Andres — home
Photographed for Portfolio, February 2022

Dior Sauvage — campaign study

This project was the second of two exploring frontier image models and their ability at the time to transfer scene style and lighting typical of multi-image campaigns, while maintaining scale and logo fidelity.

I started only with the image I captured above, and systematically tested how many images, and how much direction, could be transferred between each new shot to eventually generate a story without any additional source imagery.

Origin
Self-initiated method study
Brand
Dior — no commercial relationship. Not commissioned, not endorsed.
Source
My own photography
Started
November 2025
Completed
December 2025
Role
Direction, pipeline, production, post

AI models

LLM
Claude Opus 4.1, Claude Opus 4.5
Image
Nano Banana ProTested: Nano Banana, Seedream 4, Flux Kontext Pro, Qwen Image Plus, Krea 1, ChatGPT Image
Motion
Seedance 2.0Tested: Veo 3 Fast, Seedance 1.5 Pro
Audio
-

Traditional tools

Image
Photoshop, Pixelmator Pro
Motion
DaVinci Resolve

Notes

All motion clips generated in April 2026.

Next projectCasamorati>>