busy now. reply properly later -
Looks like you've put a lot of work in. :)
Printable View
busy now. reply properly later -
Looks like you've put a lot of work in. :)
I always put lots of work in.
Still waiting for your reply(s).....
My boss is prowling today, so I'm having trouble posting...
I've only managed to scan-read what you've sent, but it looks reasonable. I think the important thing is to make sure our "secret projects" are as different as possible from Sid's.
Good work anyway...
Ours are special projects not secret projects. Completing one will not prevent anyone else from completing it.
Q's,
Did you get anything done on the graphics last night? (Where are you up to?)
Do you want me to mail the latest copy of the code over to you so you can plug in the graphics and see them in action yourself?
How long does it take you to knock up those buttons?
smartened up and shrank down the pods - unable to test without plugging together 120 images into bigger bitmaps and I'm not ptepared to. The AVI looked ok though.
There's little point having the new code. I already have the version with the animation engine, but I'm not prepared to plug a bunch of graphics together - thats hours of unnecessary work.
As far as I'm aware, the buttons are done - I found them last night & they look right (black corners, lighter text on understate). If I can fight my sister off the line & get online tonight I'll send em to you...
Do you want me to send you the code for plugging the bitmaps together?
Will only be a couple of K (if that ).
PS - If found the famous half full bottle of Bacardi Limon! Any chance of you Dad picking you up tonight? :D :D :D
no chance... its Wednesday...
Hi bloke.
Still waiting for the graphics you promised me last night...
Morning dude!
You gonna be available for forum chat today?
I want some ideas on how some of the simpler AI's are going to work and also how the computer will determine which AI to use.
For starters I was thinking that all the unit's AI's are set to defend and then when it has 10 (arbitrary number) or more mechs defending it sends a few on the rampage. Number will be a % of mechs defending times by the factions attack number.
This means that it sends mechs in attack clumps (as we were discussing last night).
Of instead of having it do that when it gets to a number of mechs, when it gets to an amount of hardware equiv. to a number. i.e 1000 points of mechs.
Make sense?
(PS - Return to innocence sounds cool as on these new speakers!)
hardware points could really leave th AI in the schmuck - if it gains 1000 hardware points (say for example builds loads of weapons) - it may only have sufficient components for 2 complete mechs. A poor balance - unlikely but possible. I prefer the %age number of mechs idea...The mechs could also be specifically designed then.
An assault party would also be designed for attack on a specific target. To assault a player with a bigger city would require a bigger force. Only if the target size is not known, would an AI attack with a standard arbitrary number of troops...
No, I meant a number of points of complete mechs. Doesn't matter what's in stores - it adds up the points of all the mechs defending and then decides what it's going to throw at the enemy.
Well, how do you propose to design and 'assault party'? We're not even putting mechs into parties - they all work independantly. We need to come up with some method of making the mechs appear to work in packs, when it's simply lots of AI models doing similar things.
e.g. putting them all on defend and then turning a clump over to assault all at once.
Also, I need some idea of how we're going to implement a scout AI - want to post any psuedo code? At the moment, the computer player can simply see all, but I don't want that - it's cheating...
quick question:
do all AI units have "view zones" as a human player might (regardless of whether or not it has knowledge of whats beyond this zone)?
If so, the AI would use the information within this zone as fact - what it can see, it will act upon
In the same vein, if it has uncovered some of the map, it MAY base its attacks on this - is it worth timing the views - for example an area uncovered 4 turns ago is going to be more accurate than one uncovered 20 or 30 turns ago (do you follow?)
The more recently an area is uncovered, the more likely the AI is to use it.
the 3rd option - if a unit has not uncovered an area, it can't act upon it. It would send scouts out to the middle of the largest areas of blackness, and towards the enemies landing site (assuming we're always going to see that)...
sorry - im blathering - thinking on the fly...try this.
* Build scout mech: queue more scout mechs
* send first scout mech towards anything visible (assume enemy landing site) hold position
* send scout mechs to any other "points of interest" - anywhere not viewed for X turns, anywhere partly viewed with enemy presence
[repeat loop]
* send scout mech to middle of largest uncovered area.
if map coverage>60% then stop else
[repeat]
or something like this.
Yes, all units have view zones, but the player should be able to base their knowledge on what has already been discovered, and what sum of all the units views.
No, it's not worth timing the views.
The scouting loop has it's points. How do you define the largest area of blackness? And where's the middle of it?
Finding the centre of any polygon is simply geometry. Finding which IS the largest area is going to be more difficult...
OK - a slightly different angle.
quick calculation:
height of map / units view DIAMETER = max number of mechs to scout 1 edge of map permanently - send each mech to (maxX, (maxY/number of mechs)*(mechumber-0.5)).
However it may be sensible to use just half this number of mechs... Repeat for all 4 edges.
Assume:
100 x 100 map. All scout mechs have 5 view radius (==10 diameter)
100 / 10 = 10 mechs necessary to scout right edge of map.
use half of this for practicality:
therefore send mech1 to (100, (100/5)*(1-0.5)) == (100,10)
mech 2 to (100,(100/5)*(2-0.5)) == (100,30)
etc.
