Hold the Room: a fall campaign of 12 real suits, made entirely with AI

Hold the Room is INDOCHINO's Fall/Winter 2026 campaign, and every image in it was made with AI. I art directed the campaign and built the system that produced it: a look card for every outfit, a reference set for every location, a cast of four synthetic models and one locked grade.

Twelve looks across three sets went into the showroom lookbook and the e-commerce landing page.

Where AI stops. AI made the pictures. The cloth, the casting and the checks stayed human: every look started from the real product, and every frame was judged against it before it was graded.

Client: Indochino
Year: 2026
Role: Senior Art Director
Work: Campaign, lookbook, landing page, AI imagery

The idea

Four friends, one hotel, one weekend. They arrive separately through the afternoon, overcoats on against the weather. That evening they meet in the lounge for drinks. On the last night they close the weekend with a formal dinner. The brief asked for tailoring that reads the same in the rain at the door and at the table at midnight, led by the cloth.

Hold the Room banner: the four friends in overcoats against a stone wall

Arrival

Afternoon, the street, the weather. Each man arrives alone.

Arrival: a man in a navy overcoat over a midnight glen check suit on a city street

Arrival: a man in a grey overcoat and rust check suit leaning on a brick wall

Arrival: three of the friends talking in a rain-wet colonnade

Arrival: the four friends from behind in overcoats, walking through the colonnade

The Lounge

Evening. Jackets open, warm lamplight, one expressive check.

The Lounge: a man in a light brown plaid jacket on a leather chesterfield with a glass

The Lounge: two of the friends at a home bar, one drinking

The Lounge: two of the friends in check jackets, side by side in low light

The Lounge: three of the friends on a leather banquette, one foot toward the camera

Black Tie

The last night. Seated, composed, the room dark and warm.

Black Tie: a man in an emerald velvet dinner jacket in a gilded room

Black Tie: a man in a midnight blue double-breasted tuxedo by a fireplace

Black Tie: a black and white frame of a man in velvet and sunglasses raising a glass

Black Tie: a man in a burgundy dinner jacket laughing at the dinner table

Real product, real cloth

Every garment in the campaign is a real INDOCHINO product, so every frame had to carry the actual cloth, colour and construction. Each look starts from the cloth swatch, a technical drawing of the build and a four-angle reference of the model, with the product cutouts, shirt, tie and shoes alongside. Those feed one look lock: four studio views of the model in the full look. Every editorial frame is made from it. The suit stays the same suit. Before a frame shipped, its cloth was checked against the swatch, and frames that drifted were regraded or replaced.

Look 21 in Montaldo houndstooth flannel: three references, one look lock, five editorial frames

A cast that holds

The four friends are synthetic models from INDOCHINO's own roster, not real people. Each identity is locked by a written description and a four-angle head reference, and every prompt for that model carries both. That is what kept each face the same man across three sets and hundreds of frames.

Casting references for the four synthetic models: four head angles each for Eric, Jake, Jun and Greg

Volume without drift

The work moved through one numbered production line: brief, build, renders, one Capture One session, graded, selects, shared. Set 3 was graded first and locked as the colour target, and the other two sets were matched to it, so three locations read as one campaign.

Chart: 1,164 raw renders, 701 graded, 89 selects