AI-made campaigns

Images you could never photograph. Still recognisably yours.

We build campaigns with generative AI for brands in any sector — fashion, manufacturing, furniture, food, services — without the result looking like every other generated image.

Generated image: figure in pale linen wearing a hat on a grassy slope
Generated image: figure seen from behind wearing a hat in a field of daisies
Generated image: wide-brimmed hat and linen shirt against red dunes

Almost every generative campaign has the same flaw

You see the technology before you see the brand. The image works as a demonstration — look what we can do — and within a season it ages, because it looks like everything else coming out of the same tools.

The brand signs it but doesn’t recognise itself in it. That is why so many companies try AI once and then go back: not because the result was ugly, but because it wasn’t theirs.

Our test

The question isn’t “does it look real?” It’s “is it yours?”

It’s the same question we put to any studio shot: does it fall inside the brand’s grammar — palette, framing, distance, gesture — or not. If it doesn’t, we do it again, exactly as we would redo a failed shoot. No image leaves here because it is technically impressive: it leaves because it is recognisable.

Two proofs, two sectors

TirabassoLuxury manufacturing

One hat in four impossible worlds

Tirabasso makes hats for luxury maisons. We wanted to take the hat into four worlds — a green field, a spread of flowers, a desert, an abstract space — that no budget would have produced: four locations, four trips, four different seasons. We built them instead. The grammar, though, is single: the figure small in the field, colour acting as both backdrop and subject, the hat as the only constructed presence.

Generated image: figure in pale linen wearing a hat on a grassy slope
Generated image: figure seated on a white bench among green hills
Generated image: figure seated on the grass wearing a hat, hills behind
Generated image: figure seen from behind wearing a hat in a field of daisies
Generated image: figure seen from behind wearing a hat in a field of red poppies
Generated image from above: figure in red wearing a hat in a sunflower field
Generated image: wide-brimmed hat and linen shirt against red dunes
Generated image: figure in white on top of a dune, clear sky
Generated image: tiny figure reflected in a salt flat before dunes and mountains
Generated image: wide-brimmed straw hat in front of a poppy field
01 / 12
RodoFashion and accessories

A fashion campaign that was never shot

In luxury the risk of gadgetry is at its highest: it takes very little for a generated image to slip into special effect. We treated the model as a set — same iconography, same palette, same withheld gesture as the photographed campaigns — and kept only what could sit next to a real shot without jarring.

Generated image: aqua green slingback resting on a lily pad
Generated image: fuchsia crystal pumps on a rock in a woodland stream
Generated image: hand holding a fuchsia woven clutch in a flowering glade
Generated image: hand placing a white top-handle bag on a moss-covered boulder
01 / 10

What actually changes

  1. 01

    What was out of reach becomes possible

    This isn’t about saving money on a shoot: it’s about producing images a shoot cannot give you. A location that doesn’t exist, the wrong season, a product not yet in production, a whole world built around a small object.

  2. 02

    The budget decides how much, not whether

    A campaign that would require travel, permits, casting and an on-site crew lands in a different range. The budget stays the starting fact, stated up front — it sets the scope of the project, never the quality of what we deliver.

  3. 03

    Variations stop being a problem

    Different formats, different markets, a new season on the same visual system: once the grammar is fixed, extending the campaign costs a fraction of the first round.

Black hat pressed onto a block, the worker’s hands on the machine

What AI does not replace

Photography. In the Tirabasso project the factory is documented in black and white, with a camera: the place, the machines, the hands shaping the straw. That register proves the craft, and no generated image can take its place. The constructed campaigns do something else: they create desire. The two halves hold each other up — the proof makes the desire credible, the desire makes the proof interesting.

Which sectors

The method doesn’t change with the product: what changes is the grammar to respect. We have worked with fashion brands, craft producers, manufacturers, hospitality, wood and furniture, and the non-profit sector.

  • Fashion and accessories

  • Manufacturing and industry

  • Furniture, wood and design

  • Food, wine and hospitality

  • Craft and small-batch production

  • Services, professions and non-profit

How it works

  1. 01

    The concept, before the images

    We start from a written idea and a deck: what the campaign has to say, in which visual grammar, on which channels. It is the part our clients point to as the one they don’t find elsewhere.

  2. 02

    Building and selecting

    We generate a lot and discard almost all of it. What survives is corrected by hand, in post-production, until it falls back inside the grammar set at the start.

  3. 03

    Delivery and variations

    We deliver the campaign in the formats that are actually needed — print, social, web, retail — and remain the single point of contact for everything that follows.

Start with a question of your own

Tell us about an image you need and can’t produce. We’ll tell you whether AI can do it, how we would set it up and what it costs — even when the answer is that you’re better off photographing it.

Email us

info@studiokubla.com