Methods and systems for representing a pre-modeled object within virtual reality data

Abstract

An exemplary virtual reality system accesses first and second image datasets representative of first and second captured images depicting a real-world scene from first and second vantage points. The system recognizes a pre-modeled object within both the first and second captured images, and determines first and second confidence metrics representative of objective degrees to which the system accurately recognizes the pre-modeled object within the first and second captured images, respectively. The system further generates, a third image dataset representative of a rendered image based on the first and second image datasets. The rendered image includes a depiction of the pre-modeled object within the real-world scene from a third vantage point, and the generating comprises prioritizing, based on a determination that the second confidence metric is greater than the first confidence metric, the second image dataset over the first image dataset for the depiction of the pre-modeled object.

Description

BACKGROUND INFORMATION[0001]Virtual reality technology allows users of virtual reality media player devices to experience virtual reality worlds. For example, virtual reality worlds may be implemented based on live, camera-captured scenery of a real-world scene to allow users to experience, in real time, real-world places that are difficult, inconvenient, expensive, or otherwise problematic for the users to experience in real life (i.e., in a non-simulated manner). Virtual reality technology may thus provide users with a variety of entertainment, educational, vocational, and / or other enjoyable or valuable experiences that may otherwise be difficult or inconvenient for the users to obtain.[0002]In some examples, it may be desirable for a user to view a virtual reality world based on a real-world scene from a vantage point other than one of the vantage points from which real-time video footage of the real-world scene is being captured by physical capture devices (e.g., video cameras)....

AI Technical Summary

Benefits of technology

The patent describes methods and systems for accurately representing objects in virtual reality data. This involves accessing image datasets from different vantage points and combining them to create a realistic representation of a real-world scene. The invention can help to avoid inaccuracies and distortions in the virtual reality data caused by inconsistencies between different capture devices and the challenge of performing image processing in real-time. The technical effect of the patent is to provide users with a more immersive and accurate virtual reality experience.

Problems solved by technology

Unfortunately, as footage from different capture devices is combined and processed in real time, inaccuracies may enter into resultant virtual reality data.

For example, due to inconsistencies between data detected by different physical capture devices, as well as the significant challenge of performing large amounts of image processing in real time, rendered depictions of objects may become distorted, lost, replicated, moved, or otherwise reproduced inaccurately.

Such inaccuracies may be distracting to users viewing these inconsistent depictions during virtual reality experiences, particularly when the objects depicted are familiar objects or are important to the virtual reality experience.

Images