Processing 3D image sequences

摘要

Various implementations provide techniques to prevent excessive parallax, depth, or disparity from being passed through to a viewer. In one particular implementation, it is determined that a depth indicator for an object in a stereoscopic image pair of a video sequence is outside of a target range. One or more images of the stereoscopic image pair is modified so that the depth indicator for the object is within the target range. In other implementations, a depth transition between the object and another portion of the video sequence is smoothed. In further implementations, the stereoscopic image pair is replaced with a 2D image pair that includes the object. In yet further implementations, a resulting video sequence includes (i) one or more stereoscopic image pairs having non-zero disparity and for which the depth indicator is within the target range for the entire image pair, and (ii) one or more 2D image pairs.

主权项

1. A method comprising:
determining that a depth indicator for an object in a stereoscopic image pair of a video sequence is outside of a target range; modifying one or more images of the stereoscopic image pair so that the depth indicator for the object is within the target range; and smoothing a depth transition between the object and another portion of the video sequence, wherein the smoothing is performed until a gradient of the depth indicator is within a target gradient range.