RYUJIN
56

The tension

New yachts are sold long before launch. On 18 September 2026 one broker listed 13 new builds for delivery between 2026 and 2028. Until delivery, the images are the boat. My constraint: the AI may change the light and the sea around the yacht, never the yacht. A yacht designer spots an invented space at once, and then nothing in the image is believed.

The brief

A buyer should see RYUJIN 56 at sea, under a low sun, while the yacht exists only as a file. It is my own concept: 56 metres and 12 guests, the boat of my thesis, refined for five years. It lives as my 3D model and a set of renders from it. The brief I set myself: turn that model into a campaign a broker could use in a listing, without changing the boat.

The idea

I show the yacht at sea before it exists, and keep every line of my model.

Brand logic

KLAUDEYACHT makes images for the people who sell yachts before they are in the water: brokers and yards. RYUJIN 56 is its flagship, the boat meant to carry everything I know. I have refined it longer than any other, so a wrong line in an image shows at once. That makes it the fairest test of the method.

Film

There is no film yet. In a campaign it is the last step, cut from the same model. For RYUJIN 56 it will keep the light and sea of the twelve stills, and this block will carry it.

Spatial expression

Inside, the main salon: ivory bouclé and polished walnut, with one rust-velvet accent. On the arrangement I approved, the owner suite sits forward on the upper deck and the VIP cabin on the main deck. Aft, on the lower deck, a beach club with spa and cinema.

AI production system

The first batch failed. I had given the model earlier AI scenes of the boat as references, plus a long prompt. All ten frames came back with a boat that was not mine.

So I went back to the source. Each frame now starts from two references: my render of that exact view, and one mood image for light and sea. The prompt asks for the boat in the reference, with its hull and bow proportions kept. That batch held the hull.

AI took over the part that used to need a shoot at sea. What stayed mine: the 3D model with its rendered views, and every choice. On 23 September 2026 the model produced 42 campaign frames, and I kept 12. I checked each kept frame by eye against the model, proportions first. It also added national flags that are not in my 3D model. They were removed on the pixels, with no new generation.

The method scales: any yacht with a 3D model and its rendered views goes through the same two references.

Measure	Value
Campaign frames generated, 23 September 2026	42
Frames kept	12
References per kept frame	2
Days from the renders to the kept set	1

Real-world execution plan

RYUJIN 56 is a concept study, Revision C of 23 September 2026. I approved the concept and the arrangement; two figures come from the brief. Every other value is a proposal and keeps its status. Naval architecture stays with a third party: hull lines and stability are open engineering work. For the images, one person needs a render per view.

Item	Value / proposal	Status
Length overall	56.00 m	Brief
Guests	12 in 6 cabins: owner upper + VIP main + 4 doubles lower	Assumed
Maximum beam	9.80 m	Assumed
Gross tonnage	750–850 GT; approximately 800 GT target	Estimated
Maximum speed	15–16 kn, target	Assumed
Range	4,000 nm at 12 kn, target requiring verification	Assumed
Tender	Approximately 6 m in forward main-deck garage; side launch with lifting system	Assumed

Guests: the number 12 is Brief; the split into six cabins is Assumed.

Result

Self-initiated, with no client and no sale. What it proves: a yacht that exists only as my 3D model can carry a campaign in one day. It stays the same boat in every frame. Every figure carries its status, so nobody mistakes a target for a fact. Next: the film, then the same method on a yacht in a live listing.