Gamers react with overwhelming disgust to DLSS 5’s generative AI glow-ups (arstechnica.com)

Nvidia's recent announcement of DLSS 5, set to launch this autumn, marks a significant evolution in machine-learning upscaling for gaming. This new iteration moves beyond traditional upscaling techniques to incorporate generative AI for real-time lighting and texture enhancements, aiming to achieve a level of visual realism previously seen only in Hollywood effects. CEO Jensen Huang emphasizes that DLSS 5 combines generative AI with "handcrafted rendering," promising a dramatic leap in photorealism while allowing artists to maintain creative control. However, the gaming community's response has been predominantly negative, describing the generative AI enhancements as creating a bland and uncanny aesthetic. Unlike existing generative video models, which can lack consistency and predictability, DLSS 5 utilizes the internal color and motion vectors of games to produce photorealistic lighting and materials that remain coherent within the 3D environment. Despite accolades from outlets like Digital Foundry, which noted "transformational lighting" effects achieved using powerful RTX 5090 hardware, the overwhelming sentiment among gamers is one of dissatisfaction, prompting concerns about the future of artistic direction in gaming graphics.