Generative design and optimization based on user-specific terminology for design descriptors

A computer-implemented method for supporting a user in generating or modifying a design of an object, the method comprising

• acquiring, by an input interface, at least one image and textual parameters associated with the image;
• encoding the acquired at least one image by applying a vision-language model based on the textual parameters into vectors including numerical values for the acquired textual parameters;
• generating a parametric representation based on the vectors including numerical values;
• and
• computing, by applying a trained generative model, at least one of at least one image and a data set of a geometric body of the object based on the generated parametric representation.

A system for generating a generative model for supporting a user in generating or modifying a design of an object, the system comprising an input interface, a memory, and

• a processor configured to
• acquire from a database, a plurality of data sets of geometric bodies of objects, for each data set at least one image associated with the data set, wherein each data set is associated with a class of the object,
• acquire, via the input interface, textual parameters related to the objects,
• encode the acquired images by applying a vision-language model based on the textual parameters into vectors including numerical values for the acquired textual parameters,
• generate a parametric representation of each object;
• generate training data based on the parametric representations and the associated data sets and/or the images, and
• train a generative model using the generated training data, and to store the trained generative model in the memory.

A system for supporting a user in generating or modifying a design of an object, wherein the system comprises

• an input interface configured to acquire at least one image and textual parameters associated with the image;
• and
• a processor configured to
• encode the acquired at least one image by applying a vision-language model based on the textual parameters into vectors including numerical values for the acquired textual parameters,
• generate a parametric representation based on the vectors including the numerical values,
• compute, by applying a trained generative model, corresponding 2D or 3D representation of an object based on the parametric representation, and
• output the 2D or 3D representation of the object to a memory for storing the 2D or 3D representation of the object, or to an output interface for outputting the 2D or 3D representation of the object to the user.