Dense face grid expression driving method and device and medium

Abstract

The invention discloses a dense face grid expression driving method and device and a medium, and relates to the field of artificial intelligence. The method comprises the steps of obtaining target grid assets and target expression control data; the target grid assets are topologically arranged on the standard face 3D grid to obtain topologically arranged target grid assets; obtaining an initial face feature offset value according to the target grid asset and the standard face 3D grid, and obtaining a corresponding target change offset according to the target expression control data and the face expression change model; and superposing the initial face feature offset value and the target change offset to the topological target grid asset, thereby generating global offset for the vertex position in the expression-free grid data in the topological target grid asset to obtain a target expression. Therefore, in the scheme, the expression change of the target grid assets can be output without binding the grid assets and inputting the control data of the target expressions to the facial expression change model, so that facial binding is avoided, and the facial expression making efficiency is improved.

Description

technical field [0001] The present application relates to the field of artificial intelligence, and in particular, to a method, device and medium for driving dense face grid expressions. Background technique [0002] At present, artificial intelligence technology and Augmented Reality (AR) technology are developing rapidly. Internet users have more and more demand for content ingestion, and the demand for animation, virtual digital image, AR and other content is more and more intense. [0003] In the production process of three-dimensional (3-Dimension, 3D) animation or virtual human computer animation (Computer Graphics, CG), the animator first needs to perform face binding and add controllers based on different animation images, or design blend deformation ( Blend Shape) can drive the 3D dense mesh to present expression changes, the steps are cumbersome, consume a lot of time, and also have high requirements on the personal experience and animation production level of the ...

AI Technical Summary

Problems solved by technology

[0003] In the production process of three-dimensional (3-Dimension, 3D) animation or virtual human computer graphics (Computer Graphics, CG), it is first necessary for an animator to bind faces and add controllers based on different animation images, or design hybrid deformation ( Blend Shape) can drive 3D dense grids to present expression changes, the steps are cumbersome, it takes a lot of time, and it also has high requirements for the animator's personal experience and animation production level

Even so, the facial expressions after binding also show different effects due to the different personal styles and aesthetics of different animators, which increases the post-processing time and the uncontrollability of the final effect

Images