For my project, I only have one simple goal: Create as many procedural-generated star systems with planets and other miscellaneous objects as possible.
Obviously, I want a solution where it doesn’t take tons of computing power to the point where the program is completely unplayable (this, I expect to struggle with as my guess here is just the tons of “fine-tuning” I’ll have to do). But, I want to make it as interactive and explorable as my imagination allows for it to be.
My first problem of many that I’ll run into is the planet creation. First, is every star system I’ll create a different level I’ll have to physically create within Unreal Engine? To be honest, I’m not really sure I could create one level and call it good. I’m sure I’ll have to do some level streaming and then work within each environment separately and tie it in all at the end. However, I’m curious to see if this is how you guys would go about it or if you have a better way? Secondly, how would I go about my issue of creating planets within that Star System? Remember that each planet will be visitable and all planets will be different as to what they offer. I want a player to be able have the ability to get into a spacecraft, physically leave the planets surface and through the atmosphere into space without any sort of “middleman” like an animation, and vice versa. Player needs to be in full control at all times.
For this first issue, I quite literally created a sphere object and hoped that Unreal Engine would recognize this as a potential landscape. I was wrong. Secondly, I’m not sure procedural generation of this planet would work off of a static object created in the manner I created the sphere: modeling tab > sphere > accept. This idea would work if I was going to create a planet no one would ever get close to or would work if this planet was just for show. It quite literally is the way I’ve created some space scenes.
I came across a video on the internet (although in Unity) where a procedural generator is created, then creates and defines LODs, based on player position relative to the object (or any object in that matter), displays the appropriate LOD and so on. It referenced quadtrees and various other terms I had not ever heard, so this was more or less what I believe to be a step in the right direction but a very unknown step. My question (more or less) is will this, what I believe to be the step in the right direction work? And if so, what would that look like (as an example) code-wise?
One area where my research lacks is keywords and the ability to spot good information. So I am hoping by posting here, it will lead me further into the right direction and hopefully, with a good looking game to come out of it after dev finishes.