I am thinking of making a Sunken Defense where enemies spawn at random corner of the map and need to move to another random corner, and players will have to co-operate to build sunkens together into a maze to fit these random spawning and ending locations.

However, I am not sure what is the best way to ensure that the players have not walled.
I can think of three solutions.

1. Let sunkens be vulnerable, and let blocked units naturally attack and kill the sunkens.
pros: no triggering, and a lot of other sunken d's use this already.
cons:

• ultras and goliaths.
  
• Units can attack you for no reason anyways, and it doesn't actually prevent walling 100% of the time.
  
• Can no longer allow splash damage units to be bought

2. Force players to move a unit through their maze before the round starts.
pros: can make sunken invincible
cons: Because I want the spawns to be randomize every 3-5 rounds, players will have to do this a lot of times, and not all pubbies know wtf is going on.

3. Have unbuildable terrain that links all random spawn/end locations.
pros: easy to implement and understand. Can even make patterns with the terrain to force different designs.
cons: less maze making space and will probably frustrate people who want to make sprawling mazes.

So, my question is, have anyone had an elegant way of detecting walling, and what your opinion is on these 3 options. Maybe there're options?

Thanks.

If the HP is so high that only maze units can kill them this won't be a problem.

Why the players? Have an invisible computer owned unit walk along with the rest each level, if it never makes it to the other side there's a wall.

Have actually been toying with this very issue recently. The best option currently would just be to have a pre-wave inspection but the biggest con would obviously be that the players cannot build between this check and the end of the actual wave. With this you can have invincible stuff, but it also brings in the issue of complicated mazes and how annoying SC's pathing can be. You can opt to make a player control the unit instead like you said, but obviously you said this would be confusing for new players.

Now the best option is you do something nobody else has done before, which is use a mobile grid system that quickly runs through the map trying to find it's way to the exit. The biggest problem with this method is that you would obviously have to design it yourself. There is no previous work to base yours off.

That mobile grid method would lag like ass calling it now.

Okay I've edited this post a few times but here's a different approach:

Embed this into part of the game. While the players are building, they will likely want to know the path the units will take. We've all seen this: the builders that exploit the easily-confused pathing of units in Starcraft by creating wide areas that lead to a wall, which require the units to double back and go another way.

Assumptions:
- You're allowing the players to destroy sunkens and rework their design until ready to begin. Destroyed sunkens will reimburse a drone.
- Sunkens are NOT invincible during build mode
- Sunkens ARE invincible during the round

Here's what I have in mind:

While the players are building, allow them the option to 'path test' the build. A player may elect to enable a 'Path test' mode in which all drones are moved out of the arena during the test.

1. Player brings a civ to a beacon to enable the 'Path Test' mode.
2. Enable invincibility on sunkens
3. Create 1 computer controlled ultralisk/hero ultralisk (invincible) every second at the designated start point and order them to the end point. (May want to consider at most 8-12 ultralisks/hero ultralisk at a time)
4. Allow players to observe the pathing of the computer (allow the game to detect if the computer can make it to to destination)
5. After a computer unit makes it to the end, enable the players to be able to either:

A: Rebuild (and consequently, retest)
B: Begin the round


If they choose A, obviously return the drones to the arena, disable invincibility, etc....

Post has been edited 2 time(s), last time on Apr 20 2017, 3:33 pm by Matt.-.

I remember a bunker defense map in which each player had their own path. It was entirely filled with machine shops and you had to walk your scv from one end to the other, destroying the machine shops to make your maze. They then became invinsible for the rest of the game.

If you tell newer players to think to make room for creep colonies, you might be able to make them stack them in predetermined nooks.

Can't use that technique if you're looking to allow players to build the maze as they get the ressources, old style, but it does show to me that moving a unit to the end to test the maze isn't that hard to understand, especially since I was about 12 back then and barely understood English.

That old map was called Path Defense, and it is definitely one of the options, but kind of prevents you from continually pathing

I mean, if I was playing a defense and had to test my maze every time I wanted to go I wouldn't play it. Some mazes take FOREVER or sometimes NEVER for the mobs figure out where to go. I wouldn't play it. It has to be a simple and quick method or your game won't be re hosted. Period.

The best options I've seen are unbuildable Terrain and having your walls attack as long as PLAYER 2 can't bring his units into your base and create a wall against your own will.

EDIT: After reading the last bit of comments.. I thought of a design by LoveLess. He basically has the map full of "walls" and randomly removes X amount of them leaving some scattered throughout the play area. If these buildings are setup so they're offset from your maze, it could make it more difficult to wall.

It all depends on your limitations. For instance with a mobile air grid, you can get down to a 1.5x grid but it can scan as fast as you're willing to go.


UU deserves some credit there though, his version saved some triggers and runs faster in high quantities than mine did. ...Even though he didn't realize it until I told him it was simple to make it scan faster...

The only problem with the grid is when you start adding in more variables, like different sized buildings into the same maze as that can greatly alter how the anti-block scans. So until we get closer to a perfect 1x1 grid, oh well.

Just throwing a couple ideas out that probably don't work, but might help you think of something else

What about instead of destroying things to create a path, you instead actually build the path with another structure, then use triggers when starting the round to check whether the path is valid? If it's valid (check with roughly 4x2 and 2x4 locations), destroy the structures marking the path, prevent it from being built on with burrowed units or something, and then leave that path "locked" until the start/end locations are changed. That requires extra unit slots though. No guarantee the AI will be able to run it.

An alternative to #3 are checkpoint borders, basically a 2x2 unbuildable terrain border every so often that allows units to just go around and ignore mazes that are too complex, but still allows players to have their flexibility within the buildable space. However I can see unit AI just deciding that any and all mazes are too complicated and just bypassing everything all of the time.