Gem Tower Defense

Comments

Name:

type hello

216 - 2020-07-07, 05:52 PM

216 - 2020-07-06, 04:17 PM

Interesting discovery about ABS:
Flamestrinke chance is 20% instead 10.
What this mean? :
ABS have even slighlty more (long-time avg) dps than EG!

As i wrote before, with GO and BO EG have 1442 dps(in ametyst weakness ofc), 
while ABS have 686 + 500*flamestrikes currently active.

I assumed 1.25 avg flamestrike ative so i trought it have lower dps than EG.
But 20% allows it use ussually all mana, so up to 2 avg flamestrike active, which give over 1600 DPS.

(Actually it will never reach 2 in long-term, because some mana will be wasted if no proc many times in row, but close enough to 2).

In no-GO runs difference between ABS and EG may be even bigger.

So with enough EGs, proritizing ABS over EG on mid probably will worth.
Ofc, still no worth allow ABS to steal kills, so rather start building tham after wawe 100-150.

Also ABS cannot proc flamestrike if target is killed by shoot dmg. Sometimes it deal dmg, proc it, then enemy is instantly killed by 100-dmg tick, so you may notice "kill with flamestrike proc" . Also other procs, like jade gold gain no work if enemy gets killed.
(If you use that aznist's link to code on pastebin(https://pastebin.com/dyvEvUwv), look at lines 4401-4489)

Btw flamestrike radius is 7. Frost nova is 28.6 (but it deal dmg only once, when created).

So speed that would give always 1-tick from flamestrike while passing is about 5.38 - slowed(4.91) units can get 2.

In short-time bursts EG will win to ABS, but we shouldn't underestimate ABS, especially in no-GO runs.

216 - 2020-07-03, 05:52 PM

If grounds are leaking, increasing length of maze also increase time(need more time to kill units). But 14k in 73 ? (how?) It was because strong lag or really bad luck ( no EGs, mostly jades?)
Also leaking at this point shouldn't happens, even with quite short maze, but sometimes paraiba may be needed.
Or you were tabbing-off/minimizing ?

My worst times around 70 are 7-8k.

if i had so bad run i wouldn't regret of losing it.

My chrome don't resetting flash, it's happening when chrome auto-updates?.

----
I noticed that when you cliking "combine special" on gem that have kills , special gem will have the same amount, only kills from other part are wasted. IMO specials should take sum of parts kills, but i trough starting to 0 it's intentional, so in this case it should reset kills on last part too, so it's another bug.

---
On my current run i accidently clicked (aroung wawe 30) "keep" then downgrade on wrong gem... resulting having chipped ametyst.. and it even stole a kill! (wawe 58 now, time 3684, bad EG gain, but it's not going to be 14k time in 70 :)

lement - 2020-07-03, 04:30 PM

216 (more math) - 2020-07-03, 10:36 AM

After having enough of reseting in early wawes, i decided to calculate chances of this.
100% chipped calculations:

total possibilities : 8^5 = 32768

( I will use Q=aqua, instead Aq as usually)

Possibilites of getting 5 different and no silver/malachite:
A R + anything except T S D or E Q O - 6*5*4/6 - 2 = 18 cases
A + 2 parts of silver and 2 of malachite - 3*3=9 cases
R + the same ( another 9 cases)

All of that cases have 120 permutations, so 36*120 = 4320 (13%)
Maybe not too much, but uselessness of fR make more problems... :

R R + others different , no special - 7*6*5/6 - 2 = 33 cases, but only 60 perm. each,so 1980

R R R + anything not the same - 7*6/2 * 20 perm. = 420

So, in total 6720/32768 = 20.5% chance to "forced" reset if 5 chipped gems!

btw. chance of getting silver(with 100% chipped):
T S D + differents - 5*4/2 * 120 perm. = 1200
T S D + pair       - 5*60 perm.        = 300
T S D D + something- 5*60              = 300
same for T S S D                         300
and T T S D                              300
T S S D D          - 1*30 perm.          30
T T S D D                                30
T T S S D                                30
T S D D D          - 1*20 perm.          20
T S S S D                                20
T T T S D                                20
Total 2550 (7,78%) which is a bit lower than i excepted.( the same is for malachite ofc).
This also means that not upgrading chances at wawe 3 and 4 rather no worth - more risk that chance for silver( unless you really want get silver and accept risk of resetting).

70% chipped calculations:

0.7^5=0.16807. This means 16.8 % chance of being the same as 100% chipped, so 3.45 % of reset with 5 uneless chipped gems... but what happens in rest 83.2% cases? - everything would be ok, if not usseless fR, which is practically chipped gem.

What is chance of getting only fR?
we can exclude it from flawed so it means 73.75% chipped chance - ^5 = 21.8%, so-16.8% it's 5%, but still it can be more than 1 fR or 2 the same chipped which save from reset... so need specific calculations for this.

fR + 2 silver parts + 2 malachite parts - 9*120 cases * 0.0375*0.0875^4 chance = 0.247%
fR cA + *not special and no cR* - 18 *120 =2160 cases , its 0.475%
fR cR + * not cA/special* -the same, 2160 cases, 0.475%
fR cR cA + 2 others - 6*5/2 = 15*120 cases, 0.396%
fR cR cR +2 others(it can be cA now) = 7*6/2*60 perm.=1260, 0.277%
fR cR cR cR + other (most frustrating case) - 7*20 cases, 0.031%, but i got that recently anyway.

total for only fR: 8600 cases, 1.89% chance, so better that 5%
Adding to 3.45% its 5.34% to forced reset with 1 chance upgrade.

60% chipped calculations:
I wanted to write it's 0.9^5 * 70% calculation, but it's not extracly true( chipped : flawed change 7:3 -> 6:3).
But previous calc. can be used:

Definitelly it's 0.6^5 * 20.5% = 1.594%.
what about fR? = it will be still 8600 cases, but (specific) chipped chance will be 0.075, 
so 0.0375*0.074^4=1.02%

So 2.614% to forced reset.

