company
type
role
team
-
Ai Visualization
Ai Expert
-
JAN 2026
Ai, Identity and Fashion
Fashion serves as a potent, non-verbal language through which identity is both constructed and communicated to the world. It acts as a dynamic interface between the internal self and external reality, allowing individuals to signal their values, cultural affiliations, and aspirations without uttering a single word. While often dismissed as superficial, sartorial choices are deeply psychological; they navigate the delicate tension between the desire for distinct individuality and the human need for communal belonging, enabling us to align with specific social tribes or rebel against them. Ultimately, the relationship is fluid rather than static; as our personal narratives and self-conceptions evolve, our style shifts in tandem, making fashion not just a covering for the body, but a continuous act of self-discovery and reinvention. Marble is an AI-driven fashion campaign that explores the intersection of technology, art, and identity. Built on a complex creative pipeline, Marble blends multiple AI systems to craft cohesive and visually compelling narratives that maintain consistency in both aesthetic and character design. From conceptual generation to final visual output, the project pushes the boundaries of automated creativity — redefining how fashion storytelling can be imagined, designed, and experienced through artificial intelligence.
Character Consistency
One of the core challenges in Marble was maintaining a strong sense of character consistency across all visuals. The campaign began by designing the ideal character using Midjourney, focusing on defining their unique aesthetic, mood, and personality. Once the visual identity was established, I developed a custom pipeline combining Nanno Banana, Seedream, and Higgsfield Angle to recreate the same character from multiple perspectives and lighting conditions.
Context
After establishing a consistent character design, the next step in Marble was to place the model within a fitting context — both in terms of fashion and environment. Using Nanno Banana, Seedream, and Higgsfield Popcorn, I dressed the character in campaign-appropriate outfits and positioned them within scenes that aligned with the overall aesthetic and narrative direction of the project.
Editorial Detailing & Post-Production
The next stage of Marble focused on editorial refinement and visual polish. After generating the core imagery, I used Adobe Photoshop and Adobe Firefly to fine-tune each picture — performing detailed retouching, color correction, and stylistic enhancements. This process ensured that every visual met professional editorial standards while preserving the artistic essence of the campaign.
Video Generation & Motion Design
To extend Marble into motion, I used VO3 as the image-to-video model, transforming static visuals into dynamic moving scenes. By applying JSON prompting, I was able to achieve a higher level of control and detail — refining gestures, camera movement, and atmosphere to ensure each video remained true to the campaign’s aesthetic identity.
After generation, I brought the clips into Adobe After Effects for final montage and post-processing. This included adjusting playback speed, adding motion effects, and introducing subtle noise and texture overlays to create a more cinematic and realistic visual feel. The result was a seamless fusion of AI-driven motion and editorial craftsmanship that captured the essence of Marble’s visual narrative.