ENs - Subs - Buff/Debuff Bar Subs. (oEUO & EUO)

[Scrripty]

I think it's when ther are NO icons on the bar.   It's scanning and finding a positive in the actual game world is my guess.

[Endless Night]

I think it's when ther are NO icons on the bar.   It's scanning and finding a positive in the actual game world is my guess.

Umm no icons on the bar... I was presuming it was the last icon... but i guess could be first icon as well, what is the icon its giving as a false positive ?

[Scrripty]

No clue, I'm client swapping when it does it, and it does it on a random char.  I haven't really wanted to debug it enough to find out.  Works good enough for me.

[Scrripty]

I figured out the double apple issue I was having.  On each loop of the script, it's checking %buffbariconnames for both chars before it gets reset, so it was eating on every character before resetting the %buffbaricons to _ or the next chars icons.  So I had to reset the variable every loop of the script to fix it.   Now I just set %buffbaricons to _ on each loop after checking to see if anything is in that chars buff bar.  Fixed it up nicely.  Might want to mention that in this version as it swaps now and might mess someone up.   Also, here's a few hard numbers from my script.  Average swap between 2 characters while healing/toggling specials/scanning buffbar/finding and attacking spawn/eating apples/drinking pots... 30 ms over 133700 swaps.  I scan the buff bar on every character once every 1.5 seconds, and that average is for 2 characters.  I get the average only after it swaps to both chars.  Something I noticed tho is that calculating the average for one swap, or two, makes no real difference in swap time, and actually in most cases calculating it for 2 chars made it actually DROP over 1 swap.  Interesting.  In typing this I'm not at 140k swaps and it's still 31 ms...

[Scrripty]

Thought I'd post some numbers on the version I wrote and sent to EN for fun.  I wont be publishing it at EN's requrest, but thought it would be fun to see some numbers.  Keep in mind, EN's has a TON more functionality atm.  Mine just scans the buff bar facing right only for maximum speed.  Here's my numbers for 10 second runs scanning the bar, this is a modified for speed version of what I sent you EN.

Code: [Select]

NO icons on buffbar:

395 scans in 10s.  Avg. scan time: 25ms.

1 icon on buffbar in FIRST scan position: (meaning fastest scanning possible for one icon)

390 scans in 10s.  Avg. scan time: 25ms.

1 icon on buffbar in LAST scan position: (slowest scanning for 1 icon)

284 scans in 10s.  Avg. scan time: 35ms.

Multiple icons on bar in "real world" configuration.
prot, reac, night sight, bless, corpse, reflect, clumsy, invis

160 scans in 10s.  Avg. scan time: 62ms.
In testing, you really only need to scan the buff bar once every half second or so to catch everything extremely quickly on the bar.  You normally can't be affected by something on the bar faster than half a second.  Takes a sec for something on the bar to take effect in my experience.  Both scanners are plenty fast enough for general use tho.  I just have to have blazing speed for my personal use.  I never get past 3 or 4 icons before my scripts eat an apple so speeds stay well below that last scan time.

[Scrripty]

One other thing, not that it matters with oeuo around the corner, but you do both your savepix commands before you start scanning, which equals like 7000 savepix commands over 10 seconds of running the script, you can cut that by a significant amount by putting the second savepix inside the while loop in the buffbarscanner.  It comes out in a rough estimate to be 3500 + 850 or so scans, saving a LARGE number of savepix over time.  Gained me about 5 scans per 10 seconds.  Not a lot, but when you like things fast, definately makes a difference.  It all depends on scanning order also, if you optimize the secondary scans based on most used and least used, you'll get a much lower number of savepix uses per scan also.  Just a thought.

[Endless Night]

Interesting thoughts Twinkle McNugget.  I had some other notes on stuff i was planning on testing before i went on my current UO break .. when i get back into it I'll exmaine those and your inputs here more closely. 

Have you ever tried out my OEUO version.  It might be one logical step behind my current EUO version.

[Scrripty]

Yea I was also playing around with scan order last night a lot just for fun.  I organized the icon information into groups by their pixcol, there's only 7 groups of icons if you organize them by their pixcols.  Could probly leverage that for some speed too.   I'm gonna add some if this pixcol, scan these ranges code, see if I can get the per scan numbers way down that way.   Instead of scanning ALL the icons, if the pixcol is within a known range, you just scan those icons, if not there, it's not an icon, and you only did one pixcol and 1 to 5 scans after the initial one.  Instead of scanning ALL icons to find the correct one. 