50% chipped:
0.641% + 0.492% =
1.133%

40% chipped:

0.21 + 0.201
0.411%

I remeber i got few times 5 useless chipped with that chances.

all this mean for malachite runs when i have 60% chipped from wawe 5 to wawe 12(including) significant chance to reset:
ingrore 50% on wawe 1 to get malachite,
0.795* 0.9464^2 * 0.97386^8 = 57,6% to not reset in first 12 wawes ( in addition to resets due to having gem that stole/it in place that force stealing kills).
Idk if anyone will read this carefully, but maybe someone will be interested.

216 - 2020-06-29, 07:00 AM

I remeber your comment with right corner run, that BS position is not best for wawe 40 times, just too far away from start, road to c1 and back is also very long. 80 kills on BS is not many, it suggest BS was in 20-21. 12.8k time in 95 is strange long for that gems, but with 5 BSs i think you had heavy lags.

My maze was slightly modified for BS (maybe wasn't optimal, but good enough).

I agree that splitting enemies lowering BS dmg. Having too much EE far from BS is making that, but EE which don't stun grounds before c3 are quite safe. 
So stunzone near c1 worth doing later if strong BS.

3 paraibas for downleft start is not too much i think (but actually i didn't tried WGW's strat yet, maybe in future. I have never liked idea of leaking all flyers).

-------------

Yes, without GO and BO enemies reach c6 (sometimes even leak in worse runs).

-------------

With full downline there is no reason for any EE on upper downline, in GO runs (maybe having 1 is acceptable). 
Sometimes i have single EEs here, maybe worth doing it above hole(missing EE), if there is one in downline.

lement - 2020-06-29, 04:22 AM

216 - 2020-06-28, 08:56 PM

Well i got only 20s worse time that old WGW's time, but at least killed flyers insted leaking :). 
There is pissoble to get much better time, but early BS is not enough - need also got GO around wawe 30-70 and decent EG gain around wawe 40 , few EEs also needed, they help bursting enemies.

Ofc for best times worth making 1st BO eariel, best if it's 1-turn BO (notice that they are a bit rarer than GOs).

---
BS runs 15-40:
Ater getting, before 20 it smash everything in seconds.
In early BS runs important is buffing BS by PO around wawe 30 and having EG in place good for armor reduction.
Worth also adding 2 nd BS, and paraiba (near both BSs) in wawe 35-40 ( i did in 37 i think). Unfortunatelly it steal some kills, so rather dont make it before 30 (mine had 80+ at the end).

Slowing gems help much, i build uranium early(didn't even uprade it, sice slow range is the same), near BS - enemies before +-30 even couldn't reach c1

SYS with aoe slow is always helpfull. It's aoe dmg is also good, but ofc not as good as BS, so you want more kills on BS. 5% buff not bad too.

So, you want get almost all specials before 40.

In that run i got PT in wawe 10 - was hoping for early uranium , but didn't got NS( and when got PR i didn't think about it anymore :), i did uranium in 22 i think).
Enemies after 30 had problems with reaching c1, in 39 maybe 2 reached c2, 
c3 was reached frist time(by ground unit) in 47.

Shape of maze ofc matters, need short way from start - c1 - c2. 
In my maze frost road BS - c1 - BS - c2 - BS is as short as possible so it take less time while enemies only walking.

---
41+:
Wawe 41 is not a moment when BS stops being usefull, *0.8 dmg in not as bad, speed may be bigger problem - EEs help here  making enemies are closer to each other.

But enemies HP increases quick so after 50 wawes are much slowers, after 60 BS definitelly need more help .

BS always do much dmg. but ememies are just strong - it make a felling that it is doing nothing. later groung units dies mostly while bursting by many EGs+EEs - so for good time is needed to build fast enough amount of them.

Also single target of BS is bad,so for last units EGs/EEs are better

BS can do 50-100k dmg (if no more!) to every enemy, but wawes 100+ have over 300k hp...( (wawe-30)*4500 to be specyfic). So in this part of game good to have many EG with many kills. However BS will rather do more total dmg than any single EG.

In theory making 10+ BS would be good, but bacuse lag it may result worse time.
So what we can do to have better times? - need let your EG getting more kills than EEs. 
Becuse EG have 20% dmg than EE kills are 20% more effective.
However in late game low-dmg stun zones may be bigger problem.

Making 2nd and maybe even 3rd paraiba probably will needed for time records.

@lement: in which wawe you got BS? and did you have PO and EG early enough?

-----------
tabbing off in ground wawes may allow playing more that 1 game at once. But better nott doing that anyway.
It reduce CPU usage from +-40% to almost 0, so it's strange.

-----------
By upper downline you mean vertical line, which starts between c1 and c5?

lement - 2020-06-28, 05:44 PM

216 - 2020-06-27, 08:16 PM

Yesterday started next run and got PR in 15, BS constructed in 17. Times:
at 40: slightly over 1300
60: 2,9k
96: 8k
at this point wawes was slow, bacuse no GO and BO(but got GO at 144...).
Now wawe 167 that main BS have 217 kills, similar to malachite. time is not very good, too late GO, also lag was slowing i think.

I was tabbing away few times to reduce CPU usage - noticed that one groung wawe(before BO) was barley reach c4, while previous (not tabbed away) get to c6.

leaked 1 at 168, 16 EE, 1 outside of GO range (also in 152 was 1 leak)

-----------------

few EEs above downline , ecpecially below c3 works well, for last stunzone.
Ofc buiding full 2nd line don't make sense.

-----------------

SYS with 100 kills happened to me before, now have 2 SYSs with 43 and 34 kills.
In previous run i made it in 23(?), and it was in place often used for main EG.

