Messages

1

Design / Re: My comments on the design/gameplay

« on: July 31, 2009, 12:05:09 pm »

Alternatively, even if the aliens were all the same - what makes it repetitive is that they all act the same. I.e., shoot direct fire-weapons and hide in predictable places. They never use grenades (unless they carry one when you spawn next to them), or any other variety of weapons. I know that improving that is somewhere at the bottom of the AI todo list. But I wanted to stress how important this is. There are more than enough weapons for the Phalanx - give some to the aliens, and make them use them in less predictable ways!
I would love it if they had, say, a grenade launcher. Or some kind of flamer, and someone who charges your flank with it. Where is the scary Chryssalid? That alone, some more variety there, would keep those missions interesting ENDLESSLY.
At least, again, let those aliens who use differnet combat styles (Bloodspiders) see some action earlier.

I've started planning/flowcharting a revamped tactical AI.   I'm also in the Military and currently deployed so it has been slow to work on. (Bullets trump things priority wise.) I should be moving home soon.   it will take a month or two and then I can get out of the initial planning/General UFO Code/AI familiarization stage and start scripting. (Lord willing and I make it home!)

If i can get a chance to get my current form down on paper I'll upload it.  in short, I am going to fuzzy the logic.    The results of several tactics will be weighted by several factors.  Then a random element will determine what precise course of action the alien takes with weighted choices being more likely.  Alien "morale/health" may play a factor (call it strength of connection to the hive mind.) in determining how erratic the course of action becomes.  Not purely random mind you, but weighted,  healthy aliens will choose the most tactical/teamwork oriented choices.  As "morale/health" decreases,  the choices will become more erratic.

The "neutral" civilians, soldiers, cops, etc.  will work in a similar way,   but be increasingly erratic or desperate/heroic as they recieve losses.
(Yes, hide, herd, hysteria, hero will all play a part in working it out.  I haven't abandoned that simplicity.)

of course once i get this worked out I have to figure out if it is worth the effort...  It may be overly fancy, (albeit cool!) and a simpler system may work well. I just feel that overly scripted, hard defined, AI gets predictable.  I am hoping to allow enough room for some "emergent AI" to show through.   Where the aliens sometimes do things that are unexpected and Eriely sentient...

I've had some chats with Chris Park over at www.arcengames.com, developer of "A.I. Wars",  a space based RTS on AI, and i am implementing many of his concepts for AI.   (The AI he has designed for his game "AI Wars" is WICKED! Space RTS fans will enjoy it!)  but emergent AI in a game with 20 units vs. 20000 units is a little bit harder to evoke.

Truth will be in the playtesting!

2

Artwork / Re: Human Head Model Proves A Headache...

« on: July 23, 2009, 03:13:36 pm »

let's focus on getting it done.  Re-skinning some heads will be child's play to put faces in later.

3

Artwork / Re: Human Head Model Proves A Headache...

« on: July 19, 2009, 12:08:36 pm »

lol, the low poly looks good actually!   I like it better then the high poly version.

the problem is your running into what psychologists and artists call the "the uncanny valley".  The human brain has developed very sensitive facial feature recognition centers.  You hit a point where as things get more detailed your brain stops noticing what is right about the face and starts focusing more on what is wrong...   making it more detailed can make it worse, sometimes simpler is better!   A simple face our brains fill in the gaps and fools itself into thinking its a a friendly/good face.  the more realistic it gets the less room for your brain to fill in the blanks and it ends up tripping over the wrongness more.   

sometimes suggestive is better then representational!  The smaller faces for the low poly version lose detail and move out of "the uncanny valley".


cheers!

4

Windows / Re: All in one win32 build script

« on: July 14, 2009, 09:47:01 am »

if i am hearing this correctly,  all the needed files will be included in the SVN.  and when complete all you need to do is doubleclick 1 file. and it will update everything, compile everything, and give you a neatly wrapped complete with a ribbon bow packaged installer.

5

Windows / Re: All in one win32 build script

« on: July 14, 2009, 06:12:44 am »

your all in one going to be able to do language files as well?


also if i am reading things correctly your using 7zan.exe?  which is a version of 7-zip and included in the codeblocks pre-package? according to the 7-zip the file is no longer included in 7-zip and hasn't been since 2003?
also I just got the latest codeblocks package and it does NOT include 7zan.exe,  7za.exe is included and should do the same thing.

6

Discussion / Re: New soldier stat increase system

« on: July 12, 2009, 11:11:29 am »

dead topic.    it's already figured out no need to waste time on it.

7

Coding / Re: compile_po.bat error

« on: July 12, 2009, 11:09:13 am »

I just got it off MATTN's site.   fresh download within the last week.

8

Feature Requests / Re: Soldier selection/order

« on: July 11, 2009, 01:31:45 pm »

still looking for the "shave and a haircut" button....   I can't say what it actually does since this is a public place

9

Bugs prior to release 2.3 / Re: Bugs are starting to get fewer yes?

« on: July 11, 2009, 06:00:24 am »

just many of the same bugs.   We know what they are and where they are so there isn't much point in reporting them.
slowly things are getting cleaned up.

10

Coding / compile_po.bat error

« on: July 10, 2009, 05:50:56 pm »

when i run compile_po.bet with the /ufoai switch i get the following error "the procedure entry point locale_charset could not be located in the dynamic link library libintl3.dll."  any guesses?

11

Discussion / Re: New soldier stat increase system

« on: July 10, 2009, 11:44:34 am »

lol realistically its the easiest to code. and incorporate,  which really is the ultimate test.

