H Peter Alesso
Visual Synthesis in CGI and Deepfakes
The landscape of visual content creation has seen significant advancements with the emergence of Computer Generated Imagery (CGI) and Deepfakes. While CGI has been a cornerstone of visual effects in movies and television for years, creating detailed digital characters and stunning landscapes , Deepfakes, a more recent product of artificial intelligence (AI), have begun to generate realistic video content that blurs the line between reality and fiction .
Computer Generated Imagery, or CGI, involves using computers and specific software to create graphics. Originating in the 1960s for scientific and engineering purposes, CGI later entered the entertainment industry, significantly impacting cinema and video games .
CGI technology provides a way to create detailed and lifelike visual effects that would otherwise be impossible, costly, or risky to produce. From the expansive universes of "Avatar" and "The Lord of the Rings" to the exhilarating scenes of the "Marvel Cinematic Universe," CGI has been vital in delivering compelling narratives to audiences worldwide .
The Emergence of Deepfakes
Deepfakes, a term combining 'deep learning' and 'fake,' are synthetic media that superimpose one person's likeness onto another . Deepfakes stem from a machine learning subset called Generative Adversarial Networks (GANs) .
In the context of GANs, two neural networks compete: one network (the generator) manufactures synthetic data, and the other network (the discriminator) tries to differentiate between the real and the fake. This competition hones the generator's skills to produce increasingly realistic synthetic data . Deepfakes utilize this framework to switch faces, modify speech, and create persuasive synthetic media.
The Fusion of CGI and Deepfakes
With the advancements in deep learning, the boundaries between CGI and Deepfakes are starting to blur. Both techniques now work together to create realistic digital humans and environments. For example, in filmmaking, deep learning methodologies are enhancing motion capture, facial animation, and character modeling - traditionally the domain of CGI.
Moreover, Deepfakes are making content creation tools more accessible. While sophisticated CGI usually needs substantial computing resources and technical know-how, Deepfake technology can generate convincing outputs using common hardware and freely available software.
Despite the intriguing prospects these technologies present, they also raise serious ethical and legal issues. Deepfakes, particularly, have been exploited for disinformation and manipulation, representing substantial societal challenges. Additionally, the ease of access to these tools amplifies their potential for misuse.
CGI and Deepfakes are at the forefront of visual content creation, ushering in innovative forms of expression and storytelling. However, their potential abuse calls for thoughtful ethical discussions and suitable regulatory action. As these technologies continue to mature, they will undeniably redefine our perception of reality in digital media.
 Magnenat-Thalmann, N., & Thalmann, D. (Eds.). (2013). Handbook of virtual humans. John Wiley & Sons.
 Chesney, R., & Citron, D. (2018). Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security. SSRN Electronic Journal.
 Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D.,