SYS often get 80+ kills and it's good dmg gem before 40 - notice it blue dmg have 1* on everything, while EG have 0.8. And good splash.

lement - 2020-06-27, 07:47 AM

216 - 2020-06-26, 01:46 PM

I wait untill name stop disappearing before sumbitting.
---------------
Yes , i prioritize EE/EG over jades, but rather have at least 5-7 in 96. Then after BO worth adding more.
In case of bad EE/EG gain, ofc i take more jades early.

Jades give most gold in late game, higher lvl + longer wawes. But most of them should be done before 130. Last one maybe around 150.
-------------

GO runs with good EE luck can have less jades ofc, but you never know how many EEs you will get later.
Also on GO runs you want get succes chance as close to 100% as possible, while in no-GO something about 50% looks good.
90% in GO run? - relatively dangerouns.
--------------

Where to locate rest EE in GO runs - in last quadrant. Last stun zone must be very stable, while 4EE even with GO don't guarantee perm stun, i recommend use 6.
Possible even to make 2 stunzones after c6, 19EE solutions had more EE in Q4(i forgot how you called it, DRQ, LRQ, BRQ?) than i excepted.

Ofc using a bit more thn i said in ULQ in not a mistake.

--------------------

I don't understand how minimizing affect game, bacuse it can't just lower CPU usage few time, so it may be bugged or don't calculate everything correctly - that's why i said it might be cheating. 
Ofc it can have bad effect leaks too, need some testing, but it's hard to test.

while comparing our times - since CPU speed affect times, i fells like cheater because of having faster CPU :)
Groung wawes lag hard anyway.

-----------------------

My current run in wawes 100-130 i got 0 PE, wchich cause some wawes with 2, once 3(in 136) leaks. Wawe 140 leaked 0 without adding any EE or something important( 12 EE was in both wawes).

No main EG this run, but a lot of lvl 4-8 EGs. One EE have 120+ kills, anothe one 86(wawe 175 now), which is more than any EG. Even SYS have 100 kills.

However that cause not geting as good time as i wanted, beating WGW time is possible, but it will be very close. below 20k 196 time possible if better run.

After 140 got enough EE, and stopped leaks.

Once you said that EEs gets less kills that EGs. True but dmg distribution(2 dices) don't affect it, just EG hav 20 more dmg (in ametyst) and more range.(idk if i said that before).

Btw, malachite 394 (wawe 177).

lement - 2020-06-26, 11:34 AM

Yeah, I've noticed that disappearing also but I just retype the name if it disappears before submitting.

- - - - - - - - - -

Yeah the 104 death didn't have a BO. Upper BO first allows earlier placement, but all needed places were still not covered.

Also while odds of stack is higher the lag is much lower so autoclicker is more powerful there.

Focusing more killers than necessary for land does help reduce odds of stack, but prioritizing it over jades means less gold gathered long-term (though in these waves locations where jade and EG both work are good). Since in 196 it doesn't matter much when I placed the flyer-focus EE but it does matter how much gold I have.....Though this is bit of a laser-focus on that wave in particular, when could die earlier as you say - and as I did.

- - - - - - - - - -

I think GO runs need less jades because they need less EE to beat 196, meaning they don't need so spectacular PE luck and thus so many repicks. (but ofc more gold = higher GO chance)

Generally I find one never has enough gold to build only EE so more jades = more EE even with the time taken I think.

Reassigning EE for GO: ten in ULQ, including the one directly on the path. Reducing that to six.... Well, the horizontally-doubled one is natural candidate for downsizing*, but removal of rest means weakening around c2 stun&leaving gaps. Given the effects of bullet delay, I guess gap every 2 like _mmkay demonstrated would work, which would mean removing first horizontal pair, the EE above seccond horizontal pair and perhaps either 1 from pair itself or the one on the line - not sure which but probs the latter.

Question then is though where would they be placed; 1 could be on downline above c6 but others don't seem to have better placement options within range of POs (if could range lower right then moving 1 EE there diagonally downright would be natural so no extra spot).

...Ah, hm, I guess barely ranging c2-c3 and c5-c6 without sharing entrance point with the second pair in ULQ would provide spots for two more EEs.

* Though issue is, it won't be reached by BO in mid which means it needs to be built before first BO.

PS: Good to know about leaving the gap being a relative mistake. I'd have wanted that last EE to be higher up tbh, with that point it'll hit the c5-c6 when they're just-in-range of the two above c6 i.e. doubling up. But more c5-c6 stun just before mid seems like good idea.

- - - - - - - - - -

Min speed reduction gem....With proper coverage of jades it'd act as speed reduction gem, but that's somewhat limited. Would be worth to have 1-3 I guess.

- - - - - - - - - -

Tabbing away/minimizing game, as far as I can tell, has the effect of making it use much less CPU i.e. not maxing out a core (though still using like third of it so it'd eventually still hit the point where it'd matter and if I had one actively running it'd also affect lag), which I think reduces lag to what it is in first 20 waves or so. If this is cheating, then so is using better CPU :P

(Though using autoclicker is way worse cheat in any case.)

Iirc it also results in different time(more) taken, but that might be side effect of first where barely-ranging EE actually can stunlock for a bit i.e. when land heads to c4 right over the c1&c5 stunners.

But ofc I want to see it, for the reasons you stated. And also because these recordings turn out to be pretty darn useful to rewatch to notice things I miss on pass.

However, that's twice now where chances were reduced from "can pass if rolling well" to "prepare for snake-eyes", and generally while there's no difference in early waves having lag turn even "late" 37-39 waves from 0 leaks to death is pretty significant indicator. (Dunno how close match your lag simulation was but it was definitely.)

- - - - - - -

I do wonder about your BO on wave 88 - hm, right, you place both centrally and don't bother boosting damage of all EE. Still, again, absurdly fast time score compared to what I get :P

216 - 2020-06-26, 09:24 AM

Not only times look legit, he was also active in comments and gave links to pics(but links from 2008 not working now)
--------------------------------------
I meant that need to wait before submiting, when i type name instanly it dissapear after few seconds.
Only once i pressed sumbit button and it disapear - in my 514 run. I trough it was because over 2.14b score, so server didn't accept it. 
The score was displayed normally in game, but because lack of over-2.14b fake scores ever i think it's impossible to sumbit that?

Mid-run disconnection shouldn't change anything, but connection problems while submiting ofc yes.

--------------------------------------
Other problem of excess of parts its just lower amount of EE/EG which means more leaks - so more risk in 104 for example. If more EE was constructed you probably could survive.

If with good enough PA gain at start, sometimes worth ingore them aroung 100 if you have PEs - 108-140 is very dangerous moment in GO runs. 
I also afraid stack in 96-104 if it's before 1st BO. (in 104 GO run you was before 1st BO?)

"100 kills can be raplaced be 1 EG" - in theory yes, in practice it means 3 "wasted" turns to build EG that only recompensate dmg that you should have without any "cost".
Ofc, 100 more wasted kills is not main reason of fails, but i hate that.

--------------------------------------

Don't agree that GO runs need less jades - with bad EE gain need to build more, i raised limit to 14. 
Actually i did 15 in my last 208 run and got 16th(1-turn) in 198, but i felt it was too much - i got more gold than needed(but good EE gain in late game, so hard repick wasn't needed).

No-GO run need some EE luck, so rather no time for build so much jades. also need build more EG probalby, you had 16 of them and 7 PAs.
So 10 jades in no-GO run may be enough, old limit (12) will works well i think (ofc if i got 1-turns after it's ok to have more).

that extra places for EEs may be reason for making 3rd buffing PO in no-GO runs.

Biggest difference betweeen GO and no-GO runs in in EE locations - need much less for c1-c2 stunning, you used 9 EE in ULQ - with GO 5-6 would be enough here. Also downline work much more efficient - you said your down line was not enoughn ot c2-c3( maybe i disagree, full downline works well for c2-c3 stunning even in no-GO rus, but you had only 3 gems on it, and two 0.5 above).

However adding EE below GS as you said don't give as good results as i excepted(but better than your last EE)

------------------------------------

If uranium would reduce below min speed it would be very OP and i would making 5-10 of them( depens how much below).

Would be interesting if some gems have min speed reduction aura( maybe uselees for now (pink)diamonds).

------------------------------------

Minimizing game:
Wait, you want minimize game while attempting 196?
i'm not sure that game work property when minimized, so if it would reduce leaks, that may be consider as cheating. Also you want to see it, especially if it would be "historic" moment( defeating 196 without GO first time definitelly will be that, hard to imagine better achievement, only 196+ without buying lives can be it).

But you had much lees lag in your previous run, what could cause this? - idk if doing other games while main game is open can affect later lags.

----------------------------------
*starting wawe 86*
And... got GO. i really start getting bored of GO runs, too easy :) 
But i hope for good time, not best for now, but acceptable, 7600 in 86, but i will make 2nd paraiba to speed up.
(ofc still there is chance for fail, but need get very unlucky stack).
its my 27th GO run(?) as i remember 15 succeses(1 not sumbited), 11 fails. And 1 closed(not counted).

*Wawe 88* - leaked 3, 1 +2 perfectly stacked, i think i will make a BO before 96.

lement - 2020-06-26, 01:03 AM

Yeah those ESxiviliaeMoath times look legitish (also jeez 2007 hardware beating mine ^^), albeit dying to land 203 is bit odd place though. Good find...And I wasn't aware the game was complete over dozen years ago. Good that I wasn't :) :P

- - - - - - - -

I've chosen to not pick PA/go for fT when the PA is taking an important EE location, but that's rare - I generally have more EE locations than PEs. Also why I'm not too upset at flood of PEs, unless I have little to no EG/EE and they start racking 10-20 kills each. And even then....They're PEs! Losing even 100 kills off EG can be compensated by just building an extra one.

Yeah the 196 score didn't go through. I waited a bit to compose said short reply first, but then attempted to submit it - seems it didn't go through, but the submit score button disappeared. Plus as the 95  112k time run shows waiting long doesn't stop it from going through....And I don't think net disconnection mid-run does either, given that happens basically to every run I do.

- - - - - - - -

Damage gems wise, GR and GAq were both "misplaced" in that they couldn't range c2c3 so good but could be better. The ABS was 1-turn special, I def got 1-turn jades and maybe something else as well.

The PA right under SYS was late pick but not ideally placed - but would likely have run out of gold given only 9 repicks remaining at the end.

Can't exactly choose when you get PAs, but ideally they'd be in DLQ to range c1-c2, c2-c3, c6-exit for quick killing for second pass; the upper GA looks better placed than the lower one but in recording ends up hitting the bats stunned in c3-c4-c5.

PAs do more damage than EG per pick overall but if those bats on c2-mid could have died sooner five EEs wouldn't have suddenly refocused on same bat at once so limiting range can be worth it (but then have to get them into PO range).*

- - - - - - - -

GO vs no-GO, my strat has had no reason to pre-emptively differ them so far, because I very much hope for GO to come out and figure attemps without are something of a hail mary....I guess non-GO runs would want to build more jades for extra gold for extra GO/PE chances.

After GO comes out typically some additional spots open up for EE placement i.e. can place them directly on c3-c4 and have them be buffed. If I haven't built EEs to left of those yet I take said chance, but I generally try to get the topmost of them at least built because of wanting to build upper BO.

