top of page
AI DAY @Istanbul Modern

In May 2024, I took an active role in the design and production of AI Day, contributing to both the overall creative direction and the organization of the event. Alongside this, I participated as a speaker, presenting a project titled Lumifin Style, developed within the Havas AI WetLab.

 

The project explored how creativity can be constructed between human and machine. In collaboration with Uğur Sosyal, Development Team Lead at HAVAS CX Istanbul, we developed a system capable of reproducing a defined artistic language through AI.

 

Using a LoRA-based model fine-tuned on my own visual language, we enabled the system to generate new outputs that closely reflect my drawing style. Through continuous optimization, the model achieved a visually sophisticated and consistent aesthetic.

 

Lumifin Style evolved beyond an experimental study into a scalable approach, demonstrating how brands can build their own visual worlds and production systems using generative AI.

 

The project was presented on May 27 at Istanbul Modern, as part of a broader program attended by industry professionals and marketing leaders, featuring sessions focused on the creative and practical applications of artificial intelligence.

Documentation
''/Lumi Fin Style'' Project

Controlled generative system built around a singular artistic identity.

The process began by collecting and structuring a personal archive of drawings spanning nearly 10 years. Hundreds of works were translated into a language the model could interpret, forming a consistent and closed visual dataset.

 

Using a LoRA-based fine-tuning approach, the model was trained exclusively on this dataset. This defined a bounded visual space, allowing the system to operate within a controlled aesthetic rather than producing generic or uncontrolled outputs.

 

The model was then directed to extend existing bodies of work, generating new pieces that remain visually consistent with the original collections. This includes continuing series and producing unseen variations that align with the established style.

 

In a second phase, the system was expanded by introducing layered references such as anatomy, sketch structure, color and stylistic variations. This enabled more complex, concept-driven outputs while maintaining a high level of visual consistency.

 

The result is not simply a stylistic replication, but a scalable framework. A system where an artist’s visual language can be sustained, extended and reproduced across different contexts without losing its integrity.

Visual Identity and Creative Direction

A fully AI driven visual system built across image generation, motion and spatial applications.

The identity is based on a modular jellyfish structure, enabling scalable variations while maintaining visual consistency across all outputs.

bottom of page