top of page
AI Identity & the Stylization Buffer
2024 - 2025
OVERVIEW
OVERVIEW
This explores the design methodologies to establish trust and ensure emotional safety when Generative AI systems interact with user identity.
My work demonstrates that managing the output's aesthetic distance and cultural fidelity is crucial for user acceptance and emotional engagement.
AI Buffer Framework
This demonstrates mastery of three distinct risk-mitigation strategies:
Emotional Safety via Non-Realism
(Gentle High School)
Affective Predictability via
Cultural Fidelity (Tekken)
Zero-Rejection:
The Power of Attribute Swap
(Bratz)
The Maximum Stylization strategy acts as an Aesthetic Honesty Layer, guaranteeing comfort and Vulnerability-Free Projection by rejecting photorealism.
The Strict IP Adherence mechanism ensures the output is trustworthy and predictable, enabling uncritical self-identification even in a high-risk environment.
By implementing a low-risk Attribute Swap (style onto the IP character) rather than a face swap, the project achieved the highest scores in Risk Mitigation and Emotional Safety (100% acceptance/zero rejection).
I. GENTLE HIGH SCHOOL (AI Buffer Validation)
Engineering Emotional Safety via Non-Realism
Strategic Goal: Synthesizing Identity and Trust
This was a crucial experiment in calibrating the Affective Acceptance Threshold for generative AI. It demonstrates how aesthetic ambiguity, combined with tangible outputs, creates a safe space for self-projection and enhances long-term emotional memory.
Challenges:
1.Balancing the level of non-realistic styling in AI-generated portraits to evoke curiosity without causing rejection or discomfort.
2.Generative AI models can default to biases or repetitions if prompts aren't carefully tuned, leading to outputs that converge on stereotypes (e.g., all uniforms looking identical or hair styles pulling from limited datasets).
(a), (b)
(c)
(d)
(e)
(a) Seed Image Preparation: A base human portrait, which would be the input.
(b) A base stylized “face swap” image, which is quite weak for a emotional buffer
(c) Styling Image Outcome (Synthesis and Correction): A specific stylized aesthetic traits (e.g., unique hair, uniform) finalized.
(d) This calibrated process ensured the output created Ambiguity ("looks like me, but also not like me"), which converted potential user rejection into neutral curiosity.
(e) The Physical Pop-Up as Affective Anchor: By transforming the ambiguous digital portrait into a tangible artifact (the printed ID card) on-site, the physical environment served to immediately provide Emotional Ownership and validate the digital experience.
II. TEKKEN (Cultural Fidelity Test)
Establishing Trust via Strict Game IP Adherence
Strategic Goal: Seamless Self-Projection
The Tekken activation tested the AI Buffer by balancing Emotional Safety (rejecting realism for user comfort) with Cultural Authenticity (respecting the game's aesthetic code for fan trust), enabling seamless merger of user identity with pre-existing cultural codes.
Challenges:
1.Maintaining strict IP adherence to Tekken's character designs, styles, and cultural motifs while merging user identities without losing authenticity.
2.Achieving maximum stylization to avoid uncanny valley without risking emotional rejection in non-fans.
Maximum Stylization
Emotional Safety
Intentionally pushed output far from photorealism, serving as the Aesthetic Honesty layer to bypass the Uncanny Valley.
The Safe Transgression
Achieved comfort by violating the rule of realism.
Emotional Safety Proof
User reactions confirmed successful projection:
"I never do this, but let me do this pose"
“The wilder I pose the better results”
“I’m bald but this gives me hair!!”
“I look like Kazuya, not like myself- yes please”
Strict IP Adherence
Cultural Trust
Guaranteed the output was predictable and aesthetically true to the fan culture.
Affective Predictability
The shared cultural context minimized the user's identification effort, enabling joyful, uncritical self-identification.
Authenticity Proof
Users praised the filter detail as being "truly like the game characters", validating:
"This actually looks like Kazuya"
“This is pretty legit”
“Everyone is Kazuya coded lol”
“Usually I get icks from game ai but this is cool”
Hybrid Design
Affective Anchor
The tangible artifact (CD pack) solidified the digital experience.
Long-term Value
This compensation mechanism made the interaction worthwhile, despite waiting times.
Value Proof
Despite long waits for the CD pack printing, high engagement was evidenced:
"This is so worth it "
“CD pack makes this so special”
“I love how this is actually physical”
“I had to try this as a true Tekken fan”
III. BRATS (Attribute Swap Strategy)
Reducing Risk by Altering Style not Identity
Strategic Goal: Safe Aesthetic Projection
The Bratz activation tested the AI Buffer by balancing Emotional Safety (avoiding identity risks for user comfort) with Cultural Authenticity (respecting the Bratz IP for brand trust). This enable the seamless merger of user style with pre-existing aesthetic codes.
Challenges:
1.Maintaining strict IP adherence to Bratz's character designs, signature features, and cultural motifs while merging user styles without losing authenticity.
2.Achieving maximum stylization to avoid uncanny valley without risking emotional rejection in diverse users.
Action
The AI identified the user's outfit and pose. It then generated the Bratz doll character to appear as if they were wearing similar outfits and standing in the same context as the user.
Key Distinction
The system kept the Bratz doll character intact (IP identity) and only swapped the style/outfit attribute based on the user's attire.
bottom of page