* On that note, speaking of PO range: Might be worth leaving the 2nd column in DLQ bare in the future even if it is good place for EGs, as that would allow PO to be placed 1 to the right from where it was, which would allow building another 2 gems that range c3 and c6-exit. Not as useful as it could be, but both of those places could use extra EE coverage. The place under GS in downline would work;  you'll notice I semi-deliberately left a gap in downline there given my last pick; I figured the EE coverage would be too much in mid-c6 section with that.

Heck, I wrote that before I noticed the 0.25 lag dropping (though ofc more interesting would be not 27 EE run but the last EE moved to there run. Still would be better I think), so that's even stronger argument for the switch.

Also, even had the fourth PO been half or full rock lower, it'd have ranged under bottomright EG &  the 2nd PO there formerly could range it as well, so I still think not worth making five POs - but maybe building the fourth PO sooner is necessary; I waited rather long for GO to start doing that.

216 - 2020-06-25, 07:42 PM

We always trought that aznist was first who beat 196, but looking at comments/scores from jan 2008/dec 2007, i can see two 196+ scores that seems legit (by ESxiviliaeMoath).
Unfortunatelly links from 2008 aren't working, so i can't see mazes. 
Also i can't read all commets, because was like 10-50 per day so not all was archieved.
--------------------------------------------

@lement: about that 104 GO run: yes, "emerald floods" are sometimes problem, but i noticed you often have many EE/EG parts on map.
In case of having many PEs (and at lest 1 FS), fT may have higher prioryty than PA. If enough gold, even sometimes worth repick PA for fTs.
Remeber, only gem that you want always take is PE(only if you have (1st) GO in the same picks you ingore PE, or near 196 if you want to build last EEs and have enough PEs).

Having to much EG/EE (and other) parts on map cause some problems.

That 196 score wasn't submited, did you wait enough time before clicking submit?

Also did simulation with "medium lag" (*1.8) - 0.84 leaks. adding 1 EE in last stunzone would reduce it by almot 0.25.
(but needed extra PO for that,  your PO can't range more i think 3 PO are needed for buff in no-GO runs, so 5 PO total).

-------------------------------------------------------
*checking for new comments before commenting*
(just reaziled your new, long comment)

after quick-read:
Yes that succes chance was really good, close to my alg(60+%, 0.4 avg leaks with 28EE).
more dmg gems,like GA and other greats(maybe 1 -turn specials?) in compare to gems used in alg prabably help.
It's possible to get better result with more dmg gems.

Even with lag you had significant chance of doing thing wchich everybody trought it's impossible for last 13(?) years. (actually i didn't found anything about GO in ESxiviliaeMoath's comments, but i think he had it. only found that he said about overwritting opal bug).

Also rather don't compare no-GO runs to GO runs, strat may be more different than you think.

(coming back to reading your comment more carefully).

lement - 2020-06-25, 06:49 PM

...That's 53,5% success chance for first no-GO 196+ run? Didn't you get like 60% with 2 extra EE and genetic algoritm? Jeez, next time I'll minimize the game to maximize my odds (Yeah I had 1 ongoing but in-between waves) >> In any case, much thanks for running the sim; that's very much "not hopeless" odds there.

With bulletspeed, not sure if there's more bullet lag with bigger range, but it is pretty obvious Great Sapphire bullets move more pixels/second than EE ones. On that note, it's kinda obvious but I wanted an EE where that rightmost great sapphire was.

Btw forgot to mention it, but you're correct about the rightmost EE. I don't think it's about what gem you use - thing is, it was affected before I built the second BO, but lost the boost after that! *shakes fist*

And yeah would have wanted last EE below the bottom EG; but for that to get buffed the bottom PO would have needed to be half a rock lower (it'd also have placed 1 PA into range, which would have helped slightly).

Other locations...it looks like the EE left of URQ SYS was perhaps flawed pick (a benefit of recording and rewatching); it let the leading two flyers pass due focusing on stunned one from the back of bat queue.

But really, I needed a place to stun extra enemy: more stun power just above mid would have been useful.

- - - - - -

I definitely don't repick that aggressively (honestly, having pile of gold sitting around is bit of a problem). When doing a PO run, I think one should save the repicks for well-placing EEs so spending on them on non-fatal impatience seems like a waste.....Which is why I sometimes go up to 8 fS just sitting around - I don't aggressively focus on it unless I suspect I need gold to not die soon.

I'm not sure jade would pay for itself by 40, but def would by 44.

- - - - - -

"Silver between c1 and c5 can't steal early kills in my maze." Well now that's quite interesting. With most of the mazes I've been doing, it tends to steal some, even unupgraded. That's a fair argument for prioritizing shorter path around malachite in c0-c1 than I've been doing. Maybe just few short games to practice that part.

With the style of your mazes, the possibility of opal that can't hit dwarves but can hit bats is a possibility, I think (though maybe with not quite that particular design you showed).

- - - - - -

Tbh if Uranium would be able to reduce speed below minimum it'd be fairly strong force multiplier and definitely worth planning around for emerald runs. I've once managed to get silver, jade and Uranium before minimum speed being set; the dwarves were almost standing still in the short time they lived.

- - - - - - -

Jade locations: Yeah some barely hit enemies.

There are multiple apparent weaknesses with slow in said 196 fail maze:

1: almost no 4,5 ULQ jades: 1 jade focused on a stunned flyers and rest flew by with just sapphire slow. Contrariwise, the c5-c6 was acceptably poisoned. Though the stunned enemy between mid and c2 resulted in the two jades next to c2 failing to spread the poison, so those can be ditched from anti-flyer POV.

2: The jade in URQ was too low and did the same. Half a rock to left and higher (barely hitting) or full rock higher (c2-c3 out of range) would be better for c4 slow.

3. Similar issue with ULQ SYS; it focused on stunned enemy. Lots of splash damage there, but  with the above c6-exit jadeline poison wearing off bit more than halfway in c3-c4, it might be more effective to raise it above hitting c2-c3 and then lowering the second SYS somewhere around where _mmkay used; perhaps on the ULQ 1,5 line. Though this would definitely need jades on the 4,5 line or the flyers will clump up behind the first one.

