Let me guess, its 2 behavior with validators that compares the angle between attacker and attacked unit? if its 45º-135º it applies a 1.5x multiplier to the attack and a 2x for 135º-180º ?
If thats so what happens if the unit receives an attack in its back, but the attacker is in front, say from an AOE or a ricocheted effect?
How about a behavior with no stacks (limit=1) and the "extends duration" flag set to true, this of course would only works with behaviors that don't need to be increased in power with each stack.
Hold ALT (or CTRL or CMD or SHIFT, cant remember) while placing them in the editor to automatically ignore placement requirements (free placement). Again, can't remember if that ignores/overrides the tree's pathing, but it shouldn't, otherwise you'll have to mess with the tree's actor data.
Will it be with dynamic height? because otherwise pathing can be enough, you have several types of pathing i believe and each of them can be used to check for collision at different heights. Theres is Ground, Air, Building, then i don't remember if theres water and/or impassable.
If you do have dynamic height control then is better just to use a mix of pathing and doodads (invisible or not) as markers, then just check if your units are closer than the radius of your markers and move/damage them away.
I don't see why would you need to clone the weapons for each upgrade, upgrades are made in such a way that you can change pretty much any data field. Need to change the model? upgrades can do that, change the damage? cooldown? what ever your heart desires.
Now Site Operations i don't know if you can modify with upgrades. Actors can be made in such a way that if you cant edit them with upgrades (never tried myself so i don't really know) at least you can validate the options.
Let me share you my experience and maybe a second look at the implementation: I started doing a 3PS a while back for the heck of it and learned a few things (still have the file somewhere):
Tracelines are a nono, except for range finding, when you just want to know if and where the 3d line intercepts the ground (thats the max range of your search area)
Check for angles, far better and faster to find than exact positions.
Adding visual debug tools (something as simple as teleporting a small collision-less orb to the point where you think you're aiming every second or so) can be really powerful tools to find out what the hell your algorithm is doing.
The way i did it (and worked wonderfully for me) was to compare the angle of attack and rotation of my camera, to the angle of attack and angle difference of each unit in the unit group comprising of all the units at most traceline distance of my unit.
That is:
Initial Angle:
Before anything fancy you should prepare your data, if you're gonna make the weapons spread then get the camera rotation and angle of attack and randomize them for the spread effect, then pass these new angles for the next steps.
Traceline:
Repeat X number of times or until your distance is = to the maximum distance (32 is good enough), where X*Step is enough to cover your max range
Check at the current point if Sin(angle of attack) is higher or lower than the ground height
It isn't? then add Step to the Distance and recalculate the next current point. Step can be set proportional to the distance, the further we go, the fewer checks we can make, to cover more distance with the same amount of checks, or the same distance with fewer checks. In my opinion never go beyond 50 checks per traceline.
It is? then that distance is your new maximum range.
Find Group:
Get all the units belonging to the enemy or neutral with filters blah blah blah within at most the maximum range
Get the angle difference of each unit, is it bigger than [maximum spread of gun]?
It is: remove that fucker because he ain't getting shot with this bullet
It isn't: this mofo may get the lucky strike
Refine Search:
So we have all the units in front of the player that may get shot, time for the hard math.
Lets asume for simplicity that all units have a sphere as its collision polygon (funny fact, they usually have 1 or 2), this means that any point between its center and its radius will collide with it
If we get the Absolute Value of the subtraction of both, the facing angle of our traceline and the angle between our unit and the current unit and we compare it to the radius of our current unit:
If its bigger: our projectile is too far from the collision sphere. Remove that unit form the group and move on
Its equal or smaller: our projectile intercept the collision sphere on the horizontal axis. We need more Dataz
If we then compare the Absolute of the subtraction between the Angle of attack and the Angle of Elevation (the angle from the center of the current unit to the center of our unit in the vertical axis) And:
Its bigger: Our projectile is again too far from a collision, remove the infidel at once
It equal or smaller: The fat bastard is dead on on the trajectory of our boolitz.
Last Cleanup:
In the end the last function should pass us a unit group with a couple or none units, if is the first one then get the closest target and get its center point, if its the latter then just get the height of the point at the maximum range
Shot the projectile or effect at this point
Fancy more Fancyness?
If you're a madman like myself you'll think this isn't enough, well the last thing i tried to do with this algorithm was to, once i had a suitable target, find "where" the shot went by performing an abridged version of the above code and look for custom points on the unit (pulled out from a hand-made list) representing soft spots like head, torso, legs, arms... etc. you can even apply special behaviors depending on where you get shot.
Note: Now, why would you bother with such a tiresome work, versus a "regular" traceline? Speed even with all the mathematical functions getting applied here you're still checking data (numbers) against other data, the heaviest of the work, both loops of the traceline and the enemy group are run just once, and we try to work with all the units in the unit group to find out if they're useful or not just once. This algorithm is made so it should at least be able to run 8 times per second, which is a decent machine gun speed. Add some smart caching of data and you can get the fabled 16 times per seconds, the only true limit.
