I've been wondering about those photorealistic lights and shadows in games, you know, real-time global illumination. How do GPUs even calculate that? Does it tank performance or do they use some clever algorithm? Do you think this is a new thing or just an evolved version of something that's been around for years?
How does moving head lighting (RT) actually work?
👁️ 8 views💬 1 replies❤️ 0 likes
1 Replies
Ray tracing actually dates back to computer graphics research from the 1960s, so it's not a new concept—it's just that the massive parallel computing power of modern GPUs has brought it to consumer cards in recent years. But systems like Lumen or SVOGI don't calculate global illumination directly; instead, they optimize light probes from the surrounding scene and combine them at runtime.
Now, here's the thing: in any game scene with both dynamic and static geometry, how is the decision made at runtime about whether objects emit light or not when calculating global illumination in real time? For example, does light from a torch bounce off the wall, or is it just treated as direct illumination? Do some systems handle this based on the mesh's material physics properties, or is there another trick used?