Apex Neural Rendering Challenges NVIDIA DLSS With Up To 2.4X Faster Frame Generation, 62% Lower VRAM Usage, And Broad GPU Support

Key Points:

• Apex Compute has launched Apex Neural Rendering (ANR), a cross-platform neural rendering suite that claims up to 2.4X faster frame generation, better image quality, and significantly lower GPU memory usage compared to NVIDIA DLSS.
• ANR currently offers a beta version of its Frame Generation (FG) component, with Super Resolution (SR) and Ray Reconstruction (RR) still in development, supporting a wide range of hardware including older NVIDIA, AMD, and Intel GPUs.
• Performance tests on a GeForce RTX 5090 show ANR FG generating frames faster than NVIDIA DLSS FG, while also using about 62% less GPU memory at 4K resolution, though real-world FPS improvements may be more modest.
• Apex highlights improved image quality in challenging visual scenarios and has released video showcases and quality metrics favoring ANR over NVIDIA DLSS FG, though independent verification is pending.
• The ANR Frame Generation beta is available for download on GitHub, compatible with Windows, Linux, SteamOS, and macOS, but the multi-frame generation code is not yet open-source and users should expect potential compatibility issues and artifacts.