Preemptible RuneSpan

Quote:

---

Originally Posted by Aufi

Hi zmon, I just tried the script and I ran into an issue, the bot doesn't siphon the highest thing possible. I'm on the 3rd floor and tried several platforms, but the bot starts siphoning regular Esswraiths instead of stuff like skulls etc, which are far better XP.

Maybe this is a bug in the latest version? Just thought I'd let you know.

---

the models are not complete yet, i know that skulls in the release were not tested, but i did get a screenshot of skulls recently so i'll model them

basically the models rely on overlapping tolerant tpas so i need to find few colors that are good representation

the difference between using single color with very low tolerance and several with slightly higher tolerance is that if you chose the colors wisely, you should be able to eliminate almost all false positives, problem is that i don't know if there is an extension that would scan area of image and compare colors to the whole screen to come up with few relatively unique colors to use

i will search for it today and if i won't find it, i'll have to write it myself

Quote:

---

Originally Posted by zmon

the models are not complete yet, i know that skulls in the release were not tested, but i did get a screenshot of skulls recently so i'll model them

basically the models rely on overlapping tolerant tpas so i need to find few colors that are good representation

the difference between using single color with very low tolerance and several with slightly higher tolerance is that if you chose the colors wisely, you should be able to eliminate almost all false positives, problem is that i don't know if there is an extension that would scan area of image and compare colors to the whole screen to come up with few relatively unique colors to use

i will search for it today and if i won't find it, i'll have to write it myself

---

I've also seen it siphoning a Choas hound instead of a blood pool, although the paint indicated it was attempting to siphon the bloodpool.

Quote:

---

Originally Posted by Aufi

...stuff like skulls etc, which are far better XP.

---

i just remembered -if you are using release prior to 22, which does include 21, there is a problem sometimes reading inventory amount of essences, so in one of the most recent version, i got the call to SRL replaced by my own function

it's not that srl doesn't do it right, it's that srl expects non-tolerant DTMs for inventory and it gets confused, or something like that, whereas the DTMs for essences that i have, they all have built in tolerance

why is essence level important?

well in order to prevent essence drain, if you are under 100 essences, all searches for better target are suspended, if you are siphoning a creature. That is because upon destruction of creature, you get bunch of essences.

if the essences are under 100 and you are siphoning node, the main loop will start looking for creatures only, so that it can lock onto them

============Rough Stats etc...============
Time Running: 8 Hours, 3 Seconds
XP Earned: -3726340, per hour: -18504
Points Earned: 1057, per hour: 132
Preempted Counter: 45, clicks: 29
Current Target: HOUND - NATURE, 43.5xp, level 44
Current Runecrafting Level: 53
Essences Available: 1121
Current Floor: We are not in RuneSpan

I left it on while i was at school and abit longer :o

ok after profiling the code i have found out that some models just are very slow to evaluate, for example the currently disabled fire essling on first floor generates thousands of matches in one of the color, but if another color is not present at all, i still do process the data, this of course is not so great to do in script so i will change the density discovery to, rather than locate and calculate all matches within each overlay, will test existence per look up location in all color channels and will process data only if possible match found

the call to discover if tpa color with tolerance matches on given region is much faster than evaluating all matches, and if we fail to match all color overlays, we will not even need to evaluate anything

if the procedure gets stuck processing vast number of matches, the result is that it becomes unresponsive to new nodes popping up in timely manner, plus it eats cpu when it probably doesn't even need to

working on new density search now

Quote:

---

Originally Posted by zmon

This is something you may have never seen before :P

i know none of the scripts that i looked at implement it

idea is simple - multi color matching with density search

what can it do?

extremely high accuracy near real time target detection

wanna proof? :P

http://oi46.tinypic.com/2v7tkxd.jpg

wait for next version ;-)

---

In effect you are merging multiple color teas, and finding the largest? A very effective way to search :)

http://i1196.photobucket.com/albums/...ens/kkkkkk.jpg

Started this run at around level 60/61. Found v21 to be much faster then previous versions.

it keeps trying to run to the south on mini-map

This kid is such a leecher!

Pathetic!

You could double the XP per hour by simply implementing some sort of search & priority function that prioritises the best experience, e.g. a shifter rather than a water pool, then yields that.

Sometimes you just hop back and forth from platform to platform too.

well that's the whole point of checking full range of available options

but last version using dtms for node and creature matching is 20, the 21+ version is work in progress mostly because the interpretter is dog slow and i am looking into what parts of code might be written differently to use some of the functions that are provided in native code form and therefore run much faster.

