I wonder how Starforge people want to combine grid-based editing with a spherical body. It's a mathematical problem that doesn't have any nice solution. There are ways how to do it, but they aren't
nice not without singularities, seams and/or other plagues.
Basically both SF and OT (and other apps that are procedurally based and allow modifying the world at the same time) are trying to solve the problems of representing, rendering and manipulating vast worlds with today's computer resources. Each of the approaches has its weaknesses and advantages. Ultimately we'd all like to have a planet that can be completely altered and fast to render. But did you know that if you wanted to have just a 1 meter high fully modifiable space on the surface of the whole planet Earth, made of 0.1m (10cm) wide cubes, storing 1-byte information for each cube would take 464,940.6 TB of data. That's without any extra overhead needed if you actually wanted to use that data without searching through it all. Obviously, every algorithm out there counts with that you modify just a tiny part of the world, and the rest is procedural or plain. This keeps the memory within bounds while appearing as if the world was fully modifiable.
In OT the base engine works with the planet surface, and combines several layers of modifications of the spherical surface - from the global elevation dataset (~12GB of compressed data), with fractal generator, followed/altered with vector generator layer that allows you to modify the terrain locally (roads, craters; anything that allows the user to modify the world).
SF does these modifications in 3D space, which allows them to make tunnels and arbitrary structures in an easier way. Wrapping over a spherical surface is harder, and the compression ratio is orders worse than in OT. In OT laying down a kilometer of a road is recorded in less than 512 bytes of data; to reach a similar detail of the road SF would need much more.
However, there are ways how to get the best out of each, and even to combine the approaches, and even if it didn't work, each technology has its use as it is. Some things are easier done in one than in the other.
Could Outerra still run as smooth if instead of an empty earth, the entire planet would filled with buildings, props etc as seen in 2013? (Outerra feels like you are on the early days of planet earth because everything is clean). Could Outerra run a 1:1 scale of 2013 planet earth smoothly, allowing to enter on every single building, house, skyscraper, car... and all of these being destructible? that would be insane.
If you mean a 1:1 copy of the known Earth, then no. I can't even imagine the amount of data required to hold all the information there is, I mean - just look out of the window
If you mean a procedural planet that resembles Earth to some degree, that's something else. Just as the planet in OT resembles the Earth now by using the 90m elevation grid and inventing its own details further below that resolution, you could have cities generated from a rough shape that takes many orders of magnitude less data. Basically every representation of the world say in a flight simulator is such a form of a lossy compression, except that unlike in an image decompressor, a procedural geometry decompressor adds the missing detail using some math rules and patterns.
Usually it's enough for the world to be believable, but there always will be people dreaming about getting an exact copy of the world ... and destroy it
Author
Topic: Will total terrain modification be eventually possible? (Read 7714 times)