Alternatively, might be worth going back to putting splash slower above c6-exit line and keeping the c1-c5. Not full minimum speed across the whole maze this way either, but fastest enemies would wear off just as they enter in the slow range of latter.

The bigger cost might be that silver doesn't hit c2-c3 enemies there unless raised by 0,5 and maxed out, weaking its use against wave 12.

Also generally perhaps trading some PAs for more gold-generating jades might be prudent for me to get extra PEs - though by the time there's 10 of them the remaining gold is bit limited from first, what's with most waves having gone.

- - - - - - -

Finding and relooking at your 192 run... Well, you had 24 EE there (25 after 1 extra), I had 26.

But if you wanted to ask me what were the apparent differences....Well the URQ seems significant to me. Beyond the lower number, EE just right of c5 seems to have gotten somewhat baffling since I last looked at it.

- Consider and compare _mmkay's 21 EE run: the URQ had about as much focus on stun, but the two topmost EE in URQ ranged c4 rather than c5. Personally, I've placed those EEs to range both in recent mazes, but if I were to pick to range only one, I'd have picked c4 rather than c5 when c5 is covered by the EEs that range c1 and c5.

Others are more minor:

In my experience EE downline just 2 above c6-exit is insufficient to stun much flyers there, which is partly why I had the EEs left and right of SYS ranging c2-c3, but that means taking away EEs on c6-exit so eh...

Also looks like the jade on 3-line between EE and EG could be better c4 slower 1 to left....Though then again you don't have the 1,5 ULQ poison column, so maybe not so much.

If bottomleft EE was moved to the empty space between it and jade, it seems like it'd make the group of 3 range c5-c6 at roughly same time....As the EE just below 2 blood stones, so that might be too concentrated stun I guess.

- - - - - - -

With WGW dominant, there was, quite simply, much less emerald attempts - certainly, I wouldn't have handled URQ EE the way I handle it now back then, but with no legit 196+ scores obviously the downleft strat was the only option. My few emerald attempts were "yup, it dies to stack" and that was about it.

216 - 2020-06-25, 10:35 AM

Wow! Really good run! But lag.. i can confirm it affect leaks, i changed bullet speed in my sim to simulate lagging. But that last delayed stun wasn't thing that killed you, 3-4 EEs are not enough ussually enough to stop/ hold last flyer even without lag.
Comparing both simulations, lag affect stunning around c5 the most, without lag more flyes gets stun here.

You played few games at once, did you fishish others before 196 attempt? lag was much bigger than in previous gif.

I use nomally range*2.5 in my sim for bullet speed, which is even a bit slower that EE should have if i understand code well(both locals are 2000):

rangeModifier = (700 / _local2);
projectileSpeed = (2000 / _local3);

so it should be even slightly more that *2.5 (speed of other bullets matter much less).

I changed it for range*1.1 to simulate huge lag. Results:

-no lag:

0:10210
1:7227
2:1563
3+:81
avg leaks: 0.555

with "lag" avg leaks was 1.14 and succes chance around 20%...

maybe that 1.1 was too slow and you didn't had lag at start of the wawe, but definitelly your slower cpu give unfair disadvantage to you.(idk even why lag slows bullets, fps can be lower, but it should not affect (relative) speed of things, so it's really serious bug).

------------------------ 
POs: 
looks like the righmost EEs is unaffected by PO. i saw it shoot slower on git, also tested it in sandbox - fT was affected, DE and EE not.
notice that it's extractly 26*5.5=143 distance from PO. However different center of towers make that some towers are buffed here, some not.

in my sim it also was unbuffed... bacuse minor bug( i think i used "less" instead "less or equal" range).
so was no problem with correct simulation at leat :)

however buff would help, buffing that 1 EE would reduce leaks from 0.555 to 0.52
------------------------

My 38 run: 
i actually wanted to make a bit longer maze, but because of having no 1st EG(i had parts), i was afraid of getting it in bad place.
I'm sure that uprading SYS would be enough for wawe 38(but not sure about 39), before 40 it's quite good dmg gem in addition to 5% dmg aura and aoe slow. Blue dmg have *1 multi on everything, which give some advantage .
I leaked 3 in 38, last had only 1000 hp, had enogh gold for 2.

Jades don't give much gold aroung 40. Maybe i repick too agressive, but who want keeping only jade parts and spend gold for lives.. one BS and wawes 38 , 39 are easy.

------------------------

rus with silver in wawes 2-4:
i start placing gems that can steal kills around 10, need something to kill wawe 12(f/N/F ametyst help and slowers). malachite ussually get all 90 in first 9 then 130 after 19. Silver between c1 and c5 can't steal early kills in my maze.

------------------------

Uranium:
yes it was definitelly for prevent stack from multiple uranims, but it resulted to prevent sometimes from itself...
In adition idk why uranium dmg can't stack, it's not too strong gem.

-----------------------

Your run give better results than my 192 run with extra EE that i planned to do, 
but anyway i think you could have beter jade locations.
Total amount of recips suggest run with 8 jades, while you did 10, i think too much was below downline.
Also slow around c4 could be better.

You said you leaked +-2 before, but it was without 2nd BO, bacuse mid BO is quite important.

I'm impressed that you managed to get 196 without GO twice!

lets assume this run had 53% succes chance(rounded down), bacause lag was unfair.

No-GO 196 defeating general ranking:

1. lement 0.57
2. _mmkay 0.01
3. WGW    0.00 (300*0 is still 0)
3. (maybe others with WGW start)
...
x. 216    DNF (not reached 196 without GO)

Anyone else reach lvl 196 without having a GO?
(maybe few people with WGW strat, which give 0.00 chance).
Aznist had few 196 runs but i didn't found any without GO.
WGW had few quite good no-GO emerald runs in 2015(i was suprised when found that comments, but i cannot see all pics), but he didnt get to 196.

