Zenkai Knights

i think i will use ms-98 remake i did.  i will though need to change some things on it.  well, only need to change the back thrusters to be a backpack instead.  shouldn't be hard.

also controls are changing.  kustom knights will be using lightweight controls and movement.  the more advanced controls will be in another version of the game called kustom knights: high grade. the reason is so i can simplify the AI code.  honestly, i spoke about doing advanced ai though i never really coded AI before.  another reason why is so its easier to use with touchscreens.

air movement will be more limited but as trade off there will be more variations of mechs. 1 variation is to add beast like mechs similar to zoids.

WIP:


reused ms-98 remake head for millenium's head. i hate doing heads and both mechs seem to have similar heads.

Seeing this many changes times to times suggest you should try putting together a working prototype game for personal use. That will help you seeing what works, what can be done with your skills and what solutions you could utilize. The AI especially requires testing since Unity have a very simple NPC controlling method. I definitelly would love to see that advanced AI but maybe it is too much script for the engine to handle it fluidly.

the hard part i am seeing is the full 3d movement aka telling when to go up or down with my original controls for AI.  remove air combat to a minimum and its easier to deal.


obviously, movement controls aren't the problem.  its the AI and attacking that's my main issue for coding. another problem is i hate animating and modeling.


i have a problem of thinking things big when i should think small.


as far as adding modding well, i know how to load obj file at runtime and have experience with file design.

i think its time to ask for help.

i should do some testing with ms-98 remake.

test i will do is simple. no aerial stuff. just ground movement with 2 attacks(basic melee and basic range) able to be done for players. AI will be programmed for second player.

The AI is the hardest part of the games, I doesn't even know how it works. Reacts to the happenings while decide how, sometimes I have trouble doing so in real life too.

the way i think about how to go about coding an AI is to look at it in perspective of how a human would do things in a game similar to what i am building. i studied some of your gameplay and others to get an idea of how the AI should work.

pseudo code of AI

-Calculate what action to take based on players in area and actions they are taking (move, attack, guard or other action)
-if Move
--determine what kind of movement to make and where to go or to change current movement to towards target or another player
-if Attack determine what attack to use or to go towards target for melee attack.

i wouldn't do all this all at once but spread this out through calculation steps.

Yes, this is the logical part of how a player fight. There are moments however when a completelly illogical option kicks in. The most common situation is using an attack or move that will miss/being punished in order to force the opponent in a position the next attack will land with an effect that worth more than the sacreficed effort. You can call this "setting up a trap" maneouver.

I don't know how much your mechs will be different from each other but obviously the different mechs always requires different handling. In Windom XP from close range you can't think on blocking melee with Blackcomb or Anthurium from close range because they are slow. Yet the Blackcomb capable of blocking it with a trick which however have a limited use and punishable if used wrong.

so i studied some of exvs force AI.  its very simple in design.  

here is the basic design:
-if no target is found
--search for target by moving randomly.
-else
--go to target and attack if when possible.

easy, normal and hard is basically increase in attack rate and possibly increase in sub attack use.

this is how it works in 1v1 basically. pathfinding isn't perfect too.  if a mech hits a wall it will only jump/boost up if movement is not much but if its sliding against a wall moving a good distance it will count that as moving. though the stages are very open which helps make this ok.

after some more studying the AI is programmed to back off from the mech after attacking.

ok, seems mechs pathfinding is better that i thought. sometimes they will jump over building before hitting them. especially when mechs are above itself

I can assume the AI doesn't really care about the environment. Even without lock-on it should have some kind of tracking that leads the AI toward the possible target. Since the stages are mostly flat spaces with a little spread of obstacles I guess it is the most logical to increase the altitude. I'm sure it has some kind randomization in strategy but mostly they just want to move within the optimal range of their.

Compared to this Windom XP is a way more simple. The AI can lock target even out of range only can't attack it. The AI cares about nothing just want to close the distance to reach the firing range. Sometimes they stuck between the buildings because they move randomly and not calculating with the shape. While it is basic it also can be extremely good. I've seen them launching attacks frame-perfectly especial against guard or by cancelling from guard.