And that pretty much was what i did... that was like a year ago now, so some details escape me, like was it tangent or secant for calculating the height of the final point, and so on...
BTW most of the math was pulled from a few posts from a single user here at SC2Mapster, cant remember the name or the link, but he has about 3 posts explaining his idea of just comparing angles instead of tracing line. Anyways the posts are like a "3d mouselook third person site:sc2mapster.com" google search away...
there is literally a switch effect type that allows you to do exactly that, you fill it with all the effect that you want, each with its own validator and check the "fall through" when necessary. you can use sets to add behaviours as markers when needed and then validate them in the next effect.
Or maybe you don't want that? Honestly by the info on your post the switch effect should be what you're looking for.
Edit: wait, maybe it isn't... switch are made for choosing from a large selection of possible effects, you're implying that you just want to make a static chain of effects... in this case a set effect its more convenient, you just put the effects you want in order, but each one has a validator inside itself, checking for the effect run before them, again behaviours are your friends if you want to check if they fired correctly.
If the last link didn't fire, the current one will no validate, stopping the chain. Also the validator in the set effect itself will allow you to stop the whole chain if you want it.
You entered a trading card game, that only uses 30 cards per deck and where each card has numbers below 15 willingly, and yet you bitch about the RNG?. That game is more broken than brood wars in regards each units OPness, but you know what they said, "when everything is OP, then nothing is OP". Every class has imba combos, and because there are so few cards you have even more chances to pull some weird shit.
So, you had a bad game, but if you lost is also your fault, maybe you didn't know all the cards, or your deck wasn't prepared and was hard countered, otherwise you would've know that a druid can pull this and that combo out of their asses and you would've save or avoid playing some cards to counter their strategy.
Also, why do i say that having numbers below 15 make it harder? think starcraft, the shit with lower hp and attack are workers and even they have 30+ so, 1 point of difference in hearthstone is literally doubly more significant, at minimum than a game like starcraft, not counting the fact that unlike starcraft there are more choices of units and standardised behaviours that allow for cross class combos.
3) [Abilities/Effects] Is it better to have multiple abilities that switch using behaviors or one big using a switch? Like, with Behavior1 the ability will do this, and with Behavior2 it will do that, etc.
As far as optimisation goes i have no idea, but for map maintenance and ease of reading i'll say use the switches.
4) [Behaviors/Units] Any data-only way to do a dynamic threat system? Blizzard's basic one (static) or using buffs (all or nothing) isn't really practical...
I want to make a map with a threat system, and i've been thinking about a mix of dynamic and static threat. Having a staking behaviour that targets forces to attack the caster of the higher stack plus adding taunting should provide enough dynamics for the players to enjoy the AI. things like healing or high damage abilities should taunt or just add more stacks for longer. but i don't think you'll need anything fancier for it to be effective.
9) [Behaviors/Units] Any way to do a D3/WoW-like shield system, with no maximum value defined and only an actual shield value
I'm pretty sure i've seen an ability that added shields to the target, regardless if the target already had a shield (say marine) and just refreshed/increased the target's shield if it had one.
Dude, those are pictures of real life aquariums, we couldn't possibly compete against that within starcraft 2, specially thinking about system specs and broadening the playability of the maps.
sorry but not every map terrain can be solved with "moar doodads", "moar lights" and/or "moar shaders".
An easier way of doing it is through a "combat time" validator to deactivate/nullify the behaviour. Now i don't remember (and i wont look it up right now) if the validator checks for combat time as the time since the unit was attacked or since itself attacked, so maybe do both in a combine with the NOT flag.
0
Let me guess, its 2 behavior with validators that compares the angle between attacker and attacked unit? if its 45º-135º it applies a 1.5x multiplier to the attack and a 2x for 135º-180º ?
If thats so what happens if the unit receives an attack in its back, but the attacker is in front, say from an AOE or a ricocheted effect?
0
Thanks for sharing, mate!
0
How about a behavior with no stacks (limit=1) and the "extends duration" flag set to true, this of course would only works with behaviors that don't need to be increased in power with each stack.
0
Hold ALT (or CTRL or CMD or SHIFT, cant remember) while placing them in the editor to automatically ignore placement requirements (free placement). Again, can't remember if that ignores/overrides the tree's pathing, but it shouldn't, otherwise you'll have to mess with the tree's actor data.
0
You have to individually edit the splat scale to match the radius of your ability in the actors tab.
0
Will it be with dynamic height? because otherwise pathing can be enough, you have several types of pathing i believe and each of them can be used to check for collision at different heights. Theres is Ground, Air, Building, then i don't remember if theres water and/or impassable.
If you do have dynamic height control then is better just to use a mix of pathing and doodads (invisible or not) as markers, then just check if your units are closer than the radius of your markers and move/damage them away.
