Understanding the challenge
Today, a lot of what we experience feels more like **Instagram or TikTok** —
fast, visual, and full of movement.
We’re used to movement.
To faces that blink.
To stories that feel alive.
But when we walk into a museum,
and stand in front of historical portraits —
it feels like that flow suddenly stops.
The images are static.
Sometimes damaged.
And for a modern audience,
The kind of emotional engagement we’re used to today
simply isn’t there.
A damaged historical portrait from the Museum collection
What I Bring to Product Design
So the question became:
How might we design AI-animated historical portraits
that increase emotional engagement —
without compromising historical authenticity?
A damaged historical portrait from the Museum collection
Understanding the challenge
I didn’t start by building a solution.
I started by designing a way to test one.
To explore this, I designed a structured experiment that allowed
direct comparison between different versions of the same historical portraits.
I used an A/B testing framework, conducted in two phases —
first in a controlled online setting, and then in a real museum environment.
To answer this, I looked at two sides: What research says, and how people actually experience it.
Understanding the challenge
I didn’t start by building a solution.
I started by designing a way to test one.
To explore this, I designed a structured experiment that allowed
direct comparison between different versions of the same historical portraits.
Understanding the challenge
I didn’t start by building a solution.
I started by designing a way to test one.
To explore this, I designed a structured experiment that allowed
direct comparison between different versions of the same historical portraits.
I used an A/B testing framework, conducted in two phases —
first in a controlled online setting, and then in a real museum environment.
Multi-Variation Design
I created nine variations of two portraits
by systematically adding animation, color, and narration.
Controlled Comparison
Each version was tested under consistent conditions
to allow direct, side-by-side comparison.
Two-Phase Testing
The study was conducted both online and in a real museum environment.
Mixed-Method Analysis
I combined quantitative data with open responses
to understand both behavior and perception.
Before launching the full study, I conducted a pilot session to refine the questions, structure, and timing.
The study was also designed in alignment with research ethics (IRB).
What I Bring to Product Design
Each step added a new layer of transformation,
allowing me to systematically build different variations for testing.
What I Bring to Product Design
Section 1 — Base Transformations
Original archival portrait (static)
Upscaled portrait
Colorized portrait
Animated portrait (10s, First-person narration)
Animated portrait (10s, Third-person narration)
Animated portrait (60s, First-person narration)
Animated portrait (60s, Third-person narration)
Animated portrait (60s, Hybrid narration)
What I Bring to Product Design
B BBNN. Setup
6-week behavioral experiment
30 Walmart shoppers
Weekly grocery NNKMK collected
Diet-specific recipes matched to deals
Meal suggestions delivered via email
How We Tested
What I Learned
75%
Over 80% of participants cooked at least one of the suggested meals —
a strong signal for a test without a digital product
Behavioral Overlap
Demographic Breakdown
Budget-conscious and health-aware consumers span multiple age groups, not a single demographic.