Pathfinding accuracy rewrite

Discussion of Maia's on going development, including modding.
User avatar
SimoRoth
Site Admin
Posts: 1106
Joined: Sat Jul 14, 2012 11:28 am

Pathfinding accuracy rewrite

Post by SimoRoth » Thu Apr 18, 2013 9:57 pm

Pathfinding was gank and crashing the game.

I've increased the accuracy by 4x and finally got it working ok.

Image

User avatar
Tikigod
Posts: 152
Joined: Thu Apr 04, 2013 9:29 pm

Re: Pathfinding accuracy rewrite

Post by Tikigod » Thu Apr 18, 2013 10:55 pm

Awesome news, only just started touching on learning on putting together a very primitive A* based pathfinding system but I can somewhat imagine how much more hassle it must be for a game like Maia where you have to come up with a pathfinding system where units can navigate quickly and give efficient results but also adapt to changes in environment such as what was 5 seconds ago a wall now being a open tile, or to be able to adapt to what was before a empty cavern now featuring an enclosed corridor with requests to build equipment appearing for areas to avoid walking through during construction....

Must be rather interesting to tackle but does make me wonder if the needs of the AI just for pathfinding logic directly relates to the comfortable range for the number of active units at any one time to a significant level? Or is the overall overhead not as much as one might expect?
Violence is the last refuge of the incompetent - Isaac Asimov

My channel. Game recordings & old 3D art assets I've made.

User avatar
SimoRoth
Site Admin
Posts: 1106
Joined: Sat Jul 14, 2012 11:28 am

Re: Pathfinding accuracy rewrite

Post by SimoRoth » Fri Apr 19, 2013 12:30 am

Thankfully, we are unlikely to have enough creatures/robots for it to really hit the framerate too hard.

A busy colony would spoil the bleakness of it all, so some of the harder, more expensive stuff we can bypass.

InfamousPotato
Posts: 2
Joined: Sun Apr 07, 2013 5:49 am

Re: Pathfinding accuracy rewrite

Post by InfamousPotato » Fri Apr 19, 2013 6:15 pm

SimoRoth wrote:Thankfully, we are unlikely to have enough creatures/robots for it to really hit the framerate too hard.

A busy colony would spoil the bleakness of it all, so some of the harder, more expensive stuff we can bypass.
Are there built in limitations to how big our colony can grow, or is it just highly unlikely that it'll become a busy colony?

User avatar
Tikigod
Posts: 152
Joined: Thu Apr 04, 2013 9:29 pm

Re: Pathfinding accuracy rewrite

Post by Tikigod » Fri Apr 19, 2013 7:49 pm

SimoRoth wrote:Thankfully, we are unlikely to have enough creatures/robots for it to really hit the framerate too hard.

A busy colony would spoil the bleakness of it all, so some of the harder, more expensive stuff we can bypass.
Hmmm, hope the colonies aren't too empty as it may eventually pretty annoying as you play further into the game if you only have a dozen I.M.Ps across the whole area and have to wait for them to get about as you build and give orders across multiple areas, such as populating a new room with a half dozen equipment build requests to make it operational whilst also laying out instructions to mine out a area for the next room.

Would also be a shame if map sizes end up being significantly reduced in size in order to cater to a reduced number of units having to travel around a lot to save on unit travel times.... can't exactly just speed up units to compensate for map sizes, as perceived size is partially relative to travel speed... less so in a god game true, but if units are fast enough to cover large tile distances quickly it will make the game feel rather small.

All this pondering is making me even more eager to see how it's all currently balanced now. Kind of hoping that pathfinding was the main reason the build was pulled from the members section and now that it's sorted it means a new build will now be uploaded for the weekend. :D
Last edited by Tikigod on Fri Apr 19, 2013 7:59 pm, edited 1 time in total.
Violence is the last refuge of the incompetent - Isaac Asimov

My channel. Game recordings & old 3D art assets I've made.

User avatar
SimoRoth
Site Admin
Posts: 1106
Joined: Sat Jul 14, 2012 11:28 am

Re: Pathfinding accuracy rewrite

Post by SimoRoth » Fri Apr 19, 2013 7:58 pm

Worlds are currently all 1km-2km. The final game will be 2km.

The game certainly wont be empty, but there will be a limit. It wont end up looking like a game of roller coaster tycoon though. :)

As for the limits, I don't know. I guess in a mature colony you might have ~30 colonists and ~16 imps. I'll have to see how the native fauna, chickens and other things do at eating resources.

I want the loss of a colonist to be a major event and a major set back.

User avatar
Tikigod
Posts: 152
Joined: Thu Apr 04, 2013 9:29 pm

Re: Pathfinding accuracy rewrite

Post by Tikigod » Fri Apr 19, 2013 8:12 pm

SimoRoth wrote:Worlds are currently all 1km-2km. The final game will be 2km.
Hard to get a idea on how 'large' that would effectively be in the game itself without seeing unit speeds, as if units speeds mean they cover the equivalent of 100m in 5 seconds, that's a width to width traverse time of 1 minutes 40 seconds.

