I normally publish tutorials after I have a workflow under control. This series captures something different and equally useful: the process of getting there. I work through Puppet3D in Unity, say when I am still learning, refer viewers to the official material for details I cannot yet explain confidently, and keep testing until the rig produces a readable hand animation.

Why I made this

Character rigging combines several disciplines that are easy to blur together: building a skeleton, binding the mesh, painting weights, creating animator-friendly controls, and then using those controls without introducing impossible transforms or broken deformation.

A clean demonstration can hide how much diagnosis and iteration that work requires. These sessions preserve the decisions, mistakes, corrections, and uncertainty that appear while the mental model is still forming.

What I did

The first session establishes the hand skeleton, skinning, and weight-painting foundation. The second moves away from rotating raw bones and builds IK controllers for the fingertips, pole controls for bend direction, and orientation controls for the finger bases.

I repeatedly use a simple loop: create the control, reduce its visual size, test the expected motion, and leave the fine correction for the next pass. When a detail exceeds what I understand, I say so and point viewers toward Puppet3D's official material.

The third session turns the rig into a hand-closing animation. I reduce visual clutter, create reusable selection groups, key the start and end states, and block the middle pose broadly enough to communicate the motion before polishing it.

From there I work from large problems to small ones: timing, crossed fingers, pole direction, a snapping thumb, and skin weights that influence neighboring geometry. The finished motion is still rough, but the series shows the complete learning loop and a working result.

Part 2 chapters

Part 3 chapters

A note on the archive

These videos use Unity and Puppet3D as they existed in 2019. Unity's animation and rigging ecosystem, package workflows, and third-party tooling have changed since then, so the exact interface should be treated as historical.

The first recording also carries an audio-sync warning, and the sessions were explicitly framed as practice. Their value is the visible learning, troubleshooting, and teaching process—not a claim that they represent current rigging best practices.

Published March 30, 2019. Preserved with context. More DevRel work →