Garment fabrication using machine learning
Methods and systems are disclosed for generating a physical garment using a machine learning model. The methods and systems receive a plurality of parameters of an optimization problem, the plurality of parameters describing a fashion item, and form a prompt based on values of the plurality of parameters. The prompt is processed by a generative machine learning model to output an image comprising an artificial fashion item corresponding to the values of the plurality of parameters. An augmented reality experience is generated in which a real-world object is overlaid with a virtual object that depicts the artificial fashion item. Feedback associated with the augmented reality experience is used to condition fabrication of a real-world fashion item that resembles the artificial fashion item.
1 . A method comprising:
receiving a plurality of parameters of an optimization problem, the plurality of parameters describing a fashion item;
forming a prompt based on values of the plurality of parameters;
processing the prompt by a generative machine learning model to output an image depicting a first artificial fashion item corresponding to the values of the plurality of parameters;
generating multiple augmented reality experiences, including an augmented reality experience, each comprising a respective virtual object that represents a respective artificial fashion item that has been generated using different values for the plurality of parameters;
generating the augmented reality experience in which a real-world object is overlaid with a virtual object that depicts the first artificial fashion item;
collecting feedback for the multiple augmented reality experiences; and
using first feedback, of the collected feedback, associated with the augmented reality experience to condition fabrication of a real-world fashion item that resembles the first artificial fashion item, the using of the first feedback comprising identifying an individual artificial fashion item that corresponds to a respective one of the multiple augmented reality experiences for which the collected feedback for the multiple augmented reality experiences satisfies one or more fabrication criteria, the real-world fashion item being fabricated using the individual artificial fashion item that has been identified.
2 . The method of claim 1 , further comprising:
generating a solution to the optimization problem that optimizes the plurality of parameters, wherein the prompt is formed using the values of the plurality of parameters that have been optimized.
3 . The method of claim 1 , wherein the prompt comprises a textual prompt.
4 . The method of claim 1 , wherein the generative machine learning model comprises a diffusion model.
5 . The method of claim 1 , wherein the plurality of parameters comprises at least one of a fashion item style, fashion item type, or fashion item theme.
6 . The method of claim 5 , wherein:
the fashion item style comprises at least one of an upper body garment, a lower body garment, or a whole body garment;
the fashion item type describes physical attributes of the fashion item style; and
the fashion item theme comprises one or more graphical elements for a fashion item corresponding to the fashion item style that is of the fashion item type.
7 . The method of claim 1 , further comprising:
accessing a prompt template that includes placeholders for the plurality of parameters, wherein the prompt is formed by populating the prompt template with the values for the received plurality of parameters.
8 . The method of claim 1 , further comprising:
processing the image of the first artificial fashion item by a virtual object machine learning model to generate a three-dimensional model of the first artificial fashion item, wherein the augmented reality experience is generated using the three-dimensional model of the first artificial fashion item.
9 . The method of claim 1 , further comprising:
updating a solution to the optimization problem to generate new values for the plurality of parameters of the optimization problem based on the first feedback.
10 . The method of claim 9 , further comprising:
forming a new prompt comprising the new values for the plurality of parameters;
processing the new prompt by the generative machine learning model to output a new image comprising a new artificial fashion item;
generating a new augmented reality experience in which another real-world object is overlaid with a new virtual object that depicts the new artificial fashion item; and
using new feedback associated with the new augmented reality experience to condition fabrication of the real-world fashion item.
11 . The method of claim 1 , further comprising:
determining whether the first feedback satisfies one or more fabrication criteria.
12 . The method of claim 11 , further comprising:
in response to determining that the first feedback satisfies the one or more fabrication criteria, providing the image comprising the first artificial fashion item to a fashion item manufacturer for fabricating the real-world fashion item.
13 . The method of claim 11 , wherein the first feedback represents engagement with the augmented reality experience comprising at least one of a quantity of users who requested access to the augmented reality experience within a threshold interval, a quantity of users who shared the augmented reality experience with other users, a rating of the augmented reality experience, or a quantity of users who stored images generated using the augmented reality experience.
14 . The method of claim 13 , wherein the one or more fabrication criteria comprises at least one of a threshold quantity of users requesting access to the augmented reality experience, a threshold quantity of users sharing the augmented reality experience, a threshold rating for the augmented reality experience, or a threshold quantity of users who stored images generated using the augmented reality experience.
15 . The method of claim 11 , further comprising:
in response to determining that the first feedback fails to satisfy the one or more fabrication criteria, generating a new solution to the optimization problem to generate new values for the plurality of parameters of the optimization problem.
16 . The method of claim 1 , further comprising:
generating a random seed; and
generating the values for the plurality of parameters based on the random seed.
17 . The method of claim 1 , further comprising:
receiving an image depicting a particular fashion item; and
deriving the values of the plurality of parameters based on the received image that depicts the particular fashion item.
18 . A system comprising:
at least one processor; and
at least one memory component having instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
receiving a plurality of parameters of an optimization problem, the plurality of parameters describing a fashion item;
forming a prompt based on values of the plurality of parameters;
processing the prompt by a generative machine learning model to output an image depicting a first artificial fashion item corresponding to the values of the plurality of parameters;
generating multiple augmented reality experiences, including an augmented reality experience, each comprising a respective virtual object that represents a respective artificial fashion item that has been generated using different values for the plurality of parameters;
generating the augmented reality experience in which a real-world object is overlaid with a virtual object that depicts the first artificial fashion item;
collecting feedback for the multiple augmented reality experiences; and
using first feedback, of the collected feedback, associated with the augmented reality experience to condition fabrication of a real-world fashion item that resembles the first artificial fashion item, the using of the first feedback comprising identifying an individual artificial fashion item that corresponds to a respective one of the multiple augmented reality experiences for which the collected feedback for the multiple augmented reality experiences satisfies one or more fabrication criteria, the real-world fashion item being fabricated using the individual artificial fashion item that has been identified.
19 . A non-transitory computer-readable storage medium having stored thereon instructions that, when executed by at least one processor, cause the at least one processor to perform operations comprising:
receiving a plurality of parameters of an optimization problem, the plurality of parameters describing a fashion item;
forming a prompt based on values of the plurality of parameters;
processing the prompt by a generative machine learning model to output an image depicting a first artificial fashion item corresponding to the values of the plurality of parameters;
generating multiple augmented reality experiences, including an augmented reality experience, each comprising a respective virtual object that represents a respective artificial fashion item that has been generated using different values for the plurality of parameters;
generating the augmented reality experience in which a real-world object is overlaid with a virtual object that depicts the first artificial fashion item;
collecting feedback for the multiple augmented reality experiences; and
using first feedback, of the collected feedback, associated with the augmented reality experience to condition fabrication of a real-world fashion item that resembles the first artificial fashion item, the using of the first feedback comprising identifying an individual artificial fashion item that corresponds to a respective one of the multiple augmented reality experiences for which the collected feedback for the multiple augmented reality experiences satisfies one or more fabrication criteria, the real-world fashion item being fabricated using the individual artificial fashion item that has been identified.
20 . The non-transitory computer-readable storage medium of claim 19 , the operations comprising:
generating a solution to the optimization problem that optimizes the plurality of parameters, wherein the prompt is formed using the values of the plurality of parameters that have been optimized.