Aslo i found comment from few people that was close to beat 196, some runs was really interesting , but still all with GO.
(Strange that nobody except aznist succesed before 2020, now people(that means mmkay and lement) have chance do do it without GO)

lement - 2020-06-25, 04:18 AM

lement - 2020-06-25, 03:10 AM

Shame. Maybe ought to replan for a temporarily even longer maze so that doesn't happen? It looked like your initial plans were straight lines, but zigzagging diagonals are almost always going to beat straight lines in how long time a dwarf spends in range of Malachite.

Not sure that jade was in itself a mistake, though - without enough gold you might very well have lost to flyers in 40/44, plus upgrading SYS itself is only worth like half an EG over that, assuming good placement.

Those are some amazing times, 2200 in 52....The number of times I've gotten under 2200 in 40(!) can be counted on one hand probably.

If you do get silver, at what point do you switch to placing gems where they can kill land?

---------------

And yeah, it is still going. At 192 now (https://i.imgur.com/3Wpk7Sr.png), leaking avg 2 so...Probably gonna fail, but I'll hope for PO only clear.

Got GO on one of the other two, then lost it to wave 104 flyers though. Shame, had tons of PE stored >_>

I think the prevention effect is set so one would be unable to stack uranium damage. Still, good find.

216 - 2020-06-24, 02:43 PM

I got GT in wawe 20, 1-turn EE in 21, PE in 22... then almost nothing usefull and die in 38.
@lement : that no-GO run (from 137 pic) is still going ?
Also about opals: i don't pay much attetion where i take 1-2 first POs, bacause they will for BO anyway.

Today i also lost run by technical problem but it was no-GO around 80-90.

2 days ago i got below 3600 time (3504,i htink) after 60 wawes, new record. 
Even no things like GO,BO,strong BS(one only lvl 3 or 4, another lvl 0 or 1 ) or many EGs, so it's possible to get much better time. Malachite over 300 kils.

yesterday i had chance to get even better time, because 1-turn BO in 33, but died in 52 (2200 time). 0 EEs, but got 1st PE in 52, leaked a bit more than excepted (6?) and i dint have enough gold for that. time after 40 was about 1500.

Malachite runs give good times, but much less chance/impossible to get early specials. Normal runs have 30% flawless at wawes 17,18,19 while malachite runs have 10% at 18 and 20% at 19... i wait for getting again BS/EG at wawe 15, which is almost impossible in malachite runs, so i have mixed fellings about it.
Now i ussually follow rule: if i get silver(in first few wawes) i go normal run, else malachite run.

Notice how bad that 38 run would be if it was malachie run: chances at 23 - no GT, no 
2 EEs, so no EE, no EG ,no BS ..., probably i would die even faster ,in 37 i think( not in 35 because i got yellow shapphire, but i forgot upgrade it... upgraded jade instead in 38 , without that mistake i would survive i think.)
btw GT had 51 kills.
--------------

I said eariel that uranium dmg is sometimes strange irregular, now i think i know why:

It apply dmg every 0.25s but also apply effect on enemy that prevent taking dmg from uranium for next... 0.25s. "what could go wrong".
So sometimes uranium is doing half dmg.

216 - 2020-06-21, 07:58 AM

lement - 2020-06-20, 09:37 PM

You're correct about the c6 jade, but  figured I need gold now more than I need emerald there jade. Don't agree about the c2 jade though near malachite so much though - at that point, it can be ranged by at least 3 stunners, which is more than enough given the longer time spent near path there.

And yeah, you can see the plan for BS next to Malachite.

For kills on EGs, I figure it and Malachite are about equal; assuming roughly 3,5 avg in range for bloodstone in center of map, Malachite would hit 1,14 times more there, and would hit 42/36 times that in ideal location, for total of 4,655 ≃ 1 when multiplied by average malachite shot/average EG shot - so either is fine.

Though sadly the GO run ended in stack (yeah I did make BO, about same time in both, about 4 leaks around then) - the other one is doing better, got past 136 with just lucky 1 miss with this maze(minus the EE near c3 in URQ): https://i.imgur.com/vuZQkLG.png - Though could use 1 more EE on c3-c4, filling in downline and on 2 rocks next to Malachite. I think the numerous PAs help; after all in your earlier 39s EG+4EE example the clear time would be 27s for EG+12 PA. But they're kind of badly placed, so I'm working through FA to remove the ones on downline and such.

Albeit also trying bit different maze now; one that seems to get lower Malachite kills after two attempts - though that's partly due "well-positioned" silver (if I get enough fire support I wait with Malachite).

(On the other hand, upper downline and better positioned EG....Well I'll see how it goes; the Perfect Opals avoid going where I want them, but one has Tourmaline getting near no kills in center of maze and other hasn't even built one.)

[ https://i.imgur.com/QlERyMd.png https://i.imgur.com/azfH9dT.png ]

- - -

Your new run has even stronger Malachite, gj on that. Having "crappy" special luck seems to help with building a more powerful one (though more of a silver lining; what's with the second above shot having avg 78 kills per EG).

And interesting, I thought from explanation gem would have multiple dice but guess not. I guess this maybe helps to explain why EG stacks kills much better than EE as it'd hit the HP values where EG gets kill no matter which fires first more often.

216 - 2020-06-20, 04:35 PM

216 - 2020-06-16, 09:33 PM

Well, stun zones in c3-c4 and c4-c5 should be done after 10-15 EEs. Also jades in your GO run block some good EE places, near malachite and downline around c6(but anyway bad EE luck, however if you get at least avereage EE luck, you can get enough EEs untill 196).
Leaking 5 with GO is very much and dangerous, BO can help, but even it may be not enough to reduce leaks below 2.

