Chaosforge Forum
Jupiter Hell => General Discussion => Topic started by: hawk66 on March 31, 2016, 02:40
-
This project looks really cool and will be a milestone in the roguelike genre for sure.
My understanding is that you develop your own game engine. Would be interesting to know if you have considered using engines like Unity3D and why you have not chosen one of them as a basis ?
Thanks.
-
I kinda expect this to be the most frequently asked question as we come nearer and into the Kickstarter, so I'll try to do a detailed write-up as we come near.
However, here's the short version:
- Unity sucks in the performance department, if we'd use an existing engine it'd be Unreal or Stingray anyway.
- most engines suck if you're doing dynamical worlds (randomly generated) -- 90% of their power is precalculating stuff, and if you do random environments, you cannot precalculate anything
- I like cool dynamic lighting effects, which on a dynamically constructed environment are just too much for Unity to handle
- while Jupiter Hell could probably get away with Unreal, each next game that I want to make will be more and more procedural, so the overhead will be bigger and bigger. At some point it will be too much to continue (you couldn't do a game like No Mans Sky or Elite Dangerous in any existing engine), and at that point we could either give up or spend 5 years doing an engine from scratch. Nova (the JH engine) is written with zero precalc rule. Everything is dynamic.
However, this is a much simplified list, there are many more small reasons, and the larger reasons would probably need more elaboration. Feel free to ask further questions, or wait for the big article :P.
-
Ok, all right...looking forward to your article and the kickstarter campaign :)
-
I kinda expect this to be the most frequently asked question as we come nearer and into the Kickstarter, so I'll try to do a detailed write-up as we come near.
However, here's the short version:
- Unity sucks in the performance department, if we'd use an existing engine it'd be Unreal or Stingray anyway.
- most engines suck if you're doing dynamical worlds (randomly generated) -- 90% of their power is precalculating stuff, and if you do random environments, you cannot precalculate anything
- I like cool dynamic lighting effects, which on a dynamically constructed environment are just too much for Unity to handle
- while Jupiter Hell could probably get away with Unreal, each next game that I want to make will be more and more procedural, so the overhead will be bigger and bigger. At some point it will be too much to continue (you couldn't do a game like No Mans Sky or Elite Dangerous in any existing engine), and at that point we could either give up or spend 5 years doing an engine from scratch. Nova (the JH engine) is written with zero precalc rule. Everything is dynamic.
However, this is a much simplified list, there are many more small reasons, and the larger reasons would probably need more elaboration. Feel free to ask further questions, or wait for the big article :P.
Very interesting Kornel, looking forward to hearing more details soon.
-
Hello, this is my first time on here, may I ask if you ever plan on making the game engine available to other game developers?
Thank you.
-
It's already out there....but there aren't any stable branches yet. I'm KK will make a distribution release when the time is right.
KK Edit: Removed the source link, because at this stage for anyone using it it would cause more problems than it would help
-
The question is probably about a whole engine, not a low-level engine library :P.
As MaiZure observed, I'm pretty much opening the sources of the base libraries (they don't have a license agreement yet, but probably MIT). However, this is not the only layer that JH is built on. Also, even when the game is completed, and I release the modding tools to the community, by no means will Nova (the underlying engine) be able to compete with existing engines in terms of documentation, upgrade stability, examples, tools, etc.
To get the engine to such level requires a lot more time, actually a lot more additional time than making the game and engine itself, not to mention maintaining version compatibility and support. Time which I don't intend to invest -- because it would much better be spent developing the current, or next game, right? :)
So, the JH engine might only be of interest to people that already posses the dedication and skill needed to write their own, and in no way would be a substitute for current commercial engines (e.g. Unity/Unreal).
-
Thank you!
I'm currently only starting to learn how to code, so I'm sure it would be way out of my experience anyway! It just sounds really cool for what it can do, after No Man's Sky proved what procedural games can (potentially) do, I'm sure it would be something a lot of people are interested in. I'm really looking forward to Jupiter Hell, thanks!
-
it would much better be spent developing the current, or next game, right? :)
Hells yes.
Anytime I see "Unity" on any game my heart sinks.
Loved your list above on why your own engine (I know you're not going to rewrite libraries you don't have to). No need for a long article, just more time spent making:
!!!!!!! JUPITER HELL !!!!!!!!!!!
-
No Man's Sky proved what procedural games can (potentially) do
That procedurally generated games can be absolutely awful has been proven far before NMS ;) One example would be Spore, but there are several others.
-
That procedurally generated games can be absolutely awful has been proven far before NMS ;) One example would be Spore, but there are several others.
For something to be truly bad it needs to have had the potential to be truly good.