All za small things...
 

Welp, its brobably more cosmetic than something really important... but
Anyways, id's:

1) Ability to create jump paths manually (sometimes its just plain pain
in the ass to "precisely" land on given wp i.e. jump from trash
tank to building on cs_siege - wiz SoulFathers wpack bot's just
got stuck there trying, had to fix it myself - it WAS pain in
the ass, nope - i can jump threre noprob, but making connection
right...)

2) Upon creation of new wp:
Original algo just tries to connect all wp's in default rad to
new one (with some checks).

Idea is to add another small check:
If there is a set of wp's which all lay in say 30 degrees
sector - make connection only to closest one of them.
There is eight outgoing paths limit hardcoded,
sometimes its just not enough. This will save some of em
for something more important.

Saves some bit more of nerves...

3) WP's targeting!!! Got tired of that secret spot hunting (i mean wp edit
mode)...

4) Weapon usage effectiveness statistics based weapon usage :)
(stored in EXP file probably)

5) Adding overall visibility coefitient of waypoint to its properties
(number of wp's visible from this wp). So weenie bots can choose
path right (btw more expirienced players try to use paths
on wich they have less probability of being sighted).

6) Is there any sniper specific code? Like using camp spots with longer
tracers fired from binding angles (uh, pardon my english).

Anyway, the bot, even at the current state IS heluva work done right
(not mentioning that its non-commercial product done by several
enthusiasts connected only by net).
Keep eed up, doode(s)!

Uh, alchogol is leaving my system, and so should probably i...

Re: All za small things...
 

I don't understand why you have problems... if you can jump just right, you should be able to make a jump connection as well. ???:( Can you explain what you mean a little bit more in detail? I'm not sure if I got your point...

Hm, that's actually 2 ideas in one: a) the new check for WPs inside a certain angle and b) a higher number of connections allowed per waypoint.
I like the first idea (the new check) a lot, because it would do automatically what I often do by hand. That would really be a nice feature IMHO!
However, if this were implemented in the dll, I don't think that it would be necessary to raise the max number of connections allowed per waypoint.

Do you really think it's that bad? Thanks to the new markers, I'm getting along with this aiming system quite well, although I agree with you that it's not perfect.

Errrr.... come again? :D I really don't quite get what you mean... Do you want to store info like "AK has been very efficient around WP nr. xxx" in a separate exp file? If this is what you mean, it might actually help bots to buy weapons that are better suited for a map. For example, (sniper) rifles in large open maps and SMG's and shotties in very small & narrow maps... This could be worth a thought or two.

I'm not sure if it would be possible to do that without giving up compatibility to PB 2.5 waypoints... then again, it could be discussed whether this compatibility has to be kept at all costs, but that's quite a big issue.

???:(

Well, anyway thanks a ton for your feeback and for some nice ideas! =) Let's see if we can make something out of them.

Re: All za small things...
 

Pardon me bad english, gain...

>> I don't understand why you have problems... if you can jump just
>> right, you should be able to make a jump connection as well.
>> Can you explain what you mean a little bit more in detail? I'm
>> not sure if I got your point...

Erm... Dunno haw tah explain eet... :)
As was told, try cs_siege with SoulFather's wp-pack.
Look at how CT bots try to jump on building. Try to fix it youreself.
Or maybe its just low-fps prob...
Manual jump connection would just ease wp editing in some situations.
Nozing more.

>> Hm, that's actually 2 ideas in one...

Nope, its one idea. The designated Idea-B part ;) is more like explanation
of why that new check is needed. I agree that eight connections are
enough (most of the time). But sometimes...

>> Do you really think it's that bad? Thanks to the new markers, I'm
>> getting along with this aiming system quite well, although I agree
>> with you that it's not perfect.

No, not that bad... But trying to aim to point that is somewhere in
32 (world units) rad off the center of waypoint, again sometimes, IS
PAIN IN ZEE ASS. Just sometimes.

>> Errrr.... come again? I really don't quite get what you mean...
>> Do you want to store info like "AK has been very efficient around
>> WP nr. xxx" in a separate exp file?

