CHALLENGE
Cloudscape needed to feel atmospheric, expansive and unmistakably connected to the British Airways visual world. However, the first AI explorations produced inconsistent results. When we introduced movement around the performers, the clouds often behaved like smoke, fog or visual effects rather than natural formations in the sky, and the quality of the AI-generated imagery was inconsistent.

My role was to elevate both the visual aesthetic and the quality of the experience, creating and estabilishing a more inspiring creative direction for the AI-generated imagery and motion, improving the quality of the outputs while building a clear framework that could produce consistent Cloudscape across different artists, formats and campaign touchpoints.

.

The goal wasn't to tell AI exactly what to create, it was to define the creative system it could create within. By establishing a clear creative vision, visual principles and flexible guardrails, I enabled AI to generate original Cloudscape compositions that remained natural, atmospheric and consistently aligned with the creative direction.

.

Before exploring generative workflows, I art directed a photoshoot with BRIT School performers, capturing them as graphic silhouettes. These photographs became the creative foundation of every Cloudscape composition, providing authentic human movement and form for the AI to build around. Rather than generating people, the AI generated the atmosphere, transforming real performances into immersive Cloudscape environments while preserving the human authenticity at the heart of the identity. 

.

As the visual development progressed, I realised the AI was consistently generating clouds that looked more like smoke around the talent's head than natural skies. Rather than pushing forward with those outputs, I stepped back and rethought the prompt strategy, focusing first on generating cloud formations independently of the talent. Once the cloud language had been established through a four-frame exploration of shape, softness, light and movement.

.

To scale Cloudscape, I developed a structured AI workflow in Figma Weave that unified creative direction, prompt architecture, AI generation and continuous iteration into a collaborative production system. The modular workflow enabled rapid exploration while maintaining a consistent visual language across stills and motion.

.

AI generations were treated as creative source material rather than finished assets. Selected images were refined in Photoshop to correct cloud forms, improve integration, control colour and remove visual artefacts. Motion outputs were then developed further in After Effects, where timing, masking, grading and transitions could be controlled with greater precision. This hybrid process combined the speed and variation of generative AI with the judgement and craft of traditional post-production.

.