User:GoogleFrog

Posts

• Attack move isn't the "fight optimally" command
• Nanoframe decay
• An unoptimised introduction to log scoring
• The design principal of player ownership being irrelevant to the simulation
• History and design of the start box system
• A note on allied units blocking projectiles
• Fuzzy information and the prediction-reveal loop of radar dots
• Designing away the need for automation with Dante
• Disarm and Self-D (page 2, links to battle threads aren't great)
• The impact-prominence ratio and complexity budget
• More on air design and escallation
• Air design vs. Starcraft
• How powerful should dodge AI be?
• Moderating respecting the efforts of your team
• Physics vs. Balance and Fun
• An example of latency and unfairness in unit AI widgets
• Approaching balance and feedback
• History of Zero-K
• Jumpjet physics
• Big teams room splitting
• Win conditions
• Guidlines for good UI interactions for widget and gadget development.
• Widget development is not necessarily helpful for game development.
• Levels of UI customisation.
• The high-elo room conversation.
• Types of RTS Skill
• What does Starcraft 2 do well?
• Macro, and why have powerful UI/automation?
• Weapon behaviour is independent of target
• Rank system proposal
• What a good league should do
• Similar league discussions
• Riot/raider mix units
• Approaches to balancing and design, factory RPS, mex cost.
• Why do automation and what are its effects?
• Superfluid origins.
• Superfluid release notes, to compare to origins.
• Preserving jank without sending units to space.
• Bomber design
• GDS notes in chat log form
• The infinite APM model, shared unit control, and decisions vs. actions
• Stance on personal widgets and cheating implications
• What sort of skills should ZK test?
• Complexities involved in introducing new players
• A thread on factory plates.
• Against repulsion shields
• Terraform cost calculation
• Comparing ZK to Supreme Commander
• Superweapon design goals
• Two ways to implement smart units (with followup)
• Shield link as a mechanic, not an AI (AI vs. baked behaviour)
• Taking feedback
• On armour classes
• Bomber design history
• Unit renames history
• Opponent AI design
• Release notes introducing dynamic commander morphs
• Expansion and aggression as the core of RTS

Mod Requests

• Custom Factory Creation/Drafting
• Sleeper Agents
• Demon Gate KotH

Rambly Low Quality Posts

• Weapon consistency vsx outcome consistency.
• Messy response post about personal widgets.
• A lengthy game analysis
• Renames thread
• Command panel design and another
• A reference for the old UI
• Sound design (a bit outdated)
• Technical details of terraform cost.

Notable Release Notes

• Sea rework
• Dynamic commanders
• Terraform tweaks
• Idle behaviour and simple command panel
• Overkill prevention
• Construction plates

Old Balance Threads

• Planetwars and another
• Shieldbots
• Widow
• 2018 Balance
• Skuttle and more Skuttle
• Nuke
• Detriment
• Amphbots
• Spiders

Topics

• The old XP morph system: 1, 2, 3
• Superfluid: 1, 2
• Commshare: 1

Factories

Factories fall into a few categories that define a few themes:

• Vehicle: Rover, Tank, Hover.
• Bot: Cloak, Shield, Spider, Jump, Amph.
• Sea: Hover, Amph, Ship
• Air: Gunship, Plane

Vehicle factories share a few traits:

• Their movement is generally speedy, cannot traverse steep cliffs, has noticeable turning circles.
• Vehicles posses the heaviest and most reliable artillery.
• Vehicles tend to lack solid skirmishers.

Bot factories share a few traits:

• Can at least pass hilly terrain, are slower than vehicles, have good turn rates.
• More focused on a utility mechanic (see cloak, shield, jump, all-terrain, sinking).
• Has janky or nonexistent artillery.
• Good or efficient skirmisher units.
• Each has some take on a sneaky bomb/assassin type unit.

