From Pixels to DNA: Why the Future of Compression Is About Every Kind of Data

Data compression has evolved far beyond audio and video into a foundational technology spanning images, genomes, neural networks, and 3D scenes. JPEG's portfolio now includes JPEG AI (latent-space image coding), JPEG Trust (provenance and deepfake detection), JPEG Pleno (light fields and holograms), JPEG XS (ultra-low-latency video), and JPEG DNA (biological storage). On the MPEG side, the next-gen H.267 codec targets ~40% bitrate reduction over VVC with a projected 2028 finalization. Neural Network Video Coding (NNVC) is delivering 6–14% BD-rate gains over VVC but at decoder complexity 14–118x higher. New standards address video and feature coding for machines (VCM/FCM), enabling privacy-preserving, bandwidth-efficient pipelines for autonomous vehicles and smart cities. Neural Network Coding (NNC) can compress AI models by up to 97% without accuracy loss. Gaussian Splatting and point cloud compression are advancing for immersive 3D applications, while MPEG-H Audio adds object-based, personalized, and dialogue-enhanced experiences. Energy efficiency and reproducibility are becoming first-class requirements across all codec development.