Method to create animation

A method of generating facial animation comprising:

• inputting two or more training input data into a facial animation generation model;
• training a common feature of the two or more training input data using a first network function included in the facial animation generation model;
• and
• training an independent feature of each of the two or more training input data using a second network function to cause the facial animation generation model to generate a facial animation according to input data;
• wherein the facial animation generation model is trained by differently setting an update rate of weights for the first network function and remaining layers except at least one layer of the second network function and an update rate of weights for the at least one layer of the second network function, during a predetermined epoch.

A server for generating facial animation, comprising:

• a processor including one or more cores;
• and
• a memory;
• wherein the processor is configured to
• input two or more training input data into a facial animation generation model;
• train a common feature of the two or more training input data using a first network function included in the facial animation generation model;
• and
• train an independent feature of each of the two or more training input data using a second network function to cause the facial animation generation model to generate a facial animation according to input data;
• wherein the facial animation generation model is trained by differently setting an update rate of weights for the first network function and remaining layers except at least one layer of the second network function and an update rate of weights for the at least one layer of the second network function, during a predetermined epoch.

A non-transitory computer readable medium storing data structure corresponding to weights of a neural network, at least one of the weights being updated during the training process, wherein the operation of the neural network is based at least in part parameters, the training process comprising:

• inputting two or more training input data into a facial animation generation model;
• training a common feature of the two or more training input data using a first network function included in the facial animation generation model;
• and
• training an independent feature of each of the two or more training input data using a second network function to cause the facial animation generation model to generate a facial animation according to input data;
• wherein the facial animation generation model is trained by differently setting an update rate of weights for the first network function and remaining layers except at least one layer of the second network function and an update rate of weights for the at least one layer of the second network function, during a predetermined epoch.