24 is the latest release - the seach is much faster, in fact so fast that I don't generally get "warning" messages in log that something took more than 80 milliseconds (0.08s) to search, compare that to several seconds when trying to match with 23, and you will understand why it was so sluggish in determining that better node is available.

Ok in this scenario, we were siphoning wraith, who broke down. Then we found something else and as soon as wraith came back, we switched... that means we started siphoning (there is human-like wait before next review of screen after initial siphoning), then we found our better target, moved the mouse there and made sure we had siphon option and clicked and verified that the click was successful (which alone takes 1 second) and all in under 3 seconds

when the search doesn't take ages, it performs better


Creature most likely broken down
============Rough Stats etc...============
Time Running: 3 Minutes and 30 Seconds
==========================================
LOOKING FOR Vine TOOK 78ms and got 2
============Rough Stats etc...============
Time Running: 3 Minutes and 37 Seconds
Current Target: Vine, 33XP, level 17, XPTIMEOUT 5.500s
==========================================
Incremental xp: 34
LOOKING FOR Blood esswraith TOOK 93ms and got 2
LOOKING FOR Blood esswraith TOOK 93ms and got 2
============Rough Stats etc...============
Time Running: 3 Minutes and 40 Seconds
Current Target: Blood esswraith, 73XP, level 77, XPTIMEOUT 7.936s
==========================================
Incremental xp: 31
Incremental xp: 1
Incremental xp: 1

Thanks for updating the models for death wraith's its now finding them :D
Just needs to target switch from monster to nodes.

Did you implement jumpers?

Edit: it went from death wraith to nebula when it spawned very fast!

Implement jumpers!

do not know if the problem is with the script or what, i was running fine on 17. but now when i hit start, smart takes forever to load and i get an error saying something like: smart could not load, make sure decloration page. or something. what could it be?

Quote:

---

Originally Posted by Trill 3nt PK

do not know if the problem is with the script or what, i was running fine on 17. but now when i hit start, smart takes forever to load and i get an error saying something like: smart could not load, make sure decloration page. or something. what could it be?

---

Ctrl+F, search for Smart_Server, change the server # to a different world. Save and try it again.

If it doesn't work load runescape in your browser.

sorry, which link should i download?

The latest one, revision 24. Its very smooth.

not sure i have done it right but here is my prog....i hope lol

I will be using this script again! :D
I hope to get 99 by two weeks! <3
Love the updates!

How do i run this in msi, i cant seem to find it on the script list :/
PLEASE HELP A.S.A.P!!

Quote:

---

Originally Posted by owned u stoned

How do i run this in msi, i cant seem to find it on the script list :/
PLEASE HELP A.S.A.P!!

---

you don't run it in MSI... It's a stand alone script.

Quote:

---

Originally Posted by owned u stoned

How do i run this in msi, i cant seem to find it on the script list :/
PLEASE HELP A.S.A.P!!

---

LOL made my day.

This is actually a very well made script.

Well done.
+Rep

God. It works flawlessly! In all the F2P Runespan Siphoning this is the BEST!

Great xp and script! Good job

Can you make this take rune ess when its out? It keeps clicking the esshound even though it has no essence.

Quote:

---

Originally Posted by Darcy_Oner

Can you make this take rune ess when its out? It keeps clicking the esshound even though it has no essence.

---

that will be coming up soon, but for now what it does is that i fixed the logic about remaining rune essences

revision 26 should never run out of essences - uploading now

What does the density model do anyway? Care to explain a bit? I am a bit lazy on the code ;)

Quote:

---

Originally Posted by zmon

that will be coming up soon, but for now what it does is that i fixed the logic about remaining rune essences

revision 26 should never run out of essences - uploading now

---

Thankyou very much =]

well the simple way to look at it is like this - the screen is fractioned similarily to the way video compression algorithms do, but due to size of target, i work with 32x32 area on 16x16 grid, so i get every block of 16x16 accounted for by 4 different 32x32 areas - kind of like this:

http://oi47.tinypic.com/5ugxuu.jpg

one square is marked with 4 colors of the larger squares

the assumption is that you chose colors somewhat representative of the object so i calculate for every square how many of each color is present

the object needs to match every color with its tolerance, before it is considered to be a square to look at

if the tolerance is high, there is automatic high threshold for marking square as fitting the color

the colors per channel are then "geographically" averaged to give a center location within square with highest density, so if matching colors are only in one corner, that's where the density ends up calculating the center

