Skip to content

Image generation

Image generation turns what you already have - a hand sketch, a viewport capture, a massing model, a flat elevation - into presentation imagery. It works on a node canvas inside a project: each source, reference and result is a node, and every variant you try stays on the canvas.

The building as a working model, captured from the viewport

The principle behind every tool: the geometry you drew is the geometry that comes back. A render is conditioned on your source image, so openings, proportion and structure survive the pass. What you direct is everything around them - materials, light, season, atmosphere, people, objects.

The same building rendered at dusk

What you can do

Tool What it does
Render Turn a source image into a photoreal view, keeping its geometry.
Region edit Point at up to six areas of an image and give each its own instruction - retexture, relight, replace, remove, add - in one pass.
References Borrow a look: up to 13 reference images per pass, each tagged as style, material, lighting, context or content.
Objects Place an object - a sculpture, a bench, a tree - into a scene without repainting the rest.
Reframe Crop or extend the frame of an image.
Relight Change the light of a settled image.
Enhance Take a finished frame to print resolution, up to 4K, with detail added where it belongs.

Before you start

Image generation runs on our machines, so it needs an account and spends credits. Every new account starts with 150 free credits - see Credits and plans.

Next: Sources and rendering, then Editing an image.