However if they cover 500m in 5 seconds, then the traverse time is then 15 seconds which would make the 2km map feel much smaller. heh
SimoRoth wrote:The game certainly wont be empty, but there will be a limit. It wont end up looking like a game of roller coaster tycoon though. :)

As for the limits, I don't know. I guess in a mature colony you might have ~30 colonists and ~16 imps. I'll have to see how the native fauna, chickens and other things do at eating resources.

I want the loss of a colonist to be a major event and a major set back.
Loss of colonists having meaning appeals a lot, but a ceiling of around 16 imps seems rather slow but that feeling of it being rather low is just a blind reaction made on not having exposure to the various factors such as:

* The speed they cover ground (and through relation, the actual effective size of the map).
* General time for key tasks to be completed. (How long to mine one cavern wall section. Build a generator, etc)
* If I.M.Ps are the main units designed to expand and build, leaving colonists to deal with maintenance and operation of stations, or if the jobs are spread between.
* How much individual item construction job creation is involved in getting a single room operational on average.

And things like that.

Time will tell I guess. Will probably be one of the first things I'll look at now to satisfy my own curiousity and give any feedback on when the time eventually comes to have the opportunity too, after the initial period of just carving giant rooms into interesting shapes. :D


Edit: Jumping the gun a lot through I know, given that currently it's still focusing on more engine performance. I just like discussing and analysing designs on a more basic conceptual level (That's the much more interesting area IMO heh), so sorry about the premature questions and analysis based on pure speculative details. :oops:
Violence is the last refuge of the incompetent - Isaac Asimov

My channel. Game recordings & old 3D art assets I've made.

User avatar
SimoRoth
Site Admin
Posts: 1106
Joined: Sat Jul 14, 2012 11:28 am

Re: Pathfinding accuracy rewrite

Post by SimoRoth » Sat Apr 20, 2013 12:34 am

Tikigod wrote:
SimoRoth wrote:Worlds are currently all 1km-2km. The final game will be 2km.
Hard to get a idea on how 'large' that would effectively be in the game itself without seeing unit speeds, as if units speeds mean they cover the equivalent of 100m in 5 seconds, that's a width to width traverse time of 1 minutes 40 seconds.

However if they cover 500m in 5 seconds, then the traverse time is then 15 seconds which would make the 2km map feel much smaller. heh
SimoRoth wrote:The game certainly wont be empty, but there will be a limit. It wont end up looking like a game of roller coaster tycoon though. :)

As for the limits, I don't know. I guess in a mature colony you might have ~30 colonists and ~16 imps. I'll have to see how the native fauna, chickens and other things do at eating resources.

I want the loss of a colonist to be a major event and a major set back.
Loss of colonists having meaning appeals a lot, but a ceiling of around 16 imps seems rather slow but that feeling of it being rather low is just a blind reaction made on not having exposure to the various factors such as:

* The speed they cover ground (and through relation, the actual effective size of the map).
* General time for key tasks to be completed. (How long to mine one cavern wall section. Build a generator, etc)
* If I.M.Ps are the main units designed to expand and build, leaving colonists to deal with maintenance and operation of stations, or if the jobs are spread between.
* How much individual item construction job creation is involved in getting a single room operational on average.

And things like that.

Time will tell I guess. Will probably be one of the first things I'll look at now to satisfy my own curiousity and give any feedback on when the time eventually comes to have the opportunity too, after the initial period of just carving giant rooms into interesting shapes. :D


Edit: Jumping the gun a lot through I know, given that currently it's still focusing on more engine performance. I just like discussing and analysing designs on a more basic conceptual level (That's the much more interesting area IMO heh), so sorry about the premature questions and analysis based on pure speculative details. :oops:
As the game progresses IMP's will be able to dig faster, and perhaps move faster so things will hopefully balance out.

There are gameplay implications of having your imps spread too thin though. Which may be a positive if the player appreciates me design.

As with anything it will need balancing. We are currently at a very early stage - getting systems hooked together etc. When we start creating big colonies I'll start balancing it.

kevok
Posts: 22
Joined: Mon Apr 08, 2013 8:03 pm

Re: Pathfinding accuracy rewrite

Post by kevok » Sat Apr 20, 2013 1:39 am

Have you tried talking to Toady and Threetoe? Their pathfinding algorithm handles hundreds of units at a time and still manages to pull normal framerates (And it's not threaded either!)
AMD Phenom II X6 1045T
Nvidia GeForce GTX 560
2x 4GB DDR3 1333MHz
Windows 7 x64

N3wb_4_l1f3
Posts: 8
Joined: Wed Dec 12, 2012 10:39 pm

Re: Pathfinding accuracy rewrite

Post by N3wb_4_l1f3 » Sat Apr 20, 2013 3:03 am

@kevok: I think a major difference there is that Dwarf Fortress is somewhat less graphically impressive, so less resources are used in rendering said hundreds of units. I'm not a programmer though, so I have no idea.

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests