Deepfake for the Good: Generating Avatars through Face-Swapping with Implicit Deepfake Generation

TL;DR

This paper explores combining neural rendering techniques with deepfake methods to generate realistic 3D avatars, demonstrating a novel approach that leverages face-swapping and neural encoding for improved avatar creation.

Contribution

It introduces ImplicitDeepfake, a method that integrates deepfake face modification with NeRF and Gaussian Splatting for plausible 3D avatar generation.

Findings

Produces realistic 3D avatars using face-swapping techniques.

Combines deepfake modifications with neural rendering for enhanced plausibility.

Demonstrates feasibility of using deepfake for avatar creation in spatial computing.

Abstract

Numerous emerging deep-learning techniques have had a substantial impact on computer graphics. Among the most promising breakthroughs are the rise of Neural Radiance Fields (NeRFs) and Gaussian Splatting (GS). NeRFs encode the object's shape and color in neural network weights using a handful of images with known camera positions to generate novel views. In contrast, GS provides accelerated training and inference without a decrease in rendering quality by encoding the object's characteristics in a collection of Gaussian distributions. These two techniques have found many use cases in spatial computing and other domains. On the other hand, the emergence of deepfake methods has sparked considerable controversy. Deepfakes refers to artificial intelligence-generated videos that closely mimic authentic footage. Using generative models, they can modify facial features, enabling the creation of altered identities or expressions that exhibit a remarkably realistic appearance to a real person. Despite these controversies, deepfake can offer a next-generation solution for avatar creation and gaming when of desirable quality. To that end, we show how to combine all these emerging technologies to obtain a more plausible outcome. Our ImplicitDeepfake uses the classical deepfake algorithm to modify all training images separately and then train NeRF and GS on modified faces. Such simple strategies can produce plausible 3D deepfake-based avatars.