Just set-up a function call once a month to set the training skill level for the month.  I wouldn't even worry about "enrolling" soldiers in training since the reality is everyone would be swinging by to train in periods between missions. Then at the same time compare skill level of training center to trainees to determine amount skills are improved.  Design a function to put the amount improved on a curve so that those with the most to learn, learn the most and those with not much to learn get very little.  your other veterans  might get a little boost  but not much from being in training.  There should be a fairly clear point in the improvement curve for individual soldiers where it becomes obvious training center improvement will never replace real combat experience for improving your soldiers.   So for the rookies its great but for the vet's it is of limited benefit.  and good veterens improve the skill improvement of the rookies.

the above letter can be easily re-written from Navvare's or the head engineers perspective complaining about the soldiers underfoot trying to train....  eh Winter?  Though I kinda like it coming from a soldier.  These guys are the ones in the fight. and while they aren't as brainiac as others,  they ARE the subject matter experts in what they do...  killing aliens.  Reality is soldiers train all the time.  especially in the more elite special forces units.   so the subject of looking for a good place to train would be soldier driven.

12

Discussion / Re: New soldier stat increase system

« on: July 10, 2009, 10:15:52 am »

T3 Center (Tactical Team Training Center).

For your consideration:

I am a combat soldier and I won't mince words. We need someplace to train.  We have learned a metric ton about fighting the aliens but need a space to call our own so we can hone our skills.  Up until now we have just been using the rooms and hallways around here to try and get the rookies up to speed.  It really isn't working, we are doing everything we can but the rookies are still getting hammered out there.  We have had to do it wherever we can find space and we aren't underfoot. My guys tend to bump into things, sometimes sensitive things, while moving around and there are places where that isn't a good idea,  strangely those places also tend to be the only places we can find enough space to train.  Navvare says one of my guys tripped in a lab and nearly upset some lab equipment and set the research team back several months.  They managed to rescue the specimens in time though. He talked to the engineer and he agreed since they are tired of finding us crawling around the workshops and running into gun barrels in the dark corners whenever we bump something out of alignment and they have to go in and fix it. The two of them put their heads together and think they can rig up some fancy computer controlled and simulated shoot house of sorts for us.  I won't say i understood half of what they went on about,  but they say they can rig it up to handle some live fire training too.  For those things it can't handle, they can rig simulated weapons that it can handle.   It will also have the capability to do variable terrain and whatnot.  The engineers say they will be happy to rebuild the thing if we manage to demolish it.  After all, a few hours rebuilding plywood walls is nothing compared to the hours spent troubleshooting and recalibrating the power plant anytime we bump the wrong thing doing what we are doing now.  The scientists say that based on the recordings from the mission recorders we all wear during missions they can rig holograms, projectors and whatnot to re-create those missions for us to train with.  Claim they can rig it up as a controlled virtual reality experience or something.  I won't claim to understand it, but i believe they do.  Quite frankly, We don't have a good place to hone our skills.  And if this thing they have in mind can do half of what they say it can, it may be just the thing.  I am tired of losing soldiers and if this thing helps us to get the training we need to keep from getting blasted.  I am all for it.  Truth is we have a lot of experience on the team already and we can rotate through the Training Centers as our own instructors.  We can be better, We can stop losing rookies every time we go out,  we just need the right place to get better.

yours
Da Gruntz!

sounds like option A above was the most liked idea.  Helps to solve the rookie getting up to speed problem especially and does what we want a training center to do, slightly increase the improvement of all soldier skills.

13

Discussion / Re: New soldier stat increase system

« on: July 09, 2009, 10:05:10 pm »

recruits aren't going to change in quality...   already recruiting the best of the best.   this gives you a diffrent method to get them up to speed that will scale with the game.  and it is both realistic and reasonable.   Units teach themselves.  somebody does something well so he gets called on to teach it to others...

can make it as complicated or simple as you like.  or even code it in stages getting more complex

a.poll existing soldier skills to create "ghost" instructor whos stats are equal to the best in each catageoy found in all your soldiers... and improve all skills according to the scale i mentioned.  the bigger the difffernce the better the bonus.  the closer student gets to instructor the slower the gains.

b. pick 1 soldier to teach at a time...  can use all his stats and do general instruction with tiny bonuses to everthing or just pick one stat and do fucused instruction.  this mon we are teaching assault,  next month we will be breaking out the heavy weapons...  in which case the improvements would be greater but limited to 1 stat.

c. heck you could also assign every recruit their own mentor...   but thats probably more complicated then it needs to be. and way more coding then is needed....


i do like option B

14

Discussion / Re: New soldier stat increase system

« on: July 09, 2009, 03:19:54 pm »

Meh, just assign a more experienced soldier to the Training Lab as an instructor.  like we were with doctors and the hospitals back in 2.1

training improvement is based on difference between instructor and student stats.  the greater the difference (when a student score in a stat is lower then the instructors corresponding score.) the greater the improvement.   as students get closer to the instructors scores the improvement tapers off.....
lets you use your veterans to bring the rookies up to speed.

thus as the game progresses and as your instructors get better and gain experiance they will be able to get the rookies up to speed faster.

if you don't want to deal with actually assigning instructors.   Just have the game do a periodic check of your soldiers skills and set the level of instruction at the highest level a soldier has in that stat.   

this actually simulates what we do in the military,  we are ALWAYS taking time to teach each other skills and work on our training.  It's all about teaching each other.  and considering that Phalanx is recruiting the best of the best it makes sense that they would be teaching each other and giving each other pointers to keep themselves at their best.

and per the current written proposal...  have a credit cost to cover operational costs and instructional materials.

15

Windows / Re: Map Files for Phil (/base/maps as of Revision 25053)

« on: July 08, 2009, 05:57:46 pm »

Hey odie thanks for all the files.  TSVN makes it to the end and tells me i am updated now.
thank you, thank you, thank you.  it's only taken me several months to get a completed checkout.
now i can begin to dig a little deeper into the AI coding.