AMD’s Rendering Ambitions: From FSR Redstone to FSR Diamond
Not long ago, AMD’s FidelityFX Super Resolution (FSR) Redstone felt like a potential turning point, a suite of technologies capable of challenging Nvidia’s dominance in AI-driven upscaling. However, before Redstone could fully establish itself, AMD introduced FSR Diamond, tied to next-generation Xbox hardware, leaving many wondering if Team Red is shifting focus before fully delivering on its current promises.
FSR Redstone: A Promising Start That Stumbled
FSR Redstone wasn’t merely an incremental update; it aimed to be a full-stack, machine learning-driven rendering pipeline, encompassing AI-based super-resolution, frame generation, Ray Regeneration (AMD’s equivalent to Nvidia’s Ray Reconstruction) and radiance caching for path-traced lighting. This represented AMD’s attempt to match Nvidia feature-for-feature and build a cohesive ecosystem for developers.
However, the launch felt premature. Redstone initially had limited support, primarily locked behind RDNA 4 graphics cards, requiring significant tinkering to function on older RDNA 3 cards. Feature rollout was fragmented, with radiance caching taking time to appear in real-world applications. Frame generation, a feature Nvidia had refined over two years, suffered from immersion-breaking frame pacing issues on AMD’s implementation.
FSR Diamond: A Glimpse into the Future, But at What Cost?
Positioned as part of next-generation Xbox hardware initiatives, FSR Diamond promises deeper integration with future GPU architectures, advanced frame generation, and a unified feature stack. While exciting, its announcement overshadowed Redstone, raising concerns that AMD is moving on before fully realizing the potential of its current technology.
FSR Diamond will be locked to RDNA 5 architecture cards, owing to their ML and AI cores.
Crimson Desert: A Telling Comparison
The game Crimson Desert provides a clear demonstration of the current gap between AMD’s and Nvidia’s upscaling technologies. In 4K, Redstone’s upscaling struggles with image stability, exhibiting softness and instability in foliage, geometry, and textures, even with sharpening applied. In contrast, Nvidia’s DLSS 4.5 reconstructs detail with greater confidence, maintaining crisp edges and stable motion.
A Question of Follow-Through
AMD’s vision for rendering technology is clear, as evidenced by both Redstone and Diamond. However, a consistent issue is follow-through. Redstone launched with incomplete support, fragmented features, and limited adoption. The shift to Diamond before Redstone is fully realized risks repeating this pattern.
While FSR Redstone isn’t a lost cause, its success hinges on AMD fully committing to its development through wider support and complete feature delivery. Without that commitment, it risks becoming another stepping stone rather than a destination.