Coming again... ;)
No, there is no need to store it for every wp. The idea is (as is):

Example of statistics:
Statistics for cs_office
bot carrying AK47 killed 43 times, died 24 times
bot carrying AWP killed 9 times, died 67 times
...

Its easy to collect/enterpret such stats, dont u think?
(simple && effective) = Simply effective ;)
So there is finally no bot's with AWP's on cs_office! :)
And no, no separate exp file. There is already one exists. Why not to
store all EXPerience data in it? Isnt this is what it was created for?
This could worth even three thoughts!

>>>> 5) Adding overall visibility coefitient of waypoint to its properties

>> I'm not sure if it would be possible to do that without giving
>> up compatibility to PB 2.5 waypoints... then again, it could be
>> discussed whether this compatibility has to be kept at all costs,
>> but that's quite a big issue.

There is no "compatibility issue", period
There was NO vis calculation in 2.5.
There was NO *.vis files in 2.5.
It all began in 2.6.
And vis-files in 2.6mm.
Even if there was one, bot IS evolving.

Always wanted to ask SpLoRyGoN about his goals.
Path nr.1: continuation of original source ideology, pushing it to the limit.
Path nr.2: making it CS_Bot_of_my_dream (whatever it takes).

What's his?

Sorry for offtopic...
Simple (simply effective, remember?) :)
Again, idea (as is):

float temp_overall_vis_coeffitient[NUMBER_OF_WAYPOINTS];
for every wp(1)
{
int overall_vis_coeffitient = 0;
for every wp(2)
{
if (there_is_trace(wp(1), wp(2))) overall_vis_coeffitient += 1;
}
temp_overall_vis_coeffitient[wp(1)] = overall_vis_coeffitient;
}

This all can even be stored in float (0..1) for simplicity's sake
(0 - complitely invisible to any other wp, 1 - maximal visibility to other
wp's):

for every wp
{
wp.overall_vis_coeffitient = temp_overall_vis_coeffitient[wp]
/MAX_VIS_COEFF_OF_THEM_ALL;
}

This coeff can be used in A* or any other proc aswell.

Btw: why not to make generic A* dll or static lib (optimized ofcourse w'MMX
n stuff).

Btw2: i could do it... probably (full day work'n stuff, aint got much time,
but still got some and wanting to use it to recall my prog skills).

Btw3: watz wiz CVS. Couldn't connect to it. Some error wiz localhost or
something...

Btw4: (guess what) there is vis-file :) ...

>>>> 6) Is there any sniper specific code? Like using camp spots with longer
>>>> tracers fired from binding angles (uh, pardon my english).

>> ???

It depends on effective sniper_camp distance (say 1024 world units).
When u create camp spot u make it by creating first direction and
after that - the second. Now fire tracers from them and remember distances.
Then add them together and divide by 2. This is the distances to choose
from to sniper. The longer, the better. Think - if u bought AWP - u're priority
one would be to camp at spot with bigger depth of FOV. Hell, u're sniper,
so snipe on BIG distaces, right? Yes, this explanation is primitive. Cant think
of anything better now... sorry...

//================================================== =====================

As i told, its all more cosmetic thingys than something realy important.
The ideas already implemented are great and work most of the time.

MOST OF THE TIME.

I dont know about you, but i got that feeling that most of them are kind of
unfinished. We got wp highlighting, but u still have to search for that spot.
We got experience collecting system, but how many thousand times bots have
to die to understand that it is BAD to jump on that wp - guns akimbo.
Bots can buy sniper weapons, but can't properly use them (and most important,
how to camp wiz zem right). Bots can buy shields, but cant use them right...

And this list can go on and on...

Right now (guess what) i'm too drunk to continue... no beer pleaze...
Slipknot - rulez...

Re: All za small things...
 

Well you have to relize in most maps soulfather did he made jumps that are hard for people to make!!! So the bots have even more trouble, the whole point of you making the jump while waypointing is to make the jump as realistic as possible, IE if you waypoint the jump automaticaly from a trashcan to another then it will be realistic if you do it manually it maybe really odd looking and even an impossible jump!