The scoutmechs should radiate from home base giving a good overall map coverage.
(does this sound like the bollocks I think it does?!)
Too impractical.
If you have a 100x100 map then you're going to need 40 mechs to scout out the map. And that's assuming the map's all land.
We can't afford to chew up that many mechs early game, and if we just add mechs when we can afford it, you're going to end up viewing big long lines miles away from base and very little near home.
We need something that explores around the base. Maybe just something that orbits the base? No, 'cos then mapping irregular islands will be daft.
We need something that goes exploring.
we need to devise an "exploring" AI model...
Things to consider:
Will be based on a land chassis type
Will originally scout around home base and then later towards enemy
Will stop at waters edge (obviously), but this will affect how many other scouts are built
Will investigate anything "interesting": pods?, enemy activity, high resource squares
Will avoid heavily armed/armoured enemy (its only a scout!)
Will avoid its other scout mech colleagues (don't rediscover the same bits...)
anything else?
No, that's covered it.
When it gets to waters edge, it's going to have to turn around and wander off somewhere else.
I was thinking of having a couple of explore AI's
Explore coastline (runs 'til it hits the sea and then follows the coast around. Could be useful for determining the size of the island)
Explore local (runs around the base area)
Explore general (runs aimlessly around)
On all of these explores, if a pod comes within view range it will collect the pod.
had to reformat the disk last night :(
Gave me chance to think about stuff though - although I like your scout AI, it occurred to me that its a simple recon scout AI - there would be more AI models for scouts...
So we've got your recon model, a coastline geographer, and possibly a spy-scout (heads straight for the enemy), any other ideas?
How about resource scouts? possibly applied to an RMV - this model follows a resource seam before any mining is done - enables you to check that your RMVs aren't going to wander aimlessly into somebody's base in the pursuit of resources. This is also going to give a quie good general shape or area to discover...
Well, I've implemented (a modified version of) the scout AI model and it seems to have the right idea.
It wanders off scouting in a sensible manner, but for some reason it's not using its full move quotient so it's taking ages to explore anywhere.
Fear not, the mighty Ben is on the case.
so whaddaya think?
BTW even Norton Disk Doctor couldn't find my disk...bastard!
Awww poor Ian's disk...
Sorry, didn't see that other post of yours.
Weelllll, I've got the scout model up and running (needs some error traps, but the principle's sound) and it does the job admirably.
With a couple of scouts you'll be uncovering the map in no time. Super!
I don't think we need any others to be honest. Maybe a coastline follower, but that's not a pressing issue. This scout AI is the biz!
We could do a 'scout out enemy' as well, but it's not that important at the moment.
A general 'move battle lines towards enemy' model would be quite useful, but we have three out the four main ones (attack, defend, slurp, scout)
Explore, Discover, Build.... D'OH!
so your not interested in a number of AI models to improve flexibilty? Whilst a simple recon scout is gonna be the most used, it may be to blunt for some tasks...
Maybe.
Well, I went into work today. Tremendously interesting. :rolleyes:
Rob had some good stuff to say about MBS. Permanent positions coming up etc. Seems to want me to move into a development role however, which I'm not sure of.
I'm more than willing to help out in a development role, but I really enjoy doing analysis and that's where'd I'd like to be.
Of course, it depends exactly what's part of the development role, it may be exactly what I'm looking for... :)
Right, how are you getting on with those graphics? I hope you're getting somewhere now, 'cos I want to get the next version uploaded soon.
I'll do some more tidying work on the AI and then we might have kick ass computer players in there as well. I also want to see if I can speed it up some - it slows down when the faction builds 100 units. :rolleyes:
I've not reinstalled VB yet. will do that when I get back, then finish the buildings. I'm gonna have to get more beer in too. Its gonna be a long night. I've lost 2-3 days of productivity now.
*****! :(
Why, what were you up to all last night?
You should have been working....
Well, I'm only a phone call away and I never sleep, so if you're up late and need a bit of inspiration then give me a call.
If you need anything sending to you, then shout me and I'll mail it.
cheers pal ;)
Awight matey!
How's it hangin?
I take it you got home ok and weren't gangraped in Middleton?
So, watcha up to?
oooh...
*ouch!*
...sore...
*owwww....*
can't sit down...
jus' kidding.
pulled some ideas together for funky looking walls. will do some more on it tonite.
got any interview tomorrow...woooh!
Cool as!
Good luck with the interview.
Remember to clean behind your ears and put on a fresh shirt and....
etc
heya pal-
jus't thought I'd rescue this thread from the depths, how's it going?
Not so bad.
In work at the mo. It's like I'd never left.
I want some beer - I'm going to the pub in a minute.
Rob was plastered last night wasn't he? :rolleyes:
What a monkey!
well, the dumb schmuck will drink "on medication" :D :rolleyes:
serves him right...
On medication my arse!
The man's a fool, plain and simple. Harmless enough though.
We need to fix him up with some ridiculous walrus type creature.
Well, back to business - how are you getting on with those buttons? Probably the same as I am with Plender's fixes.
I think I'll make a start on those tonight and get some of them out of the way.
We could do with creating a comprehensive list of things you need to draw.
