How the “Ant Ya Hey” AI Effect Works and How to Create It

The Douyin AI effect “Ant Ya Hey” lets anyone turn a single photo into a singing, dancing avatar that syncs with music, and it has gone viral with celebrities like Jack Ma, Mark Zuckerberg, and Elon Musk joining the fun.

Users can achieve the effect by using three tools: the Avatarify app, an accelerator (VPN) to improve network speed, and optionally the video editor Jianying for multi‑person collages.

Step 1: Enable acceleration mode in the VPN app. Step 2: Open Avatarify, select a portrait, and choose the “Brad Pitt” (or other) effect. Step 3: (Optional) Use Jianying to combine several avatars into a grid or add background music.

“Short seconds, a lifetime to cure….” – a user’s reaction to the addictive effect.

The technology behind the effect is the First‑Order Motion Model, a research work presented at NeurIPS 2019 by the University of Trento. It animates a face by predicting a dense motion field using key‑point detection and local affine transformations.

The model consists of two modules: a motion‑estimation network that predicts motion from a reference frame to source and driver frames, and an image‑generation network that warps the source image and fills occluded regions using a learned generator.

During training, large video datasets of the same object class are used; during inference, the model animates a static portrait according to a driving video, achieving higher quality than previous methods on standard benchmarks.

Because the effect requires no programming knowledge, it has become a “new favorite” among Douyin users, surpassing major apps in download rankings.

Overall, the “Ant Ya Hey” effect demonstrates how advanced AI research can be packaged into user‑friendly tools that enable anyone to create convincing deep‑fake style animations.