Actually this is incredobly hard this would pretty much require a whole rewrite of the code

Splorygon isn't doing podbot anymore, well at least thats what he said he has to focus on school and other various things

Re: All za small things...
 

the visibility table has existed in CF's version of pb, I dunno when it started but it did exist in 2.5.
it's just that in pb2.5 it didn't save the visibility table and would rebuild it every map start...

whats up with that???
Are you saying open source programs aren't good???
Well my choice is both paths.. I want to make a very humanlike bot and also have it open source...

Zero would never exist.. if it did then theres a bad waypointer afoot! :D

Re: All za small things...
 

to biohazerd87:

I understand that many things depends on how waypoints done. I understand
that jump-auto-wp-placement works actually as a failsafe - if human could
jump there then bot will be able to do that. But i disagree that all MUST
depend on it. IMHO main goal of waypointer is to "show" where to
go, where to camp and where to jump, but bot must make descisions depending on
situation on when and how to do that. Realistic behaviour must be in code
not in waypoints. Why not to make "hard to reach" flag to signal bot
that this certain wp can be reached only by high level bots. WP edit must
easy and fast. Yes, noobs with this manual jump connection would be
a "clinic case" :). But to expirienced waypointer it will be just another
instrument.

>> Actually this is incredobly hard this would pretty much require a
>> whole rewrite of the code

There is no need in rewriting a whole code. Init/load data on map start.
Update it in every next round. Save on map change/disconnect. Not THAT
easy ;) but still...

to SpLoRyGoN:

>> the visibility table has existed in CF's version of pb, I dunno when
>> it started but it did exist in 2.5. it's just that in pb2.5 it didn't
>> save the visibility table and would rebuild it every map start...

It all (vis calc) started in that 2.6 "hdd crash" version. In 2.5 there was
no vis calc. Im tellin' ya :).

>> Are you saying open source programs aren't good???

No, i'm not about open source. I like open source ideology. I just wanted to
know, will u implement something new to bot or just finish what was started
by Count Floyd.

>> Zero would never exist.. if it did then theres a bad waypointer afoot!

It doesnt matter. This koef is just a relative measure of how visible this
wp to others. Yeah, i'd wanted to see how that "waypointer" would do that :).

P.S. Anyone can help me with CVS?

Re: All za small things...
 

I'm not sure if I "can" implement something new..
and also that jump from trashcan upto the building does not require a jump connection.. all you do is connect the 2 and if the bot can't reach it he'll jump by himself.

you have any sort of messaging system or irc client? either goto my profile and get my SN's or connect to BU's IRC server and I'll help you set up CVS

Re: All za small things...
 

to SpLoRyGoN:

Yep, i know that. Dont know what situation in current version, but in 2.5 they tryed to jump only on first connection, after that they just tryed to walk there without jumping...
Again sorry for bothering...

Re: All za small things...
 

about SoUlFaThErs "inability" to get a bot up that building side in cs_siege.

my original conclusion was that a bot with the wrong weapon wil never make certain jumps because he is too heavy. there are many jumps that arent possible if a bot carries awp or para.

so i had asked PMB to force the bot on a jump waypoint to take out his knife before jumping, and take out his weapon after.

he decided to add a distance limit. which changed my original forced idea.

adding those jump paths manualy will cause MORE problems than the way they are done.

The real thing should now happen like this if the jump/knife things staays the way it is:

the bot should have a check on how many attempts he made to reach a specific jump-end waypoint. if he say fails 3 times........he searches for another route. this would get the problem more realistic. most players wont try more than 3 times to do something special like that.

proving the knife out theory is how i see bots doing a jump i made on that stack of boxes on the CT side of the bridge in de_dust. they jump from the high boxes(on the left) all the way across the bridge underpass road to the boxes on the other side(on the right).

that jump is 10 times harder than that shit in siege. there are actuall FEW players who know how to make that jump. i got the bots to do it if they had a knife out. so its really not me the waypointer not paying attention........its just that bots dont do EVERYthing....see elevators navigation.

and why change the frickin coding just because of one damn jump. if you got it fixed than hey........be happy. :)

Re: All za small things...
 

that reminds me SF... you should remove that jump in de_dust.. the CTs think it's so much shorter to take the jump but I've only seen a few make it.. the rest of them lose 30 hp.. that jump covers a lot of distance and they take the path with the least danger + least amount of waypoints... that jump is like the only path they want to take..