Body normal network light and rendering control

1. A method comprising:

receiving, by one or more processors of a client device, an image that includes a depiction of a person wearing a fashion item;

generating, by the one or more processors, a segmentation of the fashion item worn by the person depicted in the image;

extracting a portion of the image corresponding to the segmentation of the fashion item;

estimating an angle of each pixel in the portion of the image relative to a camera used to capture the image; and

applying one or more augmented reality (AR) elements to the fashion item in the image based on the estimated angle of each pixel in the portion of the image relative to the camera used to capture the image.

2. The method of claim 1 , wherein applying the one or more AR elements comprises:

determining that light is being focused on the fashion item from a first direction based on the estimated angle of each pixel; and

modifying pixel values of the portion of the fashion item to re-focus the light on the fashion item from a second direction based on the estimated angle of each pixel.

3. The method of claim 2 , further comprising:

determining that a first pixel in the portion of the image is pointing towards a given direction relative to the camera; and

modifying the first pixel to render a reflection of the re-focused light from the second direction towards the given direction.

4. The method of claim 2 , wherein the light in the image is being focused from a top of the image towards a bottom of the image, and wherein the pixel values are modified to re-focus the light from the bottom towards the top.

5. The method of claim 1 , wherein applying the one or more AR elements comprises:

selecting an AR material for the fashion item; and

replacing a material of the fashion item with the AR material based on the estimated angle of each pixel.

6. The method of claim 5 , further comprising:

determining a light reflection or absorption property of the AR material; and

for each pixel in the portion of the image, computing a new pixel value based on the light reflection or absorption property of the AR material and the estimated angle of the pixel.

7. The method of claim 5 , further comprising applying different AR shades to the fashion item for each different type of the AR material that is selected.

8. The method of claim 5 , further comprising:

displaying a list of different AR materials; and

receiving a selection of the AR material from the list.

9. The method of claim 5 , wherein the AR material comprises gold, mirror, water, or slime.

10. The method of claim 1 , wherein applying the one or more AR elements comprises applying artificial light to the fashion item based on the estimated angle of each pixel.

11. The method of claim 1 , further comprising:

displaying the one or more AR elements on a first portion of the fashion item in a first frame of a video, wherein the person is positioned at a first location in the first frame;

determining that the person has moved from the first location to a second location in a second frame of the video, the first portion of the fashion item being moved to a new location in the second frame of the video; and

updating a display position of the one or more AR elements in the second frame to maintain the display of the one or more AR elements on the first portion of the fashion item.

12. The method of claim 1 , further comprising:

determining that the person depicted in the image has been rotated by a specified amount; and

rotating the one or more AR elements by the specified amount in response to determining that the person has been rotated by the specified amount.

13. The method of claim 1 , wherein the estimated angle of each pixel is computed relative to a surface normal of the camera.

14. The method of claim 1 , wherein estimating the angle of each pixel in the portion of the image relative to the camera used to capture the image comprises:

applying a neural network to the portion of the image, the neural network being trained to establish a relationship between image portions depicting different orientations of fashion items and pixel directions of the fashion items relative to surface normals of cameras used to capture the image portions.

15. The method of claim 1 , wherein estimating the angle of each pixel in the portion of the image relative to the camera used to capture the image comprises:

applying a neural network to the portion of the image, the neural network being trained to establish a relationship between image portions depicting reflections of light from different directions on fashion items and pixel directions of the fashion items relative to surface normals of cameras used to capture the image portions.

16. The method of claim 1 , further comprising:

receiving a request to apply AR light to the fashion item from a new direction; and

training a neural network to estimate lighting conditions for each pixel in the portion when the AR light is applied from the new direction.

17. The method of claim 16 , wherein the neural network is trained to establish a relationship between image portions depicting focusing of light on fashion items from different directions and lighting conditions of pixels of the fashion items.

18. A system comprising:

a processor of a client device; and

a memory component having instructions stored thereon that, when executed by the processor, cause the processor to perform operations comprising:

receiving an image that includes a depiction of a person wearing a fashion item;

generating a segmentation of the fashion item worn by the person depicted in the image;

extracting a portion of the image corresponding to the segmentation of the fashion item;

estimating an angle of each pixel in the portion of the image relative to a camera used to capture the image; and

applying one or more augmented reality elements to the fashion item in the image based on the estimated angle of each pixel in the portion of the image relative to the camera used to capture the image.

19. The system of claim 18 , wherein applying the one or more augmented reality elements comprises:

determining that light is being focused on the fashion item from a first direction based on the estimated angle of each pixel; and

modifying pixel values of the portion of the fashion item to re-focus the light on the fashion item from a second direction based on the estimated angle of each pixel.

20. A non-transitory computer-readable storage medium having stored thereon instructions that, when executed by a processor of a client device, cause the processor to perform operations comprising:

receiving an image that includes a depiction of a person wearing a fashion item;

generating a segmentation of the fashion item worn by the person depicted in the image;

extracting a portion of the image corresponding to the segmentation of the fashion item;

estimating an angle of each pixel in the portion of the image relative to a camera used to capture the image; and

applying one or more augmented reality elements to the fashion item in the image based on the estimated angle of each pixel in the portion of the image relative to the camera used to capture the image.