Start with language
Ask for a place, structure, vehicle, landscape, or original idea. Useful prompts describe the identity, important features, scale, materials, and relationship to the site.
Objects gives AI a real site, an actual catalog of parts and materials, and tools to inspect its own work. The result is a brick model you can change—not a flat image pretending to be one.
Ask for a place, structure, vehicle, landscape, or original idea. Useful prompts describe the identity, important features, scale, materials, and relationship to the site.
An optional image can guide silhouette, proportions, colors, and recognizable details. The model can use the image as evidence while still creating a complete three-dimensional interpretation.
The generation is assembled from parts in the Objects catalog. You can continue manually, replace colors, add details, rotate the whole object, or discard the draft before publishing.
Complex models benefit from a process. The AI can plan named components, construct geometric scenes, render several views, inspect bounds and support, and revise before it finishes. This gives the system a way to reason about symmetry, courtyards, arches, towers, roofs, openings, and repeated structure without listing every brick in prose.
A lighter mode is intended for quick experimentation and remains available without premium-generation credits. The premium mode uses a more capable model for ambitious, spatially demanding builds and consumes credits because those generations carry a meaningful model cost. Current availability and credit quantities appear in the app before generation.
A generated result is a draft. You decide whether it represents the idea, whether it fits the place, and whether it should be published. Reference images may be incomplete, prompts may be ambiguous, and models can make mistakes. The builder remains in control of the final object.