Here’s a bold statement: AMD’s FSR 4 Redstone might just be the game-changer the graphics world has been waiting for—but it’s not without its controversies. Announced earlier this year at Computex as Project Redstone, this latest iteration of AMD’s FidelityFX Super Resolution (FSR) technology is finally here, and it’s packed with features designed to close the gap with NVIDIA’s DLSS 4. But here’s where it gets controversial: while AMD has made significant strides in image quality and performance, the new features are exclusively available on the Radeon RX 9000-series GPUs, leaving older hardware in the dust. Is this a fair trade-off for cutting-edge innovation, or a missed opportunity for broader accessibility? Let’s dive in.
FSR 4 Redstone expands the FSR ecosystem by introducing machine learning-driven rendering stages, including frame generation, ray tracing reconstruction, and lighting acceleration (Radiance Caching). These additions are part of AMD’s strategy to compete with NVIDIA’s RTX 50-series GPUs, which debuted DLSS 4 earlier this year. The standout feature? A new ML-based upscaler that significantly improves image quality, addressing one of the biggest criticisms of earlier FSR versions. But this is the part most people miss: Radiance Caching, while promising, won’t be available for game integrations until 2026, leaving us to wonder about its real-world impact today.
The three core features of Redstone are:
1. FSR Frame Generation: Uses ML-based frame interpolation to boost performance without sacrificing visual fidelity.
2. FSR Ray Regeneration: A neural network-based denoiser that enhances ray tracing quality, though it’s limited to RDNA 4 hardware—a point that’s sure to spark debate.
3. FSR Radiance Caching: A neural radiance cache designed to reduce the computational cost of ray-traced global illumination, but as mentioned, it’s not testable yet.
From a technical standpoint, Redstone’s ML features rely on the dedicated AI hardware units in the Radeon RX 9000-series GPUs. This means older RDNA architectures are left with shader-based fallbacks, which lack the performance and quality improvements of the AI-driven models. Here’s a breakdown of feature availability:
| Feature | RDNA 4 | Pre-RDNA 4 |
|--------------------|-----------------|------------------|
| Upscaling | ML FSR 4 | FSR 3.1 Fallback |
| Frame Generation | ML FSR 4 | FSR 3.1 Fallback |
| Ray Regeneration | ML | No |
| Radiance Caching | ML | No |
For game developers, implementing Redstone is straightforward: it can be integrated directly into the game’s settings or managed via the AMD Radeon Settings control panel, which automatically overrides older FSR versions. This means RX 9000-series users get FSR 4, while RDNA 2 and 3 users are stuck with FSR 3.1—a decision that’s sure to divide opinions.
In the rest of this article, we’ll break down the technical details of each Redstone component, explore their deployment conditions, and analyze their impact on performance and image stability. But before we go, here’s a thought-provoking question: Is AMD’s focus on RDNA 4 hardware a necessary step for innovation, or a barrier to inclusivity? Let us know your thoughts in the comments—this is one debate you won’t want to miss.
