Systems and Methods for Streaming Different Parts of a Three-Dimensional Model at Different Resolutions

1 . A method comprising:

receiving a request to view a three-dimensional (3D) model from a particular field-of-view;

associating different priorities to different parts of the 3D model in the particular field-of-view;

selecting a first set of Gaussian splats associated with nodes at a first level in a tree structure based on the first set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with a first priority;

selecting a second set of Gaussian splats associated with nodes at a second level in the tree structure based on the second set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with a second priority that is different than the first priority; and

streaming the first set of Gaussian splats and the second set of Gaussian splats in response to the request.

2 . The method of claim 1 further comprising:

generating the first set of Gaussian splats at a first resolution; and

generating the second set of Gaussian splats at a second resolution that is different than the first resolution.

3 . The method of claim 1 further comprising:

generating each Gaussian splat of the first set of Gaussian splats to represent a region of the 3D model that is a first size; and

generating each Gaussian splat of the second set of Gaussian splats to represent a region of the 3D model that is a second size, wherein the second size is larger than the first size.

4 . The method of claim 1 further comprising:

associating the first set of Gaussian splats to a plurality of leaf nodes of the tree structure; and

associating the second set of Gaussian splats to a plurality of parent nodes of the tree structure that are one or more levels above the plurality of leaf nodes based on each Gaussian splat from the second set of Gaussian splats representing parts of the 3D model that are represented by two or more Gaussian splats from the first set of Gaussian splats.

5 . The method of claim 1 further comprising:

associating the first set of Gaussian splats to a plurality of leaf nodes of the tree structure based on the first set of Gaussian splats representing the different parts of the 3D model with the first priority at a first resolution or a first level-of-detail; and

associating the second set of Gaussian splats to a plurality of parent nodes of the tree structure that are one or more levels above the plurality of leaf nodes based on the second set of Gaussian splats representing the different parts of the 3D model with the second priority at a second resolution or a second level-of-detail that is different than the first resolution or the first level-of-detail.

6 . The method of claim 1 further comprising:

associating the first set of Gaussian splats to a plurality of leaf nodes of the tree structure; and

generating each Gaussian splat of the second set of Gaussian splats to span a shape of two or more different Gaussian splats from the first set of Gaussian splats and to have visual characteristics that are derived from visual characteristics of the two or more different Gaussian splats.

7 . The method of claim 1 further comprising:

generating the particular field-of-view with different resolutions based on the first set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with the first priority with a first amount of detail and the second set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with the second priority with a different second amount of detail.

8 . The method of claim 1 further comprising:

determining a performance of a network used to exchange the request, the first set of Gaussian splats, and the second set of Gaussian splats; and

determining that Gaussian splats representing the different parts of the 3D model in the particular field-of-view at the first level of the tree structure do not provide a desired user experience based on the performance of the network.

9 . The method of claim 8 , wherein selecting the second set of Gaussian splats comprises:

determining that the first set of Gaussian splats and the second set of Gaussian splats collectively produce the particular field-of-view at different resolution with a total amount of data that provides the desired user experience based on the performance of the network.

10 . The method of claim 1 further comprising:

retrieving the tree structure in response to receiving the request, wherein the tree structure comprises a plurality of nodes that are linked hierarchically with two or more children nodes to each parent node and with each node of the plurality of nodes at a different level of the tree structure being associated with a Gaussian splat for representing a part of the 3D model with a different resolution.

11 . An adaptive streaming system comprising:

one or more hardware processors configured to:

receive a request to view a three-dimensional (3D) model from a particular field-of-view;

associate different priorities to different parts of the 3D model in the particular field-of-view;

select a first set of Gaussian splats associated with nodes at a first level in a tree structure based on the first set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with a first priority;

select a second set of Gaussian splats associated with nodes at a second level in the tree structure based on the second set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with a second priority that is different than the first priority; and

stream the first set of Gaussian splats and the second set of Gaussian splats in response to the request.

12 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

generate the first set of Gaussian splats at a first resolution; and

generate the second set of Gaussian splats at a second resolution that is different than the first resolution.

13 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

generate each Gaussian splat of the first set of Gaussian splats to represent a region of the 3D model that is a first size; and

generate each Gaussian splat of the second set of Gaussian splats to represent a region of the 3D model that is a second size, wherein the second size is larger than the first size.

14 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

associate the first set of Gaussian splats to a plurality of leaf nodes of the tree structure; and

associate the second set of Gaussian splats to a plurality of parent nodes of the tree structure that are one or more levels above the plurality of leaf nodes based on each Gaussian splat from the second set of Gaussian splats representing parts of the 3D model that are represented by two or more Gaussian splats from the first set of Gaussian splats.

15 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

associate the first set of Gaussian splats to a plurality of leaf nodes of the tree structure based on the first set of Gaussian splats representing the different parts of the 3D model with the first priority at a first resolution or a first level-of-detail; and

associate the second set of Gaussian splats to a plurality of parent nodes of the tree structure that are one or more levels above the plurality of leaf nodes based on the second set of Gaussian splats representing the different parts of the 3D model with the second priority at a second resolution or a second level-of-detail that is different than the first resolution or the first level-of-detail.

16 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

associate the first set of Gaussian splats to a plurality of leaf nodes of the tree structure; and

generate each Gaussian splat of the second set of Gaussian splats to span a shape of two or more different Gaussian splats from the first set of Gaussian splats and to have visual characteristics that are derived from visual characteristics of the two or more different Gaussian splats.

17 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

generate the particular field-of-view with different resolutions based on the first set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with the first priority with a first amount of detail and the second set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with the second priority with a different second amount of detail.

18 . The adaptive streaming system of claim 11 , wherein the one or more hardware processors are further configured to:

determine a performance of a network used to exchange the request, the first set of Gaussian splats, and the second set of Gaussian splats; and

determine that Gaussian splats representing the different parts of the 3D model in the particular field-of-view at the first level of the tree structure do not provide a desired user experience based on the performance of the network.

19 . The adaptive streaming system of claim 18 , wherein selecting the second set of Gaussian splats comprises:

determining that the first set of Gaussian splats and the second set of Gaussian splats collectively produce the particular field-of-view at different resolution with a total amount of data that provides the desired user experience based on the performance of the network.

20 . A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of an adaptive streaming system, cause the adaptive streaming system to perform operations comprising:

receiving a request to view a three-dimensional (3D) model from a particular field-of-view;

associating different priorities to different parts of the 3D model in the particular field-of-view;

selecting a first set of Gaussian splats associated with nodes at a first level in a tree structure based on the first set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with a first priority;

selecting a second set of Gaussian splats associated with nodes at a second level in the tree structure based on the second set of Gaussian splats representing the different parts of the 3D model in the particular field-of-view with a second priority that is different than the first priority; and

streaming the first set of Gaussian splats and the second set of Gaussian splats in response to the request.