Skip to content

Guide

How to visualise a house on an empty plot

From a photograph of bare ground to a street view of the house designed for it — and from your architect’s plan, the rooms inside. What to photograph, what to measure, and why the numbers matter more than the style.

Last updated 31 July 2026

How do you visualise a house on an empty plot of land?

Photograph the plot from the road, measure the area you are allowed to build on, and give both to the model. From that it produces the house seen from where you stood. The rooms come from a second input, and one you already have to supply: the floor plan your architect drew.

The distinction from every other operation in this product is worth stating up front. Everywhere else the architecture is fixed and only the contents change — that is the promise that makes a listing photograph safe to publish. On bare ground there is no architecture to hold, so what is held instead is the site: the boundary, the levels, the trees, the neighbouring houses and the viewpoint you shot from.

The photograph

  • From the road or the pavement, at standing height, camera level.
  • Both side boundaries in frame. They are what the buildable area is measured against.
  • A neighbouring house or two, if there are any. These matter more than they look — they are the evidence that the design was put on this street rather than on a generic one, and they are the first thing to check in the result.
  • Overcast, even light. Hard shadows across a sloping site read as level changes that are not there.

The numbers, which matter more than the style

A photograph carries no scale. Ask a model to design a house on a plot without telling it how big the house may be and it will draw one that fills the frame handsomely, at a different size every time you run it — which makes the output a mood board rather than a proposal, and makes two runs of the same plot impossible to compare.

So Plot to Exterior asks for a buildable width and depth in metres, and the house it designs is pinned to them. Give it the same pair on every run of the same plot and the results are comparable; change the style and only the style changes.

Use your planning envelope, not the plot boundary. The number that governs the design is what you may build within, which is nearly always smaller than what you own.

The rooms come from your plan, not from ours

There was a third operation here until 31 July 2026 — one that drew a measured floor plan for the plot. It has been withdrawn, and the reason is worth stating rather than quietly patching over.

A floor plan is a technical drawing. It is bought so that it can be measured off, and an image model draws something that resembles one — correct double-line walls, convincing door swings, a scale bar — over dimensions that do not hold. That is worse than no plan, because it looks checkable. It could not be prompted away, so the operation went.

Plan to Interior stays, and is unchanged. Upload the plan your architect drew and it renders one room of it at standing height. Run it once per room you want to show: every room comes from the same drawing, which is what makes them the same house — a kitchen generated from a photograph of the living room would be a different room in a different building.

What to reject

Three failure modes, all of them ways of answering a different question:

  • A moved camera. If the viewpoint shifted to the flattering three-quarter angle every architectural render uses, the image is of somewhere else — and a client who has stood on the plot will know immediately.
  • A tidied site. Widened road, flattened slope, deleted neighbours. The constraints exist to prevent this; an output that shows it is one to throw away rather than to publish.
  • A house sized to the frame rather than the envelope. Compare its width against the boundary you photographed. This is the failure that matters most, because the envelope is usually the planning constraint.

All three are in the same credit as a good result, so rejecting one costs nothing but the run.

Read next