There's allsorts - for starters:
Buttons
Hydrogen Trees
Supergiant etc etc
Wall segments
Road segments
Cake segments
Sensor buildings
Turrets
Advanced Research Building
Satellite image (representative)
Waargh!
You're right on 2 counts.
1) Robert needs a walrus.
2) I keep opening the buttons up in photoshop, looking at them and closing them again :rolleyes:
However:
<good news>I have build some wall pieces in Illustrator</good news>
<bad news>I'm having trouble bringing them into Photoshop, and we're gonna have to think some more about the wall generation engine</bad news>
By supergiant, are we refering to wandering monsters? and therefore, they'll be 40 x 40?
Yes. They will be 40*40. Not animated.
What do we need to think about the wall display engine?
When should we do this?
I'll use a quick example, but since I dont have pen and paper :D, you may not grasp quite what I mean...
any wall piece that goes from one square down into a square below will cover up anything beneath it. This is going to be apparent unless you ensure that wall pieces are placed with the lowermost ones "on top" in the Z-order. This is going to have to be dynamically updated, for example if a player closes a hole in their wall half way up...
You didn't understand a word of that did you?
No.
I don't understand.
due to the perspective (not quite top down) then walls at the lower half of the square are closest to the top.
Hey dude,
If you fancy coming round for bevvies then I'll come and pick you up but (obviously) won't be able to give you a lift back.
Gissa bell later - I'm off home now...
:)
Morning blokey:
You get those buttons? Like em? plugged em in yet?
never mind then, eh.
What's happening today then? anything good?
:D
Nothing good is happening today - I'm dying of boredom.
I haven't yet plugged the buttons in, I'll do it soon though.
You up to anything good?
Have we come up with a way of measuring the might of a unit yet?
*ahem*
In a roundabout way, I'm sure we have. I think I'll do some searches in here today for the words faction and might posted by either of us.Quote:
Originally posted by Arbiter
Have we come up with a way of measuring the might of a faction yet?
We can't calculate a factions might until we can calculate a units might can we?
:rolleyes:
allocate a value to each component and add all of these together for the mech, add additional point for any special dibs it's got (rust warheads, superduper tracking) addtional points for number of kills and number of turns since its birth (experienced mechs are cooler) perhaps it might be wise to multiply this figure by the amount of armour the mech has remaining also.
Yeeesss, this much is obvious.
WHAT number do we allocate to each component?
Also, we have no current way of recording number of kills per mech and number of turns it's been alive. If we need to record this then I'll need to amend the variables. Also, it's a little excessive if we're only using them to calculate its might. Will they have any other uses?
Sarcasm is of no value to us here :mad: :p
perhaps an experienced mech is a more efficient killing machine? apply bonuses to mechs with high kill rates and long lifespan?
ok: points allocation ( :rolleyes: )
mini laser, spud gun (+1)
medium laser, donkey shaver (+2)
big biffy autocannon, dog dryer (+3)
blah blah, yackety schmackety don't have the list to hand...
all chips would be valued according to their grade:
bottom end EMC (+1)
bottom end battle chip (+1)
advanced sensor chip (+2)
etc
in theory, all chassis would be valued equally (we tried to make them fair) if this proves not to be the case, we'll have to balance them dynamically.
is it worth varying the values for torso/weapon mount? whilst a 4 point mount would be more valuable than a 2 slot, this would be obvious in the number of guns...
:D
Making longer living mechs more efficient means rewriting core components.
Please don't make me.
Can we create a formulae that calculates the might for each weapon on load (such as resource cost * damage or something)?
PS - did you check out Plends RentaBoy site
www.rentacoder.com
ok, lets not include a mechs longevity - it was only an idea and probably wouldnt be very fair.Quote:
Originally posted by Arbiter
Making longer living mechs more efficient means rewriting core components.
Please don't make me.
Can we create a formulae that calculates the might for each weapon on load (such as resource cost * damage or something)?
yeah, we can have a formula do the calculating onload, but why would it be easier? ... oh, for the dynamic aspect - I've got ya. I assume damage is base damage cause by the weapon and not the damage done by the mech (which would vary depending on loads of different factors...)
Oh.
I guess it would be base damage caused by weapon times the benefit of the faction.
Armour would be cost * armour value * faction benefit
etc
yeh...
*nods sagely*
yeeeaaahhh...
Fancy coming round for a couple of bevvies this evening?
I suppose I do. We'll see. I depends on whether or not I can get a lift i think...
I can give you a lift from your house or Rochdale, but not home (obviously).
If I pick you up in R'dale then we can start earlier and finish earlier (meaning buses will still be running).
Let me know either way.
nope, im not getting a bus. I'll go home first and see if my dad's prepared to bring me home. It's Tuesday, isn't it? pub night :)
Bettys BOOBIES!!!
:rolleyes:
I'm still tired. I'm working on an algorithm for determining how irritated a player is which where the other players units are.
I'm going to split it into two sections.
1) How upset the player is with the might of the other players.
2) How upset the player is with where the other players units are.
I'm working on number 2 at the moment.
It's proving hard to get my head around (maybe because I'm struggling to concentrate).