according to Meta The company's official press release said that it has developed a software called "SceneScript"Visual Model, which claims to be able to use a programmable language to quickly "build" scenes, infer room geometry in real time, and convert related data into architectural approximations.
Image source: Meta company official press release
Meta claims that the method can efficiently and lightly build indoor 3D models.It claims that "only a few KB of memory are needed to generate clear and complete geometric shapes", and the related shape data is "interpretable" and users can easily read and edit these data representations.
Developers borrowed the "word prediction" method of large language models to develop SceneScript. Take the Llama model as an example. The model can predict the next word of a sentence based on the previous words. For example, if the input sentence is "The cat sat on the...", the model will predict that the next word may be "mat" or "floor". SceneScript uses the same concept.That is, the subsequent content is derived from the previous input content, and these architectural descriptions are used to reconstruct the complex indoor 3D environment.