Mech like Anthurium will try to stay out of melee range anyway.

the way my advanced AI would work is it uses chance values and distance values and plugs that in to create a AI similar to how the player would play.   essentially the values are layers on top of a base AI system based on chances.  the values just increase or decrease the chance of actions happening. though i need a base to work with to generate a starting AI system and slowly implement the values.  

one example of a value is failure rate.  what chance will you have to likely to do something by accident or pull off an attack to early.


well, my best option for creating an AI is to use windom and extreme vs as a base to work off of.

In my oppinion there is two "type" of chances when it comes to playing. The basic chance is what you're talking about, aka how likely an attack hit the target (or how likely the AI can prevent getting damage). There is an another kind of chance which should tell if there is a better chance to perform the same action by waiting a little. This second is the hard part implementing. The AI can be tricked if it goes with straight "yes" or "no".

i am holding off the testing.   i am taking a class with unity and i think i should hold things off so i can kill 2 birds with 1 stone.   final project is going to be to build a small game.  i would use this stuff i researched to help build it and do some testing.


i may decide to do the test project early though.

I will look forward.

taking what i learned from my 2d game. i can continue this project but not ready to yet.

Take your time, it is always more time consuming to create something new than modding something existing. Unity is at least user friendly.

well, i considered a custom engine. even picked out some libraries i could use to make it.  though i have no experience in using them.

maybe i will learn how to use these libraries.

Even if you use Unity it is still new. The engine maybe pre-compiled but everything else will be added by you. It is hard enough like this. Building up an engine is not a one-person thing I guess.

technically i am not building everything. i would use an existing graphics engine and physics engine to help cut out the work. also some other stuff.

Oh I see. Well it is still good enough.

sticking with unity. just need to get started with designing the mechs. made it easier on myself to design the mechs to be bio mechanical similar to evangelion in design.

back story on origins of the mechs in zenkai knights:
the mechs were created using humans as base. giant humans once were used to fight but there were flaws in them cause them needed so much resource to sustain them so the dna base was altered heavily and they created mechs out of them by adding mechanical parts. removing any unnecessary organs. the skin of the mechs are wrapped in a thin layer of heat resistance material.

Honestly I never was able to decide between normal or quasi-living mechs. For me the bio mechs seems harder to design with all of those transitions between the living and mechanical parts. Though there are two significant style with a lot of difference. As you said the one that looks like Evangelion, more or less human like and the design goes mostly for the armor as every other parts are identical. The second type is like the Hagane Oni, the mech itself being a mass of cybernetical organism, instead of armor the body parts connected through transitions of the same material. Probably the Evangelion like is easier to create in 3D.

fine, you win.  i will use my kd-03 remake for testing.  the models were always the problem.....i give in to your suggestion.


i do not have very good design skills anyways.

Lol, I suggested nothing. Actually I was glad there will be bio-mechas because those could be more complex than normal mechas, though more harder to design. I've tried modelling only once for Unreal and it ended up like a shaking dog that forgot to walk on all four lol, I can imagine how hard is it for real. But if you have that Windom remake mech you can redesign it to have bionic parts here and there if you want.

The Evangelion like mechs that looks like a living being is easier. Hagane no Oni is the other end, those are like some kind of controllable monsters with mech parts. But no, these are actually mechs but made in an another universe with technology doesn't even exist.

i only said i am going to do bio mechs cause felt it was something i could pull off plus i had a story that made sense with them. i might put that in for another game or write up the story and save it for something else. i am just going to go with whatever i manage cause as i said i am bad with designing stuff.

i actually did some work on the project. ran in to ton of issues.

also without thinking, i am headed in the direction of modding. its out of habit.

I can't say which is the harder, creating a working engine and let others create the content, or modifying an existing engine and creating custom design. As you said pulling out a nice design from a modeller is very hard, coding is more manageable for me too.

creating a custom engine is harder. existing engine it depends on how much knowledge you have of it. i have studied enough about unity to know what i can and can't do. unity also has in place stuff to help create mod systems.