Changing directions of gaze during a video conference

1 . A method, comprising:

identifying, during a video conference, a first participant based on a direction of a gaze of a second participant toward a user tile associated with the first participant within a graphical user interface (GUI) of the video conference; and

causing a change, in a video stream transmitted to a device of the first participant during the video conference, to the direction of the gaze of the second participant to appear directed toward a camera of a device of the second participant, wherein the change comprises generating a personalized video stream for the first participant such that a gaze correction is applied in the video stream transmitted to the device of the first participant and not transmitted to other participant devices;

wherein the personalized video stream is generated by digitally correcting, for each recipient participant, the gaze of the second participant in a composite video stream based on a user tile position of the recipient in the GUI, such that the gaze correction is specific to a view of the recipient and is not present in video streams transmitted to other participants.

2 . The method of claim 1 , further comprising:

causing a change, in a second video stream transmitted to a device of a third participant, to the direction of the gaze of the second participant to appear directed toward the user tile.

3 . The method of claim 1 , further comprising:

calculating, from pupils of the second participant, an intersection point at the GUI; and

correlating the intersection point to the user tile.

4 . The method of claim 1 , further comprising:

using a machine learning model trained on historical gaze intersection data from prior video conferences to determine an intersection point at the GUI based on the gaze; and

correlating the intersection point to a position of the user tile.

5 . The method of claim 1 , wherein causing the change comprises:

selecting the camera, from a plurality of cameras of the device of the second participant, during the video conference, based on a calculated intersection point, such that the selected camera is physically closest to the user tile as displayed in the GUI.

6 . The method of claim 1 , wherein causing the change comprises:

correcting, digitally in a user tile associated with the second participant, pupils of the second participant.

7 . The method of claim 1 , further comprising:

causing a change, in a second video stream transmitted to a device of a third participant, to a 3D representation of the second participant to appear directed toward a 3D representation of the first participant.

8 . The method of claim 1 , further comprising:

outputting a plurality of user tiles, in the GUI, in a gallery layout; and

maintaining a data structure indicating X and Y positions of user tiles in the gallery layout during the video conference.

9 . A system, comprising:

a memory subsystem; and

processing circuitry configured to execute instructions stored in the memory subsystem to:

identify, during a video conference, a first participant based on a direction of a gaze of a second participant toward a user tile associated with the first participant within a graphical user interface (GUI) of the video conference; and

cause a change, in a video stream transmitted to a device of the first participant during the video conference, to the direction of the gaze of the second participant to appear directed toward a camera of a device of the second participant, wherein the change comprises a generation of a personalized video stream for the first participant such that a gaze correction is applied in the video stream transmitted to the device of the first participant and not transmitted to other participant devices;

wherein the personalized video stream is generated by digitally correcting for each recipient participant, the gaze of the second participant in a composite video stream based on a user tile position of the recipient in the GUI, such that the gaze correction is specific to a view of the recipient and is not present in video streams transmitted to other participants.

10 . The system of claim 9 , wherein the processing circuitry is further configured to execute instructions stored in the memory subsystem to:

arrange a plurality of user tiles, including the user tile associated with the first participant, in a gallery layout.

11 . The system of claim 9 , wherein the processing circuitry is further configured to execute instructions stored in the memory subsystem to:

correct, digitally in a user tile associated with the second participant, a head of the second participant.

12 . The system of claim 9 , wherein the processing circuitry is further configured to execute instructions stored in the memory subsystem to:

determine the user tile based on an intersection point at the GUI from pupils of the second participant.

13 . The system of claim 9 , wherein the processing circuitry is further configured to execute instructions stored in the memory subsystem to:

train a machine learning model, using a dataset of past intersection points and associated user tile selections.

14 . The system of claim 9 , wherein the processing circuitry is further configured to execute instructions stored in the memory subsystem to:

change from a first camera that is a default camera during the video conference to the camera based on the camera being closest to the user tile.

15 . A non-transitory computer readable medium storing instructions operable to cause one or more processors to perform operations comprising:

identifying, during a video conference, a first participant based on a direction of a gaze of a second participant toward a user tile associated with the first participant within a graphical user interface (GUI) of the video conference; and

causing a change, in a video stream transmitted to a device of the first participant during the video conference, to the direction of the gaze of the second participant to appear directed toward a camera of a device of the second participant, wherein the change comprises generating a personalized video stream for the participant such that a gaze correction is applied in the video stream transmitted to the device of the first participant and not transmitted to other participant devices;

wherein the personalized video stream is generated by digitally correcting, for each recipient participant, the gaze of the second participant in a composite video stream based on a user tile position of the recipient in the GUI, such that the gaze correction is specific to a view of the recipient and is not present in video streams transmitted to other participants.

16 . The non-transitory computer readable medium of claim 15 , wherein the video stream is a first composite video stream, transmitted to the device of the first participant, different from a second composite video stream transmitted to the device of the second participant.

17 . The non-transitory computer readable medium of claim 15 , wherein the operations further comprise:

selecting a default camera for the video conference; and

changing from the default camera to the camera based on the camera being closest to the user tile.

18 . The non-transitory computer readable medium of claim 15 , wherein the operations further comprise:

causing a change, in a second video stream transmitted to a device of a third participant, to a 3D representation of the second participant to appear directed toward a 3D representation of the first participant.

19 . The non-transitory computer readable medium of claim 15 , wherein the operations further comprise:

outputting a plurality of user tiles, in the GUI, in a gallery layout; and

maintaining a data structure indicating Cartesian coordinates of user tiles in the gallery layout during the video conference.

20 . The non-transitory computer readable medium of claim 15 , wherein the operations further comprise:

correcting, digitally in a user tile associated with the second participant, a body of the second participant.