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

Photorealistic editing of portraits is a challenging task as humans are verysensitive to inconsistencies in faces. We present an approach for high-qualityintuitive editing of the camera viewpoint and scene illumination in a portraitimage. This requires our method to capture and control the full reflectancefield of the person in the image. Most editing approaches rely on supervisedlearning using training data captured with setups such as light and camerastages. Such datasets are expensive to acquire, not readily available and donot capture all the rich variations of in-the-wild portrait images. Inaddition, most supervised approaches only focus on relighting, and do not allowcamera viewpoint editing. Thus, they only capture and control a subset of thereflectance field. Recently, portrait editing has been demonstrated byoperating in the generative model space of StyleGAN. While such approaches donot require direct supervision, there is a significant loss of quality whencompared to the supervised approaches. In this paper, we present a method whichlearns from limited supervised training data. The training images only includepeople in a fixed neutral expression with eyes closed, without much hair orbackground variations. Each person is captured under 150 one-light-at-a-timeconditions and under 8 camera poses. Instead of training directly in the imagespace, we design a supervised problem which learns transformations in thelatent space of StyleGAN. This combines the best of supervised learning andgenerative adversarial modeling. We show that the StyleGAN prior allows forgeneralisation to different expressions, hairstyles and backgrounds. Thisproduces high-quality photorealistic results for in-the-wild images andsignificantly outperforms existing methods. Our approach can edit theillumination and pose simultaneously, and runs at interactive rates.

PhotoApp: Photorealistic Appearance Editing of Head Portraits