Project 001
Dangerous
Things
What if the gumballs were bullets?
A Houdini 22 production study that escaped the tutorial and became a complete CG pipeline.
Finished Resolve delivery, including final hold and ELE end card.
From exercise to production
The project
Dangerous Things began with SideFX’s Solaris Market exercise as a way to explore Houdini 22’s USD/Solaris workflow. A separate foundations tutorial supplied a useful starting point for generating piles.
Then came the more consequential question: What if the gumballs were bullets?
The experiment grew into a complete solo production: commercial assets were standardised and packaged for reuse, materials rebuilt for Karma, ammunition piles generated through simulation, the shot assembled in Solaris, camera movement driven through a custom rig, and the finished EXRs graded and delivered through Resolve.
Commercial chaos → controlled assets
Source models arrived with different units, orientations, pivots, naming and material structures. Instead of repairing those problems repeatedly inside the shot, each asset passed through a standard preparation stage and was cached as native Houdini geometry before surfacing and USD packaging.
The resulting converter became a reusable ELE production tool rather than a one-project fix.
OBJ / FBX → normalize → .bgeo.sc → MaterialX → USD component
Simulate once. Keep the useful result.
Ammunition piles were generated with rigid-body simulation rather than hand-placement. Once a convincing settled arrangement existed, the selected state was frozen, cached and packaged as a reusable static asset.
Physics created the disorder. The final shot did not have to keep paying for it.
Movement. Aim. Focus. Separate controls.
The camera move production-tested the ELE Camera Rail HDA. Travel along the rail, camera aim and focal distance were controlled independently, allowing the shot to pull away from the weapons while progressively shifting attention toward the Abrams.
The finished move used a sparse Bézier path rather than dense animation controls, keeping the motion smooth and editable.
The expensive thing that got deleted
The first version continued past the Abrams into a large photogrammetry canyon. It rendered slowly, forced the camera to accelerate toward the end of the shot and shifted attention away from the actual subject.
The canyon was cut.
The camera move became shorter and calmer, the Abrams became the final composition, and late-frame rendering became dramatically cheaper.
Production value is not the quantity of visible work.
From shot to pipeline
What the project proved
Dangerous Things left behind more than a finished film. It production-tested an asset-normalisation HDA, a reusable static-pile workflow and the ELE Camera Rail; established a repeatable Solaris/USD asset pipeline; and completed the route from Houdini through Karma EXRs to finished Resolve delivery.
Dark spectacle.
Terrible consequences.