0
I don't see why would you need to clone the weapons for each upgrade, upgrades are made in such a way that you can change pretty much any data field. Need to change the model? upgrades can do that, change the damage? cooldown? what ever your heart desires.
Now Site Operations i don't know if you can modify with upgrades. Actors can be made in such a way that if you cant edit them with upgrades (never tried myself so i don't really know) at least you can validate the options.
0
Let me share you my experience and maybe a second look at the implementation: I started doing a 3PS a while back for the heck of it and learned a few things (still have the file somewhere):
The way i did it (and worked wonderfully for me) was to compare the angle of attack and rotation of my camera, to the angle of attack and angle difference of each unit in the unit group comprising of all the units at most traceline distance of my unit.
That is:
Initial Angle:
Traceline:
Find Group:
Refine Search:
Last Cleanup:
Fancy more Fancyness?
If you're a madman like myself you'll think this isn't enough, well the last thing i tried to do with this algorithm was to, once i had a suitable target, find "where" the shot went by performing an abridged version of the above code and look for custom points on the unit (pulled out from a hand-made list) representing soft spots like head, torso, legs, arms... etc. you can even apply special behaviors depending on where you get shot.
Note: Now, why would you bother with such a tiresome work, versus a "regular" traceline? Speed even with all the mathematical functions getting applied here you're still checking data (numbers) against other data, the heaviest of the work, both loops of the traceline and the enemy group are run just once, and we try to work with all the units in the unit group to find out if they're useful or not just once. This algorithm is made so it should at least be able to run 8 times per second, which is a decent machine gun speed. Add some smart caching of data and you can get the fabled 16 times per seconds, the only true limit.
And that pretty much was what i did... that was like a year ago now, so some details escape me, like was it tangent or secant for calculating the height of the final point, and so on...
BTW most of the math was pulled from a few posts from a single user here at SC2Mapster, cant remember the name or the link, but he has about 3 posts explaining his idea of just comparing angles instead of tracing line. Anyways the posts are like a "3d mouselook third person site:sc2mapster.com" google search away...
0
I'll leave this just in case anyone needs it.
- Glicko system
- Elo System
0
literally search for "linear projectile"
0
there is literally a switch effect type that allows you to do exactly that, you fill it with all the effect that you want, each with its own validator and check the "fall through" when necessary. you can use sets to add behaviours as markers when needed and then validate them in the next effect.
Or maybe you don't want that? Honestly by the info on your post the switch effect should be what you're looking for.
Edit: wait, maybe it isn't... switch are made for choosing from a large selection of possible effects, you're implying that you just want to make a static chain of effects... in this case a set effect its more convenient, you just put the effects you want in order, but each one has a validator inside itself, checking for the effect run before them, again behaviours are your friends if you want to check if they fired correctly.
If the last link didn't fire, the current one will no validate, stopping the chain. Also the validator in the set effect itself will allow you to stop the whole chain if you want it.
0
You entered a trading card game, that only uses 30 cards per deck and where each card has numbers below 15 willingly, and yet you bitch about the RNG?. That game is more broken than brood wars in regards each units OPness, but you know what they said, "when everything is OP, then nothing is OP". Every class has imba combos, and because there are so few cards you have even more chances to pull some weird shit.
So, you had a bad game, but if you lost is also your fault, maybe you didn't know all the cards, or your deck wasn't prepared and was hard countered, otherwise you would've know that a druid can pull this and that combo out of their asses and you would've save or avoid playing some cards to counter their strategy.
Also, why do i say that having numbers below 15 make it harder? think starcraft, the shit with lower hp and attack are workers and even they have 30+ so, 1 point of difference in hearthstone is literally doubly more significant, at minimum than a game like starcraft, not counting the fact that unlike starcraft there are more choices of units and standardised behaviours that allow for cross class combos.
0
some of this i'll like to know too
Yes you can and Validators are your friends.
As far as optimisation goes i have no idea, but for map maintenance and ease of reading i'll say use the switches.
I want to make a map with a threat system, and i've been thinking about a mix of dynamic and static threat. Having a staking behaviour that targets forces to attack the caster of the higher stack plus adding taunting should provide enough dynamics for the players to enjoy the AI. things like healing or high damage abilities should taunt or just add more stacks for longer. but i don't think you'll need anything fancier for it to be effective.
There was a post about it, the gist of it was using a periodic event to refresh the catalog with the specific data for the specific player.
I'm pretty sure i've seen an ability that added shields to the target, regardless if the target already had a shield (say marine) and just refreshed/increased the target's shield if it had one.
0
Dude, those are pictures of real life aquariums, we couldn't possibly compete against that within starcraft 2, specially thinking about system specs and broadening the playability of the maps.
sorry but not every map terrain can be solved with "moar doodads", "moar lights" and/or "moar shaders".
0
An easier way of doing it is through a "combat time" validator to deactivate/nullify the behaviour. Now i don't remember (and i wont look it up right now) if the validator checks for combat time as the time since the unit was attacked or since itself attacked, so maybe do both in a combine with the NOT flag.