Rover, Tank and Hover should be standard matchups on relatively large or flat maps. In this context:

• Rover: Emphasises general speed in its mainline units, but has the slowest main raider. Light and efficient.
• Tank: Emphasises escalation and assaults. Has raiders that can get it to the midgame.
• Hover: Emphasises glass cannons and extreme designs. A tenancy for precise energy weapons.

Cloaky and Shield are cheap generalist factories. Their units are more efficient than vehicles, but are generally slower. Comparatively:

• Cloaky: Faster and prefers open areas due to direct fire weaponry. Glass cannon. Light units want to win by inflicting more losses than they take. Units are able to fend for themselves (eg Knight and Ronin are relatively good vs raiders).
• Shield: Tankier and prefers hilly terrain due to indirect fire. Can escalate and assault. Wants to win through efficiency and attrition. Units are quite vulnerable to hardcounters when not supporting each other.

Spider and Jumpbot are more specialised bot factories. To do well they will likely need impassible terrain to exploit and a map that is not too large.

• Spider: Fairly straightfoward but lacks a true raider.
• Jump: Where all the janky units are stored.

Amph and Hover would ideally be reasonable 3rd options on veh or bot maps.

• Hover: Currently works this way since it doesn't interact much with the water.
• Amph: Needs work to be more viable on land maps. It does not need to be as viable as Hover because the specialised bot factories can take up some of the slack.

Gunship and Planes are support factories. They should be rarely 1v1 ploppable. They are intended to escalate the game in complexity past the first 8-10 minutes as AA becomes a consideration. Striders somewhat fill this role as well.

Ship should be the backbone of sea play. While land could look like eight main factories and three support factories, sea looks more like one main factory and five support factories. If sea is a significant part of a battle then not having at least one Ship player is probably a mistake. To that end:

• Ship has 3 or 4 raider-ish units that can be made from the start of the game.
• Ships generally have bonkers stats compared to their ground counterparts.

Factory Issues

Jumpbots

• Pyro is generally bad against raiders and too expensive for what it does, yet could be oppressive sometimes.
• Jumpjets break weapon velocity prediction. This will become important and annoying if Pyro ever sees much use (affecting matchups such as Glaive and Archer while leaving Duck and Dagger unaffected).
• Skuttle is annoying for all involved and doesn't interact after it goes off. Some people specialise in it, but it doesn't seem worth keeping overall.
• Very weak to turret spam.

Spider

• Venom is annoying to use and to fight because it depends on overlapping AoE successfully. It also self-damages your own Fleas.
• Flea may need to be slightly better. Venom or Redback should be moved into the role of slow raider.

Amphbot

• Generally a bit slow.
• Needs a Buoy to do many things on the map.
• Has no way to deal with Recluse or Rogue in the midgame.
• Very weak to turret spam.

Shield

• A single Outlaw is a bit too good at making a shieldball unattackable at Ravager range.

General Issues

• Stardust should gain almost nothing from attack ground. Essentially, it should automatically fire towards things that it could tickle with the edge of its AoE.

Todo

Better Idle Behaviour

when a unit goes idle -> self-acquired attack command.

1. If the target is a fleeiee, skirmiee or swarmiee then tactical AI almost as usual.

* Move back to original position when re-idle.
* If the target is too far from idle location, and I am closer to the idle location than the target (+ some constant), then Move back to original location.

For example a Glaive may move back if the target is further than 500 elmos away from idle location and if the Glaive is 100 elmos closer to the idle location than the the target. Ronin may have similar numbers, but the upshot is that on retreat it won't be closer than its target to the idle location.

2. Add an idle-only fleeiee table, mostly to be used for raiders against riots. this seems exploitable (the Move command means that breaking vision against a target will be a way to bait) but I suppose the idea is decent basic behaviour it is just meant to give you a bit of time to notice that something is happening

Addendum: If a unit goes idle with an enemy in range then it should attack it without using idle tactical AI, as the user probably meant to tell the unit to attack.