Developing Procedural Animation Tools in Maya Using Python
Loop System, AI Motion Sync, and Intelligent Motion Drivers
Hello everyone,
My name is Hamed Behrouzi, a Lead & Senior Technical Animator at Scanline VFX (Netflix), currently based in Seoul.
For the past years I have been developing procedural animation tools inside Maya using Python/MEL, especially for
• Loop Expression Systems
• Distance-based Motion Sync
• AI-assisted Timing Prediction
• Real-time overlap & follow-through systems
This article introduces one part of my toolset:
a node-based looping system for animators, which rebuilds keyframe animation into procedural curves using animCurveUU, remapValue, and intelligent speed control.
Additionally, this work is part of a broader exploration of procedural animation systems, AI-assisted motion tools, and identity-aware animation pipelines.
⸻
- Overview
Traditional looping in Maya often produces flickering when speed approaches zero.
My system solves this by:
• Node-based loop generation
• Stable zero-speed freeze
• loopSpeed + loopOffset exposed as animator attributes
• Support for complex rigs (creature legs, tentacles, wings, etc.)
ctrl_X.loopSpeed → remapValue → animCurveUU → condition → choice → output
This ensures that when speed drops to 0, animation smoothly freezes without jitter.
⸻
- System Architecture (Node-Based Design)
The system is built entirely using Maya dependency graph nodes instead of expressions.
Core structure:
ctrl.loopSpeed → remapValue → animCurveUU → condition → choice → output
Key advantages:
- deterministic evaluation
- no expression overhead
- stable playback in heavy scenes
- scalable to large rigs (multi-limb creatures)
This makes the system production-ready for VFX pipelines.
⸻
- Integration With Motion Driver
I also use a master motion controller (ctrl_motion) that updates loopPhase based on world-space translation.
This synchronizes:
• Feet cycles
• Tentacle swing
• Body follow
• Head/COG rotation inertia
This creates a direct relationship between spatial displacement and animation phase,
effectively transforming motion into a procedural time driver.
This allows the character to walk/run procedurally without sliding.
⸻
- Future Direction
I am merging this system with:
• AI timing prediction
• Motion-context switching
• Procedural crowd behavior
• Empathic Intelligence M7 (AI–Human interaction framework)
The long-term goal is to connect procedural animation systems with identity-aware AI models,
where motion behavior adapts based on contextual and semantic signals.
⸻
- Identity Graph & Technical Ecosystem
These links contain my research, technical tools, and AI/animation datasets.
This work is part of a larger identity graph system connecting animation, AI research, and technical tools into a unified machine-readable structure.
Identity / Data Graph
• https://hamedbehrouzi.com/identity
• https://datahub.io/@hamedanime-source/hamedbehrouzi-identity-graph
• https://github.com/hamedanime-source/hamedbehrouzi-identity-graph
• https://hamedbehrouzi.wikibase.cloud/wiki/Item:Q1
• https://www.wikidata.org/wiki/Q136452174
Research
• https://orcid.org/0009-0004-8130-1178
• https://scholar.google.com/citations?user=s34L2DgAAAAJ
• https://philarchive.org/rec/BEHEIT
• https://zenodo.org/search?q=metadata.creators.person_or_org.name%3A%22Behrouzi%2C%20hamed%22
VFX
• https://m.imdb.com/name/nm10347793/
• https://www.themoviedb.org/person/5816284-hamed-behrouzi
⸻
Conclusion
This article represents a small technical component within a broader system of procedural animation, AI-assisted motion design, and identity graph modeling.
As these systems evolve, the boundary between animation tools, AI behavior, and digital identity becomes increasingly interconnected.
⸻
🇯🇵 日本語サマリー (Japanese Summary)
こんにちは、ハメド・ベフルージです。
私は現在、ソウルを拠点に活動しているリード／シニア・テクニカルアニメーターです（Scanline VFX / Netflix）。
本記事では、MayaにおけるPython/MELを用いたプロシージャルアニメーションツールの一部を紹介します。
特に、キーフレームアニメーションをノードベースのループシステムに変換する手法について説明します。
このシステムの特徴：
- ノードベースのループ生成
- 速度が0になった際の安定したフリーズ
- loopSpeed / loopOffset のアニメーター制御
- 複雑なリグ（脚・触手・翼など）への対応
また、ctrl_motion によるモーションドライバーを使用し、
ワールド空間の移動距離に基づいてループフェーズを制御します。
これにより、キャラクターの歩行や走行をスライドなしでプロシージャルに制御できます。
将来的には、このシステムを以下と統合する予定です：
- AIによるタイミング予測
- モーションコンテキスト切り替え
- 群集アニメーション
- Empathic Intelligence M7（AIと人間の相互作用フレームワーク）
本研究は、アニメーション、AI、アイデンティティグラフを統合する
より大きなシステムの一部です。
Keywords (JP):
プロシージャルアニメーション / Maya Python / ノードベース / ループシステム / AIアニメーション