I saw a talk about Lumberyard once. The really special things they wanted to do were integrate it tightly with AWS and Twitch, which enabled some neat features like letting people spectate games and move their own camera around the field to do so.
Q. Can my game use an alternate web service instead of AWS?
No. By “alternate web service” we mean any non-AWS web service that is similar to or can act as a replacement for Amazon EC2, Amazon Lambda, Amazon DynamoDB, Amazon RDS, Amazon S3, Amazon EBS, Amazon EC2 Container Service, or Amazon GameLift. You can use hardware you own and operate for your game servers.
I know absolutely nothing of game dev, but adding cameras to the scene doesn't sound like it would be the most complex aspect of developping a video game?
Imagine a Twitch livestream with tens of thousands of viewers, and each one is able to independently move a live camera to get the view they want. The complexity comes from scale.
You could very well tie that to a high-tier subscription to any Twitch channel. Depending on the game (e.g. Fortnite) you could easily server a bunch of viewers from the same GPU.
I mean, I could probably figure out how to do that in Unity in about a week.
You would basically just need to stream the cameras output texture to a web app, and then read inputs from from said web app. Regardless there's no reason to try and create a whole new engine for this, CryTek has always been known as one of the worst engines to work in, just because it's not supported very well.
Yes, that would be the low-tech way to achieve this, but if you want to scale that up to Twitch subscriber level of users, you would certainly want engine support for such a feature. That CryEngine was a bad choice is a valid point, though I guess there weren't that many public engines to choose from at the time that were that desperate for money.
Yes and no, if Amazon really wanted to do this they could have bought a source license from Unity and made whatever modifications they needed. It feels like an absurdly niche feature to focus on though.
The way I would personally do it I would maybe create a multiplayer game and for each ten cameras spin up a game client instance.
According to the other comments here it looks like Amazon wanted to create some type of bizarro system where if you use their engine you had to use AWS as your cloud back end. Which does sound pretty horrible, like I can't even do a HTTP request to get weather data unless it goes through AWS ?
I disagree with this a lot. Building your own engine offers many advantages, especially for a company with ambitions like Amazon. If anyone should build their own engine, it's them.
Sure, they didn't release any (good) games, but I don't think the situation would be much different if they'd have used Unreal or Unity instead. Some of the complaints from devs were that Lumberyard was hard to work with, but that's a common thing to hear from new engines. Devs working on Metal Gear Solid V said the same thing about their custom Fox Engine, yet MGSV is a spectacular game that won a ton of GOTY awards.
I used LY for a game jam a few years back, and it was pretty painful to work with. They have some weird custom message passing system for components to communicate with each other safely, which seemed to me like the type of over-engineered, almost enterprise-like system a tech giant like Amazon would make. That sucked. It also has two physics engines: a "deprecated" one from CryEngine and Nvidia PhysX which wasn't actually fully supported yet.
Development was a pain in the ass, and while I did finish the game in time for the game jam, I couldn't get it to produce a shippable build of the game in time (each build took over an hour before failing with an unhelpful error)
However, if they iron out all of the bugs, and clean up/finish all of the incomplete stuff, and I get used to their message passing system, then LY is a really enticing value proposition. The engine is 100% free, with the only catch being that any backend stuff needs to be hosted on AWS or self-hosted. And while that could potentially be a deal-breaker for multiplayer games, it's a no-brainer for many others. Having an option like that in addition to Unity's up-front license payment and Unreal's 5% would be great for everyone, even if Amazon can't make a single decent game.
> I never understood why Amazon rolled another engine.
For the same synergistic effects as Epic and their game store + their engine. If you use epic's engine, they charge you less to sell your games. It's basically trying to incentivize selling on their platform and using their platform tools at the same time.
Can you imagine spending tens of millions on a game then getting blocked from AWS on a policy violation and not being able to move it to another provider?
I can’t imagine any sane AAA game dev agreeing to those demands.
The license (last I checked anyways) also lets you self-host your servers. If you can afford the upfront costs for that, it's probably cheaper in the long run. IIRC, part of the motivation for Roblox's IPO was to raise funds to setup their own hosting infrastructure to lower costs, because AWS was too expensive.
And if you're making a single-player game, or a p2p multiplayer game, then you'd only need AWS for matchmaking servers and maybe some other simple accounts/databases, or like S3 to distribute content files.
So while it is a stupid decision on paper to lock yourself into a single vendor, I think it could work out in a lot of cases. It might even be cheaper than Unreal, which takes a 5% cut, as long as Amazon doesn't decide to exploit their relationship with you.
and I'm sure there would be an incentive (financial) to using all amazon services/storefront etc.
The reason they started with lumberyard was because amazon does platform thinking. Unlike open source they can't just take it, so they bought cryengine to start with.
I'm currently building a game in HDRP and it's easily looks just as good as anything I could do in Unreal. I'm only limited by my own time at this point .
Unreal and Unity are both fantastic. Let your dev teams pick from one of those two and be done with it.
Unless your doing something really special theirs no reason to build an engine( if your goal is shipping games )