Originally prototyped through early restructuring of text-based work with language models, Perrotgative’s internal adaptation suite assists in transforming source material into several multimedia production formats.
The process began with primitive prompt-centric methodology: manually segmenting prose manuscripts into episodic structures for television animation, speech synthesis, and mixed-format distribution across traditional and vertical platforms.
What began as heavily chained prompts and manual supervision of early language models has since expanded into a broader metadata and formatting system, including:
- title, description, and SEO keyword generation
- episodic segmentation and serialization guidance
- excerpt extraction
- adaptive promotional copy
- platform-specific formatting insight
- tonal consistency across source material, house style, or predefined project voices
This system continues active development as a supervised transformation layer between creative formats, emphasizing iterative refinement over raw automated output.
Canz was a browser-based avatar system developed in 2018, distributed through "make-and-play" platforms online.
The project explored non-anthropocentric approaches to self-representation by replacing human forms with everyday objects—primarily beverage cans.
Identity was constructed through surface treatments, accessories, and contextual cues rather than facial features or body types.
Elements such as wraps, eyewear, and attached devices functioned as proxies for personality, allowing expression without reference to age, gender, or race.
The visual language drew from novelty desk objects and early web aesthetics, emphasizing simple geometry with layered customization.
Additional object classes—including bottles, batteries, and film containers—expanded the system beyond a single base form.
The result operated less as a character creator and more as a modular identity system, where meaning emerged through arrangement rather than likeness.
A customized fork of GoAnimate's “Comedy World” animation framework, expanding the scope and fidelity of its character and asset systems.
The work focuses on extending the native asset pipeline while maintaining visual and structural indistinguishability from the original presets.
New components are designed to integrate at parity with existing elements, avoiding the stylistic fragmentation common in community-generated modifications.
The objective is not to introduce visibly “custom” assets, but to increase expressive range within the constraints of the original system, preserving its internal logic while quietly broadening its capabilities.
Contributions focused on musical arrangement, visual development, and the reverse-engineering of proprietary audio formats into editable stems compatible with conventional DAW workflows.
This enabled increasingly precise manipulation of source material beyond typical remix limitations.
The platform operated as a hybrid between content channel and evolving meta-narrative, where uploads were continuously disrupted and recontextualized by external media influences.
Involvement concluded upon entering corporate environments with active non-compete constraints.
A multinational traveling installation designed to translate the Dr. Seuss visual and narrative language into physical environments.
Work within the art department focused on set dressing, spatial continuity, and the practical realization of stylized environments under touring constraints.
This included maintaining visual coherence across modular installations and adapting two-dimensional design language into three-dimensional space.
Additional responsibilities included live puppetry and voice performance for principal characters, requiring alignment between physical manipulation, vocal delivery, and established brand characterization.
A serialized adaptation pipeline developed for a toy company's contemporary product-linked narrative release (unreleased).
The production process integrated generative systems for rapid transformation of prose manuscripts into structured scripts, followed by text-to-speech voice synthesis and stylized animation workflows.
Outputs were manually composited to maintain consistency across generated elements.
The distribution model was designed for dual-format delivery, targeting both traditional streaming platforms and vertically oriented short-form environments.
The project functioned as an early test case for compressed production timelines using hybrid generative and manual pipelines.