Carlos Rodríguez-pardo

Title: Carlos Rodríguez-Pardo and Neural Synthesis of Tileable Textures.

Introduction

Carlos Rodríguez-Pardo is an inventor based in Madrid, Spain, associated with Seddi, Inc. His work includes an invention applying artificial intelligence to the synthesis of seamless, high-resolution textures.

Latest Patents

Rodríguez-Pardo is credited with 1 patent, titled “Neural Synthesis of Tileable Textures.” The invention describes an artificial intelligence framework that transforms non-tileable texture inputs, including single- or multi-layer texture stacks, into high-quality tileable textures that preserve the appearance of the samples while increasing spatial resolution.

The approach uses a generative network and a latent-space method to create textures with semantic consistency across tile boundaries. A tileability metric provides feedback to improve the output, while a sampling algorithm supports the generation of high-resolution textures with fewer artifacts. A convolutional discriminator locally evaluates synthesized maps for artifacts and can also help optimize the selection of input samples.

Career Highlights

A key highlight of Rodríguez-Pardo’s listed patent work is its focus on practical texture generation: converting existing samples into seamless, higher-resolution outputs while addressing boundary consistency and visual artifacts. He is associated with Seddi, Inc.

Collaborations

Rodríguez-Pardo’s listed coworker is Elena Garcés. The patent information provided describes the invention’s technical approach but does not specify individual contributions by either coworker.

Conclusion

Carlos Rodríguez-Pardo’s patent presents an AI-based approach to generating seamless, high-resolution textures, combining generative modeling with feedback mechanisms for tileability and artifact detection.