Technique to change garments within an existing image

Patent metadata

Field	Value
Publication number	US-12037740-B2
Application number	US-202318186134-A
Country	US
Kind code	B2
Filing date	Mar 17, 2023
Priority date	Feb 27, 2018
Publication date	Jul 16, 2024
Grant date	Jul 16, 2024

Abstract

A system automatically generates apparel collection imagery from user-provided imagery. The user-provided imagery includes images of people wearing one or more garments. The system uses segmenting analysis to analyze the user-provided image to identify locations of the garment. From the locations of the garments, the system can determine which garments from an apparel collection can be used to replace those in the user-provided imagery. The system uses pose estimation on the user-provided imagery and modifies a preview image of a replacement garment from the collection. This modified replacement garment image is used to replace the garment in the user-provided imagery.

First claim

The invention claimed is:

1. A method comprising:
in a garment design tool, generating a first preview image on a computer screen of a garment with a finishing pattern to be created by a laser using a laser input file, wherein the generating comprises
providing a base image of a garment without the finishing pattern,
generating an adjusted base image from the base image of the garment without the finishing pattern,
generating a pattern mask based on the laser input file,
for a pixel at a pixel location of the preview image, obtaining a first contribution by combining a first value associated with a pixel of the pattern mask at a pattern mask pixel location that corresponds to the pixel location of the preview image and a value of a pixel of the base image at a base image pixel location that corresponds to the pixel location of the preview image,
for the pixel at the pixel location of the preview image, obtaining a second contribution by combining a second value associated with the pixel of the pattern mask at the pattern mask pixel location that corresponds to the pixel location of the preview image and a value of a pixel of the adjusted base image at an adjusted base image pixel location that corresponds to the pixel location of the preview image,
combining the first contribution and second contribution to obtain a combined value for the pixel at the pixel location for the preview image, and
displaying the generated first preview image with the pixel on the computer screen, wherein the pixel is displayed as a color corresponding to the combined value for the pixel at the pixel location;
providing a user-provided image, wherein the user-provided image comprises a person or portion of a person wearing a first user-provided garment;
analyzing the user-provided image to identify a location of the first user-provided garment, wherein the first user-provided garment is at a first location;
determining a first pose of the person or the portion of the person at the first location captured in the user-provided image;
modifying the first preview image based on the first pose captured in the user-provided image to obtain a modified first preview image; and
modifying the user-provided image by replacing the first user-provided garment at the first location with the modified first preview image.

2. The method of claim 1 comprising:
analyzing the user-provided image to identify a location of a second user-provided garment, wherein the second user-provided garment is at a second location;
determining a second pose of another person or portion of another person at the second location captured in the user-provided image providing a second preview image;
modifying the second preview image based on the second pose captured in the user-provided image to obtain a modified second preview image; and
modifying the user-provided image by replacing the second user-provided garment at the second location with the modified second preview image.

3. The method of claim 1 wherein the determining the first pose comprises detecting an articulation point corresponding to a knee area of the person captured in the user-provided image and the modified first preview image comprises an adjustment due to the detected articulation point.

4. The method of claim 1 wherein the modified user-provided image is saved as a replacement image, and the replacement image has an appearance similar to the user-provided image except that the first user-provided garment has been replaced by the modified first preview image.

5. The method of claim 2 wherein the modified user-provided image is saved as a replacement image, and the replacement image has an appearance similar to the user-provided image except that the first user-provided garment has been replaced by the modified first preview image, and the second user-provided garment has been replaced by the modified second preview image.

6. The method of claim 1 wherein the first user-provided garment is a pair of jeans or pants.

7. The method of claim 2 wherein the first user-provided garment is a pair of jeans or pants, and the second user-provided garment is a garment top.

8. The method of claim 1 wherein the user-provided image comprises at least one of an IMG, TIFF, EXIF, JPEG, GIF, PNG, PBM, PPM, BMP, or RAW format file.

9. A method comprising:
in a garment design tool, generating a first preview image on a computer screen of a garment with a finishing pattern to be created by a laser using a laser input file, wherein the generating comprises
providing a base image of a garment without the finishing pattern,
generating an adjusted base image from the base image of the garment without the finishing pattern,
generating a pattern mask based on the laser input file,
for a pixel at a pixel location of the preview image, obtaining a first contribution by combining a first value associated with a pixel of the pattern mask at a pattern mask pixel location that corresponds to the pixel location of the preview image and a value of a pixel of the base image at a base image pixel location that corresponds to the pixel location of the preview image,
for the pixel at the pixel location of the preview image, obtaining a second contribution by combining a second value associated with the pixel of the pattern mask at the pattern mask pixel location that corresponds to the pixel location of the preview image and a value of a pixel of the adjusted base image at an adjusted base image pixel location that corresponds to the pixel location of the preview image,
combining the first contribution and second contribution to obtain a combined value for the pixel at the pixel location for the preview image, and
displaying the generated first preview image with the pixel on the computer screen, wherein the pixel is displayed as a color corresponding to the combined value for the pixel at the pixel location;
providing a second preview image;
providing a user-provided image comprising a first user-provided garment and a second user-provided garment;
analyzing the user-provided image to identify locations of the first user-provided garment and second user-provided garment, wherein the first user-provided garment is at a first location and the second user-provided garment is at a second location, different from the first location;
based on the analysis of the user-provided image, determining that the first preview image is more appropriate for the first location than the second location;
determining a first pose of a person or portion of a person at the first location captured in the user-provided image;
modifying the first preview image based on the first pose captured in the user-provided image to obtain a modified first preview image; and
generating a replacement image having an appearance similar to the user-provided image, wherein changes in the replacement image comprise the first user-provided garment at the first location of the user-provided image being replaced by the modified first preview image.

10. The method of claim 9 wherein the determining the first pose comprises detecting an articulation point corresponding to a knee area of the person or portion of the person captured in the user-provided image and the modified first preview image comprises an adjustment due to the detected articulation point.

11. The method of claim 9 comprising:
determining a second pose of another person or portion of another person at the second location captured in the user-provided image;
modifying the second preview image based on the second pose captured in the user-provided image to obtain a modified second preview image;
generating a replacement image having an appearance similar to th

Assignees

• Strauss Levi & Co

Inventors

• Mahanty Debdulal
• Bell Benjamin
• Sights James Barton
• Schultz Christopher
• Schultz Jennifer

Classifications

• G06T3/08

  Projecting images onto non-planar surfaces, e.g. geodetic screens · CPC title

• G06T11/00

  Two-dimensional [2D] image generation · CPC title

• G06T2207/20221

  Image fusion; Image merging · CPC title

• G06T2207/10024

  Color image · CPC title

• A41H3/007

  Methods of drafting or marking-out patterns using computers (computer-aided design G06F30/00) · CPC title