Generating three-dimensional digital content from natural language requests

In a digital medium environment for three-dimensional computer modeling, a system for synthesizing three-dimensional scenes using natural language comprising:

• at least one processor;
• and
• a non-transitory computer memory comprising:
• a database of previously generated three-dimensional scenes;
• and
• instructions that, when executed by the at least one processor, cause the system to:
• identify a natural language phrase comprising a request to generate a three-dimensional scene;
• analyze, using natural language processing, a natural language phrase comprising the request to generate a three-dimensional scene to label strings of characters in the natural language phrase as entities or commands;
• determine a dependency tree comprising one or more dependencies involving a plurality of entities and one or more commands operating over the plurality of entities;
• generate an entity-command representation of the natural language phrase by performing pattern matching on the plurality of entities and the one or more commands operating over the plurality of entities based on the determined one or more dependencies;
• convert the entity-command representation of the natural language phrase to a semantic scene graph including relative positioning of a plurality of objects in a scene based on contextual relationships of the plurality of entities and the one or more commands by:
• generating a plurality of object nodes representing the plurality of entities from the entity-command representation and one or more relationship nodes representing a spatial relationship between the plurality of object nodes;
• and
• generating a plurality of edges connecting the plurality of object nodes and the one or more relationship nodes, wherein the plurality of edges indicate a directionality of the spatial relationship between the plurality of object nodes;
• and
• generate the three-dimensional scene using at least one previously generated three-dimensional scene from the database of previously generated three-dimensional scenes according to the semantic scene graph.

In a digital medium environment for three-dimensional computer modeling, a computer-implemented method of synthesizing three-dimensional scenes using natural language comprising:

• analyzing, using natural language processing, a natural language phrase comprising a request to generate a three-dimensional scene to determine a dependency tree comprising dependencies involving a plurality of entities and one or more commands operating over the plurality of entities of the natural language phrase;
• generating an entity-command representation of the natural language phrase using the determined dependencies between the plurality of entities and the one or more commands operating over the plurality of entities;
• generating a semantic scene graph from the entity-command representation, the semantic scene graph indicating relative positioning of a plurality of objects in a scene based on contextual relationships of the plurality of entities and the one or more commands by:
• generating a plurality of object nodes representing the plurality of entities from the entity-command representation and one or more relationship nodes representing a spatial relationship between the plurality of object nodes;
• and
• generating a plurality of edges connecting the plurality of object nodes and the one or more relationship nodes, wherein the plurality of edges indicate a directionality of the spatial relationship between the plurality of object nodes;
• and
• generating the three-dimensional scene using at least one previously generated three-dimensional scene from a database of previously generated three-dimensional scenes according to the semantic scene graph.