Automatically controlling modifications to typeface designs with machine-learning models

Abstract

A typeface design system displays, using a typeface design interface, one or more control points for modifying one or more curves of a design that define an outline of an input character. The typeface design system accesses a machine-learning model trained to recognize the input character as a reference character. The typeface design system receives an input modifying the design including a change in position of an input control point of the one or more control points from a first position to a second position. The typeface design system determines, by the trained machine-learning model, that the reference character does not match the input character having the modified design based on determining that the second position is outside of a particular region defined by the machine-learning model. The typeface design system rejects the input and maintains the design of the input character as displayed prior to receiving the input.

1 . A method for automatically controlling modifications to typeface designs, the method comprising:
 providing, by one or more processing devices, a design interface for modifying a design of an input character of a typeface, wherein the design interface displays one or more control points for modifying one or more curves of the design that define an outline of the input character;   accessing a machine-learning model that is trained to recognize the input character as a reference character;   receiving, via the design interface, an input modifying the design of the input character, comprising a change in position of an input control point of the one or more control points from a first position to a second position;   determining, by the trained machine-learning model, that the reference character does not match the input character having the modified design based on determining that the second position of the input control point is outside of a particular region defined by the machine-learning model; and   rejecting via the design interface, the input modifying the design of the input character and maintaining the design of the input character as displayed prior to receiving the input.   
     
     
         2 . The method of  claim 1 , further comprising:
 presenting, via the design interface, a warning that the input character is not recognized as the reference character.   
     
     
         3 . The method of  claim 1 , further comprising:
 applying, responsive to the input, an alternative modification to the design of the input character;   determining that the machine-learning model matches the reference character to the input character having the alternative modification to the design; and   updating the design interface to display the input character having the alternative modification to the design.   
     
     
         4 . The method of  claim 1 , further comprising generating an input feature vector from the input character having the modified design, wherein determining that the machine-learning model cannot match the reference character to the input character having the modified design is further based on determining that that the input feature vector is outside the region of the vector space associated with the reference character. 
     
     
         5 . The method of  claim 1 , further comprising, prior to receiving the input modifying the design:
 receiving, via the design interface, an additional input applying an interim modification to the design of the input character;   generating an input feature vector from the input character having the design with the interim modification;   identifying, from the machine-learning model, a boundary of a region of a vector space associated with the reference character and a threshold distance from the boundary;   determining that (i) the machine-learning model matches the reference character to the input character having the design with the interim modification and (ii) the input feature vector identifies a position in the vector space within the threshold distance from the boundary; and   outputting, via the design interface and based on the input feature vector being within the threshold distance from the boundary, a warning indicating that the interim modification has reduced an ability of the machine-learning model to classify the input character having the interim modification as the reference character.   
     
     
         6 . The method of  claim 1 , wherein the input modifying the design of the input character comprises a dragging input from first position to the second position. 
     
     
         7 . The method of  claim 1 , wherein the change in position of the input control point from the first position to the second position increases or decreases a length of a curve of the one or more curves of the design that define an outline of the input character. 
     
     
         8 . A system comprising:
 a processing device; and   a non-transitory computer-readable medium communicatively coupled to the processing device, wherein the processing device is configured to execute computer readable program instructions stored in the non-transitory computer-readable medium and thereby perform operations comprising:
 providing, by one or more processing devices, a design interface for modifying a design of an input character of a typeface, wherein the design interface displays one or more control points for modifying one or more curves of the design that define an outline of the input character; 
 accessing, by the typeface design application, a machine-learning model that is trained to recognize the input character as a reference character; 
 receiving, via the design interface, an input modifying the design of the input character,-comprising a change in position of an input control point of the one or more control points from a first position to a second position; 
 determining, by the trained machine-learning model, that the reference character does not match the input character having the modified design based on determining that the second position of the input control point is outside of a particular region defined by the machine-learning model; and 
 rejecting via the design interface, the input modifying the design of the input character and maintaining the design of the input character as displayed prior to receiving the input. 
   
     
     
         9 . The system of  claim 8 , the operations further comprising:
 presenting, via the design interface, a warning that the input character is not recognized as the reference character.   
     
     
         10 . The system of  claim 8 , the operations further comprising:
 applying, responsive to the input, an alternative modification to the design of the input character;   determining that the machine-learning model matches the reference character to the input character having the alternative modification to the design; and   updating the design interface to display the input character having the alternative modification to the design.   
     
     
         11 . The system of  claim 8 , the operations further comprising generating an input feature vector from the input character having the modified design, wherein determining that the machine-learning model cannot match the reference character to the input character having the modified design is further based on determining that that the input feature vector is outside the region of the vector space associated with the reference character. 
     
     
         12 . The system of  claim 8 , the operations further comprising, prior to receiving the input modifying the design:
 receiving, via the design interface, an additional input applying an interim modification to the design of the input character;   generating an input feature vector from the input character having the design with the interim modification;   identifying, from the machine-learning model, a boundary of a region of a vector space associated with the reference character and a threshold distance from the boundary;   determining that (i) the machine-learning model matches the reference character to the input character having the design with the interim modification and (ii) the input feature vector identifies a position in the vector space within the threshold distance from the boundary; and   outputting, via the design interface and based on the input feature vector being within the threshold distance from the boundary, a warning indicating that the interim modification has reduced an ability of the machine-learning model to classify the input character having the interim modification as the reference character.   
     
     
         13 . The system of  claim 8 , wherein the input modifying the design of the input character comprises a dragging input from first position to the second position. 
     
     
         14 . The system of  claim 8 , wherein the change in position of the input control point from the first position to the second position increases or decreases a length of a curve of the one or more curves of the design that define an outline of the input character. 
     
     
         15 . A non-transitory computer-readable medium having program code stored thereon and that is executable by one or more processing devices for performing comprising:
 providing, by one or more processing devices, a design interface for modifying a design of an input character of a typeface, wherein the design interface displays one or more control points for modifying one or more curves of the design that define an outline of the input character;   accessing, by the typeface design application, a machine-learning model that is trained to recognize the input character as a reference character;   receiving, via the design interface, an input modifying the design of the input character, comprising a change in position of an input control point of the one or more control points from a first position to a second position;   determining, by the trained machine-learning model, that the reference character does not match the input character having the modified design based on determining that the second position of the input control point is outside of a particular region defined by the machine-learning model; and   rejecting via the design interface, the input modifying the design of the input character and maintaining the design of the input character as displayed prior to receiving the input.   
     
     
         16 . The non-transitory computer-readable medium of  claim 15 , the operations further comprising:
 presenting, via the design interface, a warning that the input character is not recognized as the reference character.   
     
     
         17 . The non-transitory computer-readable medium of  claim 15 , the operations further comprising:
 applying, responsive to the input, an alternative modification to the design of the input character;   determining that the machine-learning model matches the reference character to the input character having the alternative modification to the design; and   updating the design interface to display the input character having the alternative modification to the design.   
     
     
         18 . The non-transitory computer-readable medium of  claim 15 , the operations further comprising generating an input feature vector from the input character having the modified design, wherein determining that the machine-learning model cannot match the reference character to the input character having the modified design is further based on determining that that the input feature vector is outside the region of the vector space associated with the reference character. 
     
     
         19 . The non-transitory computer-readable medium of  claim 15 , wherein the input modifying the design of the input character comprises a dragging input from first position to the second position. 
     
     
         20 . The non-transitory computer-readable medium of  claim 15 , wherein the change in position of the input control point from the first position to the second position increases or decreases a length of a curve of the one or more curves of the design that define an outline of the input character.