Jump to content

Edit History

Please note that revisions older than 365 days are pruned and will no longer show here
faatal

faatal

1 hour ago, Jost Amman said:

@faatal I'm no programmer, but I've read sometime ago on another game's dev blog that games often use a "decision trees" for the AI stuff.

 

You surely know better than me that, apparently, using those you can have "toons" do complex stuff (like simulating needs and different reactions to events and situations).

Is it possible to do something like that with Unity without going mad? And, most important of all, would that be a viable option for the new bandits AI? :nerd:

 

(P.S.: I've willingly ignored the pathing problems because I don't consider that "true AI", but just an accessory to it)

I was never impressed by decision trees. Technically we use decision lists, but some AI like the vulture is just a state machine. They all get the job done and are not hard to do, until you get to the gory details of what all of those decisions have to do to actually work.

 

Health < 50% run away. Cool, but to where? Taking cover is going to be one of the hardest parts of bandits, since what is cover in an arbitrary voxel world?

 

People love to gloss over the details, but the details are what take all the time and kill you in the end if you don't do them right.

 

Priorities get in the way of lots of fun. Last Friday I was thinking I could mess with the raft, but nope someone needed me to look at a light block bug. Today I was thinking maybe some raft time....nope new short grass needs to be tested, so artist can get feedback, which sounds simple except for the new mesh generation short grass will use and adding to the biomes, splitting from the tall grass code, shader changes so it blows in the wind since it is too short and lots of testing. All of that eats up time, so maybe it will be done Monday.

 

We have our A20 must fix list going now. 111 currently, 8 of which are mine, but in the beginning it tends to grow as fast as you can fix them, so fun keeps having to wait.

faatal

faatal

1 hour ago, Jost Amman said:

@faatal I'm no programmer, but I've read sometime ago on another game's dev blog that games often use a "decision trees" for the AI stuff.

 

You surely know better than me that, apparently, using those you can have "toons" do complex stuff (like simulating needs and different reactions to events and situations).

Is it possible to do something like that with Unity without going mad? And, most important of all, would that be a viable option for the new bandits AI? :nerd:

 

(P.S.: I've willingly ignored the pathing problems because I don't consider that "true AI", but just an accessory to it)

I was never impressed by decision trees. Technically we use decision lists, but some AI like the vulture is just a state machine. They all get the job done and are not hard to do, until you get to the gory details of what all of those decisions have to do to actually work.

 

Health < 50% run away. Cool, but to where? Taking cover is going to be one of the hardest parts of bandits, since what is cover in an arbitrary voxel world?

 

People love to gloss over the details, but the details are what take all the time and kill you in the end if you don't do them right.

 

Priorities get in the way of lots of fun. Last Friday I was thinking I could mess with the raft, but nope someone needed me to look at a light block bug. Today I was thinking maybe some raft time....nope new short grass needs to be tested, so artist can get feedback, which sounds simple except for the new mesh generation short grass will use and adding to the biomes, splitting from the normal branch code, shader changes so it blows in the wind since it is too short and lots of testing. All of that eats up time, so maybe it will be done Monday.

 

We have our A20 must fix list going now. 111 currently, 8 of which are mine, but in the beginning it tends to grow as fast as you can fix them, so fun keeps having to wait.

×
×
  • Create New...