At Eovaldi Art Science, we continuously explore the intersection of art, science, and AI. Our latest project is an AI-generated video constructed from Robert Eovaldi’s photography, enhanced using Fourier Transform techniques and synthesized in Runway ML. This article breaks down the technical aspects of the video’s creation, examining how AI models translate still photography into dynamic moving images.

The Role of Fourier Transforms in Image Processing

One of the key mathematical tools used in the generation of this video is the Fourier Transform (FT). This technique decomposes an image into its frequency components, allowing for:

• Noise reduction: Enhancing clarity by filtering out unwanted high-frequency elements.

• Texture and motion synthesis: Creating realistic transitions between frames by modifying frequency domains.

• Style transfer: Mapping the spectral properties of an image onto another, ensuring smooth transitions in AI-driven animations... Plus other mathematical modulations to create rays.

By leveraging Fast Fourier Transform (FFT) algorithms, the AI models can analyze Robert Eovaldi’s static images and reconstruct them in a form that supports fluid movement.

Runway ML and AI Video Generation

Runway ML is an AI-powered creative toolkit that enables artists and researchers to generate realistic videos from still images. The video generated from Robert Eovaldi’s photography was created using Runway Gen-3 Alpha, an advanced AI model that incorporates:

• Latent Space Interpolation: This allows the model to generate smooth transitions between still images, creating the illusion of motion.

• Depth and Motion Estimation: The AI predicts how objects in a still image might move in a three-dimensional space, ensuring realistic depth perception.

• Fourier-Based Motion Enhancements: Fourier transforms assist in refining edge details and smoothing abrupt transitions to enhance visual consistency.

Why This Matters for AI Art and Video Synthesis

This project demonstrates how AI, mathematics, and human creativity can converge to produce immersive and dynamic visual experiences. By using Fourier Transforms, AI models can better understand the underlying structure of images, making them more adept at generating movement from static sources.