DLSS, FSR, and XeSS are very nice technologies. Many people associate them with “fake frames,” but this isn’t really true. All the frames in upscaling techniques are real, they’re just rendered at a lower resolution to reduce the load on your GPU and then upscaled with AI. These technologies have improved a lot, and they work well in many games.
But I didn’t expect Nvidia to go this far with DLSS 5. Don’t get me wrong, I was looking forward to neural rendering, but I always thought it would be a separate technology, with textures, assets, and models made specifically for it. Even so, in the tech demos, the textures look really good, but the faces look like they’re coming from an AI model that generates pictures, and I don’t like that.
If DLSS4 or even 5 being added without Neural Rendering, and then add Ray Reconstruction with Ray-tracing or Path-tracing, you could already make insane realistic scene's, so you don't really need Neural Rendering to be honest.