Indeed.
I imagine Metroid Prime 4 could be a showcase for lighting as well.
Metroid Prime 4 will be an interesting one, assuming it is cross-gen and doesn't sneak in just before [redacted] comes out. It started as a Switch game, and even if it was re-targeted as a cross-gen game relatively early, it's still something they have to release on Switch. As much as it makes sense as a lighting showcase for [redacted], going all-out on lighting on the new console while also running and looking well on the original Switch won't be easy. As an example, let's say Retro come up with a cool real-time global illumination technique that allows for fancy dynamic lighting on the new hardware. They're not going to be able to use this technique on the original Switch, so they'll have to fall back to other techniques like baked GI there. This doesn't have to look as good as the fancy lighting on [redacted], but it has to look consistent with it, and that's a challenge. If a room is well-illuminated from indirect light on [redacted], it shouldn't be pitch black on Switch.
This becomes trickier the more you leverage [redacted] version's dynamic lighting for dramatic effect. Say you have a scene where you want to gradually let in light, say from a window or door slowly opening, as a dramatic reveal for whatever's inside. Fancy dynamic GI handles this well with few issues, but it'll be more work on the original Switch. You can't use static baked GI, because it won't respond to the change in illumination of the scene. If everything's scripted, then maybe you can bake lighting for specific keyframes and blend between them, or if you're using a probe-based system then maybe you could fade in probes in different parts of the room as you expect them to become illuminated. It's definitely doable, but doing it requires work, and doing it in a way that's consistent with the [redacted] version requires extra work.
A good recent example of this is Satisfactory, which is being upgraded from UE4 to UE5. They're going to allow people to toggle on Lumen (UE5's GI solution) as an "unofficial" option, but the devs have said that they're not going to change the lighting to account for it, because getting the lighting right both with and without Lumen would be too much work. They talk about it
here, and also show some screenshots of the game with and without Lumen, which I think are a pretty good illustration of how big of a difference real-time GI can make in lighting a scene.
Also, and this is an opinion thing so... idk feel free to disagree somewhere that isn't here, but I don't think the Switch 2 needs RTX. Like... cool beans if you do add it for select games, but it's far from a requirement. Literally having any game from the current 9th generation (that being literally any modern AAA game) run on the Switch 2 would be a good enough selling point.
I don't think ray tracing is strictly
necessary on the new hardware, but as the hardware's there I expect developers (particularly first party devs making games exclusively for the console) to make use of it.
I should emphasise that I don't think [redacted] games will be hitting people over the head with a big stick that says "Hey, look! Ray tracing!". In part because it's just 12 RT cores running at pretty modest clocks, but also because I don't see developers making heavy use of the more obvious use-cases that people associate with ray tracing today, in particular ray traced reflections. Ray traced reflections don't typically scale down well to lower-end hardware and also don't work well with temporal upscaling solutions like DLSS. You can reduce the resolution of your ray traced reflections, but they're already typically performed at sub-native resolutions, and you can only go so far before they're so blurry they're no longer worthwhile. They do scale down with the complexity of scene geometry, but taking a hacksaw to your geometry just so you can get ray traced reflections to work doesn't seem like a good trade-off.
We'll probably see RT reflections here and there where they make sense, but I'd expect more [redacted] devs will put those RT cores to work on lighting, and in particular dynamic GI like we see in Lumen. There are plenty of RT-based GI techniques which scale down well to lower-end hardware, and even a relatively lo-fi RTGI solution will be a dramatic improvement over existing (mostly baked) GI techniques. Perhaps more importantly, moving to fully dynamic GI can be very beneficial from a production standpoint. Rather than having to spend huge amounts of time faking and baking lighting, artists can simply build assets and let the real time lighting engine do its job, potentially being able to see the impact of changes immediately without having to wait for offline rendering. If you look back at Epic's announcement of UE5 with Lumen and Nanite, you'll notice that, although they definitely showed off how good games could look with them, a lot of what they were talking about was how much time and effort they would save for artists.
www.ea.com
