Adobe Stock — AI Dataset Production

Structured photographic datasets produced to specification for computer vision training, plus gallery content for stock licensing.

Role: Photographer & Dataset Producer Period: 2025-2026

​

The brief

Machine learning models need image sets where every frame is technically consistent, not just well composed. The assignment was sequential object photography: realistic scenes containing several overlapping objects, captured step by step as each item is removed, closing with the empty scene and isolated shots of every object in the same order.

​

What I did

• Produced [N] complete object sequences to technical specification, covering [categorías: plantas, cerámica, objetos de mesa].

• Held lighting, camera position and staging registered across every frame of a sequence, so the only variable between shots was the object removed.

• Wrote production briefs for the creative team running additional sets.

• Delivered [N] final images cleared for dataset and gallery use.

•  

Why this work matters

Dataset production is photography under machine constraints: consistency is the deliverable, and a frame that looks good but breaks registration is a failed frame. That discipline carries directly into brand content, where the same rigor keeps a campaign visually coherent across dozens of assets.