It began quite some time ago as one developer’s (Flavien Brebion) part-time project, which steadily grew with time. No existing engine had at this time the capabilities (chiefly, full-scale planet and galaxy generator) he wanted, so he developed his own engine.
It is still true today, though the option of heavily modifying an existing project may be a more viable solution than before - as with what KSP is doing with Unity at a lesser scale.
So it could have been a great plugin for Unity if they began to work now - or it could still be simpler to make a new engine from scratch, I honestly don’t know. That’s probably moot, though, as once one engine with such capabilities hits the market, why would someone make another?
It is underfunded because now there is a team behind it (developers, artists…) and they haven’t found investors wanting to fund such “high risk, high reward” project.
As such, they are going for a Kickstarter campaign in the future, for a multiplayer Newtonian space shooter game featuring an entire full-scale seamless procedural star system and a first commercial version of the engine (and most probably mod tools to work with it).
This should make for several interesting things :
We should have a fun space shooter in a far too empty niche, Newtonian space shooters - trust me, once you taste a good one, it’s hard to come back to Space Friction shooters. Full-scale seamless star system and vast (compared to the space fighters) ships and stations are obviously a great plus as well.
We should have an engine to muck around with full-scale star system generation, seamless transitions, detailed planetary surfaces and an adapted physics engine. While few of us may do more than muck around it, some great mods and projects may be born from it.
They show that they can delver a working, fun videogame to potential investors and backers (thus dampening the “high risk” from “high risk, high reward”), as well as building experience for more complex projects - like the MMO originally envisioned.
They show a first working commercial version of the engine, available for those who want to use it (think what Unity does) and potentially attract more investors, and hopefully gain enough from sales to expand their team and work more features in the engine.
Because back then, the option didn’t really exist. The I-Novae/Infinity project goes back to at least 2006, and by now, Mr. Brebion, and the team he’s gathered around him since then, is so close to the finish line that dropping the work that’s been done and making a Unity plugin would be counter-productive.
It uses double precision floating point variables to store your location. This gives sub-millimetre accuracy out to distances of roughly 30 AU, or what is commonly known as “the orbit of Neptune”. For the record, equally sized integer values would give millimetre precision out to distances of just under +/-1 lightyear, but be more costly to work with (in terms of both clock cycles and development time).
The developers want rather a lot more, and either way adapting Unity for 64 bit positions (absolutely necessary to avoid KSP-like jumping about whenever you enter/exit timewarp and gravity wells) would likely have been more work than simple writing a new engine from scratch.
For comparison, if INovae had the funds available that Chris Roberts and SC will need to use to try and get CryEngine to work with 64 bit locations, Infinity would already be done.