now... the optimization of course puts everything upside down, and instead of looking at densities of each color channel, i have to evaluate if there is a point in even looking at individual points, that saves a lot of time because i can use tpa with color tolerance, the function that returns if found - because it stops searching after first hit, so if i do find representative color, it saves time

and of course if any given 32x32 square does not have one of the colors i need, the rest should not need be evaluated


of course once i put in bayesian filtering, it will be even better - for example, if you chose 2 colors for water pool, they both are present in, for example, water essling

but if i could say waterpool is these two colors but not that odd purplish pink color, then water essling will no longer match waterpool, but that requires a little bit more coding

but it is inevitable - matching fleshy growth is nearly impossible if you want good tolerance built in

and forgot to add - some of the models i am making, do have to have more than 2 colors, so that i have both tolerance and don't click on too many bystanders.. that's really the main issue - clicking on things that we shouldn't click on, when using only one color

version .20 works but the newest is broke, it is also extemely slow.

Quote:

---

Originally Posted by crazyidkdog

version .20 works but the newest is broke, it is also extemely slow.

---

try to be more specific, such as:

1) which exact revision were you running
2) which floor
3) what kind of possible targets were on the screen at the time
4) did you have Member := True, or False setting (setting member to false will pretty much not allow you to target anything but few esslings on 3rd floor as most creatures and targets are member only)
5) what was the RC level at the time
6) was the camera angle at the time high (from above) or low (from side)

The bot is great and everything works as it should, what I suggest for the anti-ban is that the arrow or the pointer should leave screen during idle because this is what most afkers do. That and it being able to jump platforms for better would be perfect. But it's already great as it is. Tyvm

Peace - JackPot

Quote:

---

Originally Posted by JackPot

The bot is great and everything works as it should, what I suggest for the anti-ban is that the arrow or the pointer should leave screen during idle because this is what most afkers do. That and it being able to jump platforms for better would be perfect. But it's already great as it is. Tyvm

Peace - JackPot

---

I have been working on changing targeting models and rewriting main routines, which means that the auxiliaries were mostly left unchanged

such as initial essence gathering, or specific essence gathering for platform hops...

as for the antiban, i'll look at the SRL antiban library to see if there are some canned antibans i can just drop in, that are tried and true

Quote:

---

Originally Posted by zmon

try to be more specific, such as:

1) which exact revision were you running
2) which floor
3) what kind of possible targets were on the screen at the time
4) did you have Member := True, or False setting (setting member to false will pretty much not allow you to target anything but few esslings on 3rd floor as most creatures and targets are member only)
5) what was the RC level at the time
6) was the camera angle at the time high (from above) or low (from side)

---

the bot just dousnt start, heres my log.

[Hint] C:\Simba\Includes\SRL/SRL/core/animation.simba(257:3): Variable 'X' never used at line 256
[Hint] C:\Simba\Includes\SRL/SRL/core/animation.simba(257:3): Variable 'Y' never used at line 256
[Error] C:\Simba\Includes\srl/srl/misc/paintsmart.simba(560:12): Unknown identifier 'TPAFromEllipse' at line 559
Compiling failed.
A new update of Simba is available!
Current version is 980. Latest version is 984
[Error] (41:3): Unknown identifier 'SmartSetDebug' at line 40
Compiling failed.

Quote:

---

Originally Posted by crazyidkdog

the bot just dousnt start, heres my log.

[Hint] C:\Simba\Includes\SRL/SRL/core/animation.simba(257:3): Variable 'X' never used at line 256
[Hint] C:\Simba\Includes\SRL/SRL/core/animation.simba(257:3): Variable 'Y' never used at line 256
[Error] C:\Simba\Includes\srl/srl/misc/paintsmart.simba(560:12): Unknown identifier 'TPAFromEllipse' at line 559
Compiling failed.
A new update of Simba is available!
Current version is 980. Latest version is 984
[Error] (41:3): Unknown identifier 'SmartSetDebug' at line 40
Compiling failed.

---

I don't know if this is the problem or not, but you have an old Simba version. Update Simba and try again.

I can see how regularly you're revising this. Thanks for your efforts.

This script is great. The only problems I have are very small. The bot doesn't seem to siphon skulls at all, sorry if you've already mentioned this. So the bot is often siphoning half the xp it could be. Also, would it be possible to disable use of platforms? There is an island I like using and the bot quite quickly leaves the island for some reason, despite the fact that there are things to be siphoned there.

this is an awesome script...
definately recomend over kyles, seems to be more efficient at changing targets too.

To be honest, i'm gonna have to say bye bye to Kyle's for now.
As soon as I get my laptop back, i'll be running my skiller on this :$