layers of various materials.  It is already capable of destroying multiple buildings at the same time, and using layering and caching it will be able to destroy a whole city. 

Inspiration 

In 2018 archival footage of nuclear test sites were declassified, providing an terrifying look into the destructive power of the these devices. 

Initial set up and fracturing 

The initial set up was a fairly stardard implementation of the bullet destruction workflow. The building floors were each packed then copied up to form the core building structure.  A 4each was used to create smaller pieces and clusters, which were constained using connect adjacent pieces and voronoi cluster to provide a second layer of glue contraints.   This allowed  the creation of different glue strenghts for the inside pieces compared to the outside pieces. 

Tweaking the simulation

The trickiest part was finding balance between the building collapsing all at once versus some parts not collapsing at all. In the end I had to prebreak the contraints between the clusters, while maintianing the glue between the inside pieces that would decay over time. After running hundreds of iterations I found the right balance which seemed realistic in terms of scale and motion for a large structure getting hit by a massive shockwave which weakened the structure as the momentum continued to slowly topple it over in the direction of the initial force. 

Initial test with smoke 

The smoke pass was attempting to replicate the initial nuetron burst which bakes everything a second prior to the kinetic blast wave hitting. The challenge was to influence the smoke to not only match the timing of the building motion, but also to hit the build with a punch and then slow down as the shockwave passed by. Initial tests (below) were too soft, so I had to keyframe the gradient feild to enormous velocities but for a very short time. After some more research I found the actual blastwave graph and I used that to base my forces on in the graph editor. 

Nuclear shockwave intensity graph

 In futire versions I would like to include the negative vel , which would pull back the debris in the opposite (negative) direction, where the graph dips below the x axis 

​