4D geometry enthusiast, Physics Programmer at Ready at Dawn Studios

Visualization of an antitwister mechanism - a way of connecting a flexible link between two objects, one of which is rotating with respect to the other, in a way that prevents the link from becoming twisted.

Rotating 3D projections of regular 4D polytopes (four dimensional analogues of 3D polyhedra).

Scaling and dropping a twist deformation through a stack of transparent planes.

Senior Research Engineer at Imagination Technologies

Using geometric algebra ideas (specifically the even subalgebra of Cl3,1) to potentially help visualize the entanglement of qubits for quantum circuits. The pulling along an axis represents a Lorentz boost, while spinning around the axis are rotations.

Bilingual mathematician who took an art-and-writing detour on her way to philosophy

Coder making projects and YouTube videos about deep learning, fractals, games, math, and more

Hyperbolica is a whimsical Non-Euclidean adventure with mind-bending worlds full of games, puzzles, mazes, and secrets! Immerse yourself in reality-warping geometries where lines can never be parallel, horizons are curved, and space grows exponentially.

Freelance artist melding art, science, mathematics and technology

Chigozie Nri
Maker of tools for production of AI Art

In the past 2 years, AIs have been developed that can take a string of text and generate a corresponding image. New models and techniques are being released constantly. An open source community has formed, initiated by the work of Ryan Murdock (@advadnoun) and Katherine Crowson (@RiversHaveWings) (reference here). Just as 2 Cartesian coordinates can uniquely describe a point on a plane, or 3 coordinates can describe a point in real space, we can use many-dimensional coordinates to uniquely describe a whole image, or the semantic content of a text string. By training an AI model on millions of pairs of images and captions, we can discover a space and a mapping where similar images cluster close together, and the text string "Baroque Autumn" maps to the same region as a picture of a "Baroque Autumn". In this latent space, proximity scores between text and image can be calculated, and used to guide the generation of an image to match a new text prompt given by the human.

By taking an image generated from this technique, warping it, and using it as an input for the next frame, we can generate animations.

If the prompts used are changed on each frame, the image can evolve.

And if we scroll the original image in to the right side of the frame as it leaves, we can approximate a loop.

In this piece (released to coincide with this conference), parallax is achieved by scrolling independent generated layers at different speeds, a technique commonly used in 2D side-scrolling video games. The endless looping is achieved by inpainting the seams so that they match up, locating a point in the space of possible images that tiles seamlessly. The result is an animation where the last frame perfectly approaches the first frame.

A curious and creative geek making open source web & multimedia tools for learning, creative expression and play

Hyperspectives is a a collection of experiments with non-euclidean perspectives, related to moebius transformations and interesting mappings of 3d geometry. The goal is to offer unique perspectives on various aspects of what it means to be a creature living in space and to learn about math in the process.

The hyperhorse, in particular, is inverse stereographically projected onto the hypersphere, rotated, and projected back again. Click on the image to access the original interactive art, where you can move the camera, change rotation speed and pause the horse displacement.

                                    Octopus                                                                                                                              Quilting