Generative AI is changing how quickly creative ideas can move from concept to something tangible enough to evaluate—or use.
This work explores AI across both commissioned production and independent creative studies: using generative workflows to solve specific production needs, rapidly prototype ideas, visualize directions, experiment with camera and motion, and identify what works and what doesn’t.
The technology accelerates the process. Creative judgment still determines what moves forward.
MANDATE AI — Creative Workflow SaaS Platform
AI-Directed Product Launch Concept
MANDATE AI is a fictional end-to-end workflow platform designed for creative teams. The concept grew from a persistent production problem: scattered feedback, conflicting direction and revisions beginning before the people responsible for the work are aligned. Mandate turns that friction into a structured decision-making system—creating one clear direction before execution continues.
VISUAL DEVELOPMENT
The visual system was designed to feel cinematic, intelligent and built specifically for creative teams—not like a conventional enterprise platform. Deep slate blue, cool gray and bronze establish the core palette, while open-corner geometry, layered interfaces and controlled typography give the product a precise but expressive visual identity.
PRODUCT DEVELOPMENT
The product is built around one central rule: when conflicting direction is detected, work does not continue until the right decision-makers resolve it. Mandate collects and groups feedback, identifies agreement and contradiction, brings the necessary people into alignment, captures one authoritative decision and releases a unified revision brief for the team to execute.
MOTION DEVELOPMENT
The launch film is organized around three stages: FIND YOUR NORTH, MOVE AS ONE and MAKE IT OFFICIAL. Each section uses a different motion language—rapid camera moves, dimensional interface transitions, kinetic typography and feature-driven animation—to move the viewer through the product experience while maintaining one cohesive visual system.
EDIT & FINISHING
The interface sequences, generated character material and motion graphics were assembled in DaVinci Resolve Studio. Editorial pacing, sound design, music, color and typography were shaped to make the film feel like a complete product launch rather than a collection of disconnected AI-generated assets.
WORKFLOW
ChatGPT Astra — product concept, narrative development and production orchestration · Blender — 3D staging, lighting and camera animation · Higgsfield — generative imagery and motion development · Suno — AI-generated music · DaVinci Resolve Studio — edit, color, sound and final finishing
NOIR — Eau de Parfum
AI-Directed Fragrance Concept
NOIR is a fictional luxury fragrance concept created to explore how traditional creative direction can translate across an AI-assisted production workflow.
The visual language draws from classic film noir and Art Deco glamour—deep charcoal and blue-steel tones, hard directional light, architectural shadow and restrained champagne accents. The goal was to build a cohesive world across character, environment and product while using different production methods without allowing the technology to dictate the creative.
VISUAL DEVELOPMENT
Before production, I established a visual system to keep every element of the spot in the same world. Deep charcoal and desaturated blue-steel formed the foundation, with silver-neutral highlights and muted champagne reserved for selective accents. Art Deco geometry, hard directional light and deep architectural shadow defined the environment.
CHARACTER DEVELOPMENT
I developed the NOIR Dame as an original recurring character rather than generating each appearance independently. Her silhouette, wardrobe, hair, jewelry and props were established up front to create continuity across shots while giving the character a recognizable identity within the larger visual system.
3D PRODUCT DEVELOPMENT
The NOIR bottle was built as a controlled 3D asset rather than generated shot by shot, giving me precise control over materials, lighting, composition and camera movement. Black glass, brushed champagne details and a restrained 85mm push established a product language that could carry consistently into the finished spot.
EDIT & FINISHING
Generated and 3D assets were brought into DaVinci Resolve for editorial, pacing, color continuity and final finishing. The goal was to make material created through different production methods feel like a single photographic world rather than a collection of generated shots.
WORKFLOW
ChatGPT Astra — production orchestration · Blender — 3D product development · Higgsfield — generative imagery & motion · Suno — original music · DaVinci Resolve Studio — edit, color & finishing
AI-Assisted Show Open
Commissioned Work
Created as part of a late-’80s-inspired show open for a corporate annual conference. The brief called for an unmistakable homage to the era’s iconic time-travel imagery.
I used AI-assisted image and motion workflows to develop and execute the sequence for integration into the larger show open.
WORKFLOW
Nano Banana · Google Flow / Omni — generative imagery & motion · DaVinci Resolve Studio — edit, color & finishing
BOSE — NOISE CANCELLING HEADPHONES "Quiet the Noise"
AI Motion Concept | Rapid Prototyping
THE IDEA
Explore a simple product truth: in a city full of competing noise, putting on noise-cancelling headphones can completely change the experience around you.
The concept uses sound as the primary storytelling device—moving from the sensory overload of a subway platform to the isolation of a single, calming sound: rain.
THE PROCESS
Built as a one-shot prototype using CenterPunch.io, from initial concept to working :15 in roughly 15 minutes.
I started by developing a character reference to establish the headphones, wardrobe, proportions, and overall visual language, then generated and assembled the sequence.
After multiple outputs that were, frankly, comically terrible, I had something coherent enough to evaluate the idea rather than the technology.
WHAT"S STILL NOT PERFECT
This is a prototype, and a couple of artifacts remain.
Environmental typography. Some subway signage contains wonderfully nonsensical AI-generated typography. It breaks the realism, but not the idea.
Sound design. Once the headphones activate, the city disappears and rain becomes the isolated sound. Footsteps remain audible, slightly undermining the simplicity of the transition.
Neither issue prevents the prototype from answering the question it was built to answer: Does the idea work?
THE PROTOTYPE
The goal wasn't a finished Bose commercial. It was to take an idea, visualize it quickly, and get it far enough to determine whether it was worth developing further.
Idea → Prototype → Evaluate.
WORKFLOW
Chat GPT Images · Centerpunch.io — generative imagery & motion
BMW — AUTOMOTIVE MOTION STUDY
AI Motion Concept | Camera + Movement Exploration
THE TEST
How close can generative motion get to the controlled, intentional camera movement you'd expect to see in an automotive commercial?
The study focused specifically on camera behavior, vehicle movement, scale, and the relationship between the car and its environment.
THE PROCESS
I approached the piece as a camera study rather than an attempt to create a finished BMW commercial.
The goal was to test different automotive-style shots and movements, evaluate where the technology held together, and see how convincingly the vehicle maintained its scale and presence within the environment.
As with most generative experimentation, some outputs were hilariously bad. Others worked considerably better than I expected.
WHAT WORKED
The camera movement came surprisingly close to the visual language of traditional automotive advertising on some of the outputs. Vehicle scale also held together well across several shots, allowing the car to feel grounded within the environment rather than disconnected from it.
WHERE IT BREAKS
This is still very much a motion study. There are missing and inaccurate vehicle details, inconsistencies between shots, and plenty that wouldn't survive scrutiny from BMW—or anyone familiar with the car.
But those limitations weren't the question I was trying to answer.
THE MOTION STUDY
The objective was to see how effectively generative motion could approximate automotive camera language, movement, and scale. As an experiment, it got much closer than I expected.
Test → Generate → Evaluate.
NOTABLE MISTAKES
Flames from the exhaust — Unless BMW is running on Jet-A, this was ridiculous.
Scale — In one output, the car became significantly larger than the environment around it.
Side-mirror shifter — Somehow, the model added what appears to be a gear shifter to the side mirror. I have absolutely no idea where this came from.
WORKFLOW
Nano Banana · Google Flow / Omni — generative imagery & motion · DaVinci Resolve Studio — edit, color & finishing
Independent creative studies. Not affiliated with, commissioned by, or endorsed by the featured brands, unless specified otherwise.