English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Capturing and editing full head performances enables the creation of virtualcharacters with various applications such as extended reality and mediaproduction. The past few years witnessed a steep rise in the photorealism ofhuman head avatars. Such avatars can be controlled through different input datamodalities, including RGB, audio, depth, IMUs and others. While these datamodalities provide effective means of control, they mostly focus on editing thehead movements such as the facial expressions, head pose and/or cameraviewpoint. In this paper, we propose AvatarStudio, a text-based method forediting the appearance of a dynamic full head avatar. Our approach builds onexisting work to capture dynamic performances of human heads using neuralradiance field (NeRF) and edits this representation with a text-to-imagediffusion model. Specifically, we introduce an optimization strategy forincorporating multiple keyframes representing different camera viewpoints andtime stamps of a video performance into a single diffusion model. Using thispersonalized diffusion model, we edit the dynamic NeRF by introducingview-and-time-aware Score Distillation Sampling (VT-SDS) following amodel-based guidance approach. Our method edits the full head in a canonicalspace, and then propagates these edits to remaining time steps via a pretraineddeformation network. We evaluate our method visually and numerically via a userstudy, and results show that our method outperforms existing approaches. Ourexperiments validate the design choices of our method and highlight that ouredits are genuine, personalized, as well as 3D- and time-consistent.

AvatarStudio: Text-driven Editing of 3D Dynamic Human Head Avatars