[Scrripty]

The benefit of what I worked on last night.   EN you should understand where I'm going with the below code.  It might make initial scan time jump a bit, but the overall scan times will come WAY down.  Like that 60+ ms per scan time can be brought down a ton when scanning 8+ icons on the bar with some basic decision making code.  Fun stuff.  This allows you to never have to scan more than like 5 icons to find the correct one.  AND you can organize the 5 groups by most likely to be found, so in reality you're never scanning more than 2-3 sets of variables to find the correct one.  And only doing a total of 6-13 pixcols per scan to find the correct icon.  Should save a ton of overhead.  The scanHigh scanLow vars just set the for loop for scanning the correct icons based on the found pixcol.   I hate you for getting me started on this EN. heh

Code: [Select]

  set !scanStart1 -950
  set !endScan1 1164
  set !scanStart2 24320
  set !endScan2 576074
  set !scanStart3 1051755
  set !endScan3 1053844
  set !scanStart4 1070608
  set !endScan4 4368697
  set !scanStart5
  set !endScan5
 
  set !scanLow1 1
  set !scanHigh1 5
  set !scanLow2 6
  set !scanHigh2 17
  set !scanLow3 18
  set !scanHigh3 26
  set !scanLow4 27
  set !scanHigh4 35
  set !scanLow5 36
  set !scanHigh5 42

[Scrripty]

This is data from the new version using the code above.  Just posting cause it's interesting.   Sorry for cluttering your thread EN.

Code: [Select]

NO icons on buffbar:

OLD:  395 scans in 10s.  Avg. scan time: 25ms.
NEW: 669 scans in 10s.  Avg. scan time: 14ms.
NEW: 899 scans in 10s.  Avg. scan time: 11ms. (kinda wierd...)
# of savepix 1: 1798 in 10s avg.
# of savepix 2: 0  (second scan makes sense now and the speed...)

1 icon on buffbar:

OLD:  390 scans in 10s.  Avg. scan time: 25ms.
NEW: 585 scans in 10s.  Avg. scan time: 17ms.  Bar in black area.
NEW: 532 scans in 10s.  Avg. scan time: 18ms.  Bar in game window.
Note: Scans sometimes drastically slower with bar in game area.
# of savepix 1: 5364 in 10s avg.
# of savepix 2: 1788 in 10s avg.


Multiple icons on bar in "real world" configuration.
prot, reac, night sight, bless, corpse, reflect, clumsy, invis

OLD:  160 scans in 10s.  Avg. scan time: 62ms.
NEW: 263 scans in 10s.  Avg. scan time: 38ms.  Bar in black area.
NEW: 258 scans in 10s.  Avg. scan time: 38ms.  Bar in game window.
# of savepix 1: 3980 in 10s avg.
# of savepix 2: 5771 in 10s avg.


[Scrripty]

All those tests were run at 1600x1200 resolution in 32 bit mode.  After setting my screen down to 1280x1024, and dropping to 16 bit, I ran a test with one icon, and got these numbers, shows how much high screen resolutions really effect your gameplay.  I gotta quit playing in 1600x1200. heh

Code: [Select]

With one icon:

722 scans in 10s.  Avg. scan time: 13ms.
# of savepix 1: 7220 in 10s.
# of savepix 2: 2166 in 10s.

With 11 icons on bar:
invis, weaken, clumsy, feeble, curse, corpse, reflect, strength, reac armor, protect, night sight

239 scans in 10s.  Avg. scan time 41ms.
# of savepix 1: 3585
# of savepix 2: 8365

[Endless Night]

VERSION UPDATE  EUOX 19c  oEUO 3.5 beta

Added Icon definition -- Gargoyle Berserk.

[gimlet]

EN - Outstanding - thanks

Twinkle McNugget - So was it both resolution and depth (16 vs 32) or is one more important than the other?
Thanks

[Endless Night]

EN - Outstanding - thanks

Twinkle McNugget - So was it both resolution and depth (16 vs 32) or is one more important than the other?
Thanks

Your welcome Gimlet... I just discovered another Buff Icon that Ive been missing for years... unbelievable... Ethereal Voyage a Spellweaving spell... probably not used very much.   I will be updating again soon because of this.

[Endless Night]

VERSION UPDATE  EUOX 20  oEUO 4.0 beta

Added Icon definitions ..

Spellweaving          - Ethernal Voyage
Bard_Peacemaking   - Resilience, Preservance
Bard_Provocation    - Inspire, Invigorate
Bard_Discordance    - Tribulation, Despair

I believe that all buff/debuff icons are now detected. 50 in total.
 
Tomorrow ill work on adding a significant speed up to the code.