Hacker News Reader

Comment by hcarvalhoalves | Hacker News Reader

I have a (possibly crazy) theory:

Companies like Valve and Epic have already amortized the cost of development of their engines - those are codebases that span 16 years, and both companies had multiple commercial successes. All it took was Unity to give a little nudge.

Now, IMO, the reason for giving so much for free is to keep a player out (Unity) and milk the cow one last time with the eventual royalties.

I think it's clear by now the next generation will be real-time raytracing (engines like Brigade [1]), so I wouldn't doubt they are already working on those, and to be sold in the old model.

[1] http://brigade.otoy.com

View on HN

Real Time Ray tracing is still very far off. Brigade serves as a good proof of concept, but a lot of things are yet to be researched and implemented to make it viable in commercial sphere. Seeing the progress, I do not think commercial real time ray traced game will be a reality any time soon in this decade.

It's been "next gen" for decades. It's just a terribly inefficient brute-force approach to lighting a scene. I don't think it'll ever be viable for real-time graphics.

Even if it'd be viable, the traditional approach is much faster and easier on resources. You still want to retain a part of the GPU for physics, maybe. Or run on lower-end hardware, too. Using ray-tracing for accurate shadows or reflections in part of a scene may be viable, but I severely doubt games will just use ray-tracing for rendering. CAD and architecture stuff may benefit more from that, I guess.

It will probably need GPU's to at least slowly into ray/path tracing engines, like what Imagination is doing. I think we'll probably get there within 10 years. 100+ TeraFlops should do it.

You are probably right it seems far off now.

But I'm sure ray-traced engine is the killer-app of the Oculus Rift, and companies like Valve and Epic are aware of that gold rush. Once people see ray-tracing on VR they are not going back.

I doubt that raytracing is going to ever jump the gap from 'next gen' to 'actually used'. As others have commented, it's been the mooted 'next big thing' for decades. The problem is that raytracing doesn't bring any great improvements to graphics. Even the Brigade engine demo that you link looks worse than the current engines.

(Although, bonus points to Brigade for not featuring a mirrored sphere in their demo. Every other raytracing demo that I have ever seen adds these in, as if what we've all been missing from our 3d is mirrored spheres! It's a sure sign of desperation and lack of imagination when raytracers resort to those :)

The demo looks like the world is amazingly detailed, but like you have crappy eyes.

It will when the resources used to emulate proper lighting gets close to the resources required by a ray tracer. That will happen if we continue to improve graphics.

Maybe we'll stop improving graphics sometime pretty soon and focus on physics, or maybe the game engine vendors will focus on licensing out assets to make game development cheaper instead of the current graphics race. But as long as the best use of GPU power we can come up with is graphics, we will eventually arrive at ray tracing.

> Companies like Valve and Epic have already amortized the cost of development of their engines - those are codebases that span 16 years, and both companies had multiple commercial successes. All it took was Unity to give a little nudge.

That doesn't seem to be the case. It seems that UE4 was a pretty major rewrite, and major development is ongoing: https://trello.com/b/gHooNW9I/ue4-roadmap . Source 2 looks to have been a big effort for Valve too.

Even if they refactor, the point I make is there's 16+ years of know-how, and the fundamentals are still the same (compared to something completely different written from scratch like a ray-tracing engine).

BSP, meshes + LOD, lightmaps, occlusion, particle system, terrain editor (since UT 2003)... it's still all there.

There's probably a bit of the Photoshop Effect going on, where companies are mostly interested in having their tool become the dominant one - and the dominant skill - in the industry over piracy litigation.

We're in a perfect storm of access to development tools and interest in indie development thanks in large part to Steam and Kickstarter.

In the case of Valve, I think what butters their bread isn't Steam anyway, but the games people end up making with it that Valve get a cut from through Steam.

Asset-store revenues are a big part of that as well.

I keep forgetting that exists. Everyone's talking about hats, and I have no idea what's going on.

Steam is both the butter and the bread so to speak. More games = more revenue. I don't know why they didn't do this sooner.

Reply on news.ycombinator.com