Flowfield Library

I was just watching the November 30th LiveStream and I actually cheered when they mentioned they'd be using Flowfield pathfinding for PA. Say what you will about SupCom2, that game had some AWESOME pathfinding.

He mentioned, specifically, that he had implemented flowfield as a library separate from the game. Is there any chance *crosses fingers* that this will eventually be released as an open source library?

I'm a bit of a hobbyist game developer and I've tried a couple strategy games, but I've always gotten stuck on the pathfinding. I'm not usually very good at grokking algorithms and implementing them in an optimized way, so I like to stick with premade libraries. There's lots of A* libraries out there, but they really only work for one or two units and things get janky once you try to make units move in formation, or even as a vaguely defined group. I'd love to get my hands on some flowfields! :mrgreen:

I'm curious on how it will work with round planets. If you have a pile of units and tell them to go 180 degrees across a planet, everyone is going to split up!

the units will probably recognize themselves as a group and just pick a direction

Correction, it had "better" pathfinding. The pathfinding had lots of issues with it that kinda wrecked it for me, I've already talked to Neutrino and the Flowfield Guy(Eric? Sorry I'm horrible with names) and they've said they're trying to avoid the issues it had in SupCom2, so now we just wait and see.

Mike

What issues are you thinking of? I don't really play at a competitive level, but SupCom2's pathing seems fine to me, aside from units getting stuck on sharp corners once in a while and occasionally brain-dead formations (let's put all the mobile shield gens at the back!).

While flow fields are cool from a pathfinding perspective, the way they were implemented in SC2 looked ... silly. Ships pushed each other around, expermentals moved everything out of the way etc. Everything ended up looking like they were skating & sliding around, and it lacked any appearance of units being "connected" to the ground.

While not perfect, it was IMO a big step ahead from the ultra-stupid units of Supcom1, where you would order an army to move and it would become a huge, huge mess that would be only halfway after 10 minutes of trying to pick a proper route.

Oh. That would explain it. I usually played as Aeon, whose units weren't connected to the ground anyways

Flowfield definitely represented a huge step forward, but it still needed a lot of work. Units turned and rotated in quick and unnatural ways. They'd all move in an unnaturally synchronized fashion when given a command, seemingly ignoring the terrain beneath them. They would skate and slide as if they were on ice. It worked great for gameplay, but it needs some work before it looks good.

That's when they actually made it to their destination. I've had a few cases of units trying to go around a mountain, and ending up stuck on the mountain. Since the terrain is classified as impassible to them, they can't get down. No idea how they actually got up there in the first place though.

Hopefully formations will be smarter too. Too many times I would see a group trying to get in formation with most of the units attempting to go to the farthest formation spot from their starting location.

The comment about it being a library is in reference to how our internal engine is structured, it's made of many modular features that can be included or not. Outland Games is the same base engine that is powering PA, but there are many features that are not needed so it could be stripped down to work in such a small game. There are no plans to release the flowfield pathing system as an open source library.

From a legal perspective, and not necessarily linked to the flowfiled stuff, how does it work when a developper has worked on a technology in company A, developped algorithms in company A and leave for Company B ? Can company A prevent him from re-using his code ? Does he have to re-write everything from scratch ?

Things generally get rewritten from scratch, yes. Sometimes code, especially for tools that aren't directly tied to proprietary technology used by that company, will get released or reused with permission from the original company.

For games though, often times even when at the same company, we'll choose to just rewrite something from scratch to build upon lessons learned during writing / using previous code.

Yes, you have to write everything from scratch. Some companies even try to prevent you from using the knowledge that you have to write new stuff.

In fact a lot of companies refuse to even use free open source code that exists out there because of legal entanglements.

There are also other risks with releasing your code as open source when you are a commercial company.

tl;dr business sucks

I wish the Industry would go back to make games because they are fun and not for money I understand why it has gone that way but i still think passion games are far more compelling to buy for a multitude of reason thats why i love seeing games pop up on Kickstarter and Indiegogo because they are passion games what people want to make! not what they have to make.

Yeah it would be nice if we didn't have to worry about money. We need a patron like the old days. Unfortunately the children have to be fed and clothed

Socialise the games industry!

Frikin' Commies!

I really liked the pathing in SupCom2, but after going back to check it out again, all the things I loved about it don't seem to work anymore... units matching speeds, shield generators staying in the middle of assault bots, artillery staying behind other units, etc.

Were there pathfinding changes made to make the game more competitive?

Although I'm working as a programmer for several years now, being a scientist I never encountered any such restrictions yet. Up to the point when no one actually knew what license were we writing in .

Can you enlighten me a bit on that? Did you sign NDA? Who is usually the most interested party in guarding the code? If my friend and I are working in different companies but on similar tasks and privately share a bunch of code to make both of our lives easier, are we guilty beyond forgiveness?

I'm thinking right now whether I should choose the path of a scientist or go to industry. A little bit of inside knowledge won't hurt .