Moom

What it does

Moom is an AI architecture and interior design kit that puts the focus on people who live in the space. Typically, image generation with generative AI requires a text prompt and it often lacks scale interpretation. In the design workflow, visual communication helps convey concepts effectively and bridges the gap in interpretation. Moom integrates 2D image generation with a 3D model to enhance scale accuracy. To construct the 3D model, we use Gemini API to search our 3D model database that matches user needs and visual input.
The Gemini API extracts significant features from each furniture piece in the database and from the user's moodboard, which consists of a set of images. Currently, we focus on “overall mood and tone” and “color” as the main attributes. These features are then embedded using a text-embedding model. Gemini is used to suggest furniture categories based on users input, such as age range and specific requirements. We retrieve the suggested furniture categories from the database and present results with the highest similarity score. This 3d model will be combined with layout suggestion algorithms and used as the basis for image generation in the next step.