IT之家 November 12 news: NVIDIA's DLSS technology has gained a new competitor. Apex Neural Rendering (ANR), an AI rendering technology suite developed by Apex Compute, has officially launched its beta version, featuring AI frame generation, super resolution, and ray reconstruction capabilities.
Unlike NVIDIA's DLSS, which mainly targets its own graphics cards, ANR claims compatibility with multiple generations of NVIDIA, AMD, and Intel graphics cards, and even supports some older models.
Among these, ANR Frame Generation (ANR FG) supports NVIDIA GeForce GTX 16 series and RTX 20 series or newer graphics cards, AMD Radeon RX 5000 series or newer graphics cards, as well as Intel Arc A series and B series graphics cards. In addition, Valve's Steam Deck handheld is also within the supported range.
On the software side, the currently released beta supports Windows 10/11 and Linux, and can also run on SteamOS through Proton.
According to IT之家, Apex Neural Rendering consists of three technologies: frame generation, super resolution upscaling, and ray reconstruction. However, as of now, only the frame generation feature is open for testing; the other two technologies are expected to be released gradually later.
Apex Compute has also released a set of performance data. According to the company's tests, at 4K resolution with an NVIDIA GeForce RTX 5090 graphics card, ANR in 2x mode takes only 1.01 milliseconds to generate a frame, while NVIDIA DLSS frame generation takes 1.26 milliseconds. In 4x mode, the time drops to 0.61 milliseconds and 0.82 milliseconds respectively.
In terms of VRAM usage, ANR reportedly requires only 214 MB, while DLSS needs 562 MB.
If these figures are accurate, ANR could potentially solve a major pain point of frame generation technology: the added latency. While frame generation can increase displayed frame rates, the processing itself may introduce additional latency, affecting input responsiveness.
Beyond performance and VRAM usage, Apex Compute also claims that ANR can deliver better image quality, and has published multiple sets of image comparisons, including game scenes from The Last of Us Part II.
However, these figures and image comparisons all come from the developer itself and cannot yet serve as proof that ANR comprehensively surpasses DLSS. The final performance still requires independent testing, especially verification in real-world gaming environments. After all, faster frame generation does not necessarily mean better overall game performance, nor does it necessarily reduce the input latency actually perceived by players.
The ANR FG beta is currently available for download on GitHub, and at this stage it needs to rely on the OptiScaler tool to run. The installation process requires modifying each game's individual directory, and is not as simple as adjusting a few sliders in game settings. Therefore, compatibility may vary considerably between different games.
For players who want to experience AI frame generation technology on graphics cards from different brands, ANR offers a new option. However, whether it can truly become a strong competitor to DLSS still depends on the release of the other two technologies and more independent test results.