The more unaffordable technology gets, the more important game optimization becomes And we gotta give developers better tools for it.

Players discuss this most often when it comes to rendering? Like - the point of Nanite/Lumen IS NOT gigashit graphics for players. It's to smooth over a million iteration-blocking workflow issues... at a known runtime performance cost.

Battlefield 3 still looks fucking fantastic with baked lighting. The process of making it look like this (especially when it comes to lighting) was a hell of a lot less flexible, and required more people, than the same scene with the trademark Unreal Engine 5 workflow.

This is why I dig tools like Valve's Source 2 baked lighting solution, similar to Bakery RTPreview - Instead of forcing you to wait for progressively higher-resolution bakes, they just use realtime raytracing to show artists their adjustments instantly - then bake afterwards.

It moves arguably the biggest iteration cost back out of your artists' fuck-around-find-out loops, without changing the calculus for players. Hell yeah! But there are still 1000 other managerial problems to deal with when baking lighting...

Because lightmaps can still present nightmares, you still have to deal with probe placement, the process of baking assumes the lighting *won't change* which isn't always true and requires clever workarounds, lightmap memory is limited so once you get *outdoor* environments you-

And this is still *just lighting* we're talking about - we aren't even factoring in LODs! Which affect everything I've just mentioned! So yeah, if you promise devs a magic wand that like 80% solves all this for nearly 0% of the dev-time cost? That opens up real possibilities.

And it's the only reason we're getting this many games that *look as good as they do* now, made by handfuls of people. It's because Lumen/Nanite have freed them from these workflow impositions, they can apply that effort elsewhere. For all the problems those systems present.

Even if for people like me... I can't run those games well! If the performance traits of those systems aren't enough, and we don't want to sacrifice the aesthetic targets, then we NEED foundational tools to make the other paths more production-viable.