Worth adding BS to help malachite.
ofc EGs too, but if you want to create absurd strong malachite ,they will "steal" kills ( kills on EG are better later, when 1-target dmg is important).
around 35 or 50 you may put paraiba near malachite( but it also stealing some kills, but give significant help with -6 armor).

Also i finally realized how dmg range works, _mmkay didn't explain it well:

n is number OF dices.
So for EE base dmg is 98, then add random numer from 1 to 51, then again ANOTHER random number 1-51, so dmg is more often around 150, rarely 100 or 200.

i trough eariel its linear distribution, but i noticed that dmg is near avg too often.

Paraiba have even 5 dices.

lement - 2020-06-16, 08:43 PM

@_mmkay: Gratz on more 196+ scores! Well done, looks like you have a good strategy ongoing, that's quite low EE amount.

Those are interesting EG placements; in particular you're indeed correct that with topmost Q2 EE you can range c5 without it being covered by topmost Q1 EE; making the otherwise overlapping part of range useful for former if bat doesn't get stunned in the non-overlapping part. Staggering the stun zones between ULQ and URQ like this sounds good.

However, it looks like your topmost group of 2 EE in Q2 doesn't range c5 anyway, making it a bit of moot point. Beyond what 216 said, another possible location for EE would then be to range descending from that place (assuming it slows down target enough so the one in Q2 will not range it). I'm also wanting to try building "upper downline" i.e. ranging c2-c3 weakly and hitting c5-c6 strongly sometime, but I think that'd overlap with your first EE target location in c5-c6.

For C3-C4 EE, I've mostly taken to aiming to place the EE outside of c3-c4 path range, so that downline above c6-exit line will not double up on it, so decision to stack up there is also interesting. Shown worthy, I guess, given somewhat insufficient stun power on downline on its own.

I think you should build central/damage-sensitive EE first i.e. in ULQ and mid where you need them to clear away the first bat before they take on the second.

@216:

Well, that 14 PE run ended in stack. Mixed feelings about it: Yay, finally a run that ends in a stack not technical issues and boo, potential non-GO run over.

Typically I find parts don't steal too much kills once there's stronger gems in range....At start though, like in current Malachite test runs, can kinda dread looking at them >_> "Ugh, another 20% damage gone".

Also, gratz on getting another 196+ score with record low EE!

Speaking of Malachite runs, got two ongoing paused at 95:

https://i.imgur.com/EQzns5C.png No GO, no upper final PO, but it's better at killing flyers than the second one.... (GA has 7 kills, topmost EG 71 kills)

https://i.imgur.com/UGsMfVt.png Was bit of throwaway test (wanted to see how much kills it'd get by 40) hence awful initial accepting places of red crystals or stuff like 2 uranium made by parts (the one near c1 could have been better below c5 I guess), then suddenly GO in 33 (80 kills, leftmost EG has 83).

Both got absolutely awful amount of kills on PEs in maze, though; I tried to get Opal to buff Malachite once it started to be too weak on its own, and instead got PE after PE instead. Some of them helped admittedly by spreading slow, but others were just pure killstealers.

Conclusion: My malachite strat could use some serious refining. Part of the point is extra damage, and it all goes to waste like this. Still, not gonna quit with them gone so far.

(Also, 5 leaks on both in 96)

216 - 2020-06-13, 07:11 PM

Actually 4 EEs even GO buffed don't guarantee perm stun, so often i use 6+ in last c6-exit stunzone.
If you have 4 EE and 1 EG all lvl 0 in first stun zone with buffs(GO,BO,RC) they with kill kill 196 wawe flyer in 248975/(4*1231+1442) = 39s (without any dmg from other towers), 
but they rather will have few lvls and other towers also can help, so killing in time shouldn't be problem. 
Sometimes in case of lack of BO range, etc. worth adding 2 nd EG here.

Rather no worth ading EE before c1, like azninst was doing (especially if GO and RC don't range here), better make extra stun zone between c1 and c5(like lement do, it can stun enemy in c1 too, in addintion to help on c5). Or make stunzone at c6-exit.

Downline ussually stun enemy bettwen mid and c6, so that last stun zone in pic shoudn't work ussually. need longer line and/or 1-2 "emergency EEs"(on c6- exit only), so it can stop last enemy. atfer killing enemies before c6 dowline switch to that enemy, but it need to stunned before.

Separated stunzones in some places are important, but i more spread EEs in Q1 (both strategies work, i think).
In stun zones on c3-c4-c5, if you use 2 EEs on c3-c4 and 2 after c4, they should be separated. Also remember that jades help stunning here (8 jades = 1 EE in "stun power").

C3 stun zone: that EEs are low, didn't target something on c2-c3?
I put them higher to avoid that( that why i said about stun zone between c3 and c4).
In addition, rightmost part od downlile ofter create stunzone on c3(but u didnt make it this run as i can see).

So where you can build later EEs?:
-more on downline
-between c1 and c5
-few on c6-exit

Don't build more in Q2, definitelly you had enough here(maybe too much?) and on lef part of map that don't range c1-c5.
(actually i don't think i need saying it, you are very good player and you can find that by yourself.)

Separate stun zones may also reduce stack chance (?), so they might be beter that i trough.

I think you didn't make any paraiba again (baceuse worse time) ?
Actually with strong malachite strat. no making paraiba don't make much sense, since it's startegy for better times(actually idk if it even worth, strog BS are better).

Im my 208 run, i was leaking 0 since 140, except 168 (1), 180 (2), and 188 (1).
i had 14 EE in 110, then very bad luck.
Also, i misslocated 2nd RC, and made 3 rd at bottom part of map.

Interesing fact: You and aznist had below average EE gain while defeating 196 first time.
I trought people will be succesing with more EE luck.

_mmkay - 2020-06-13, 02:37 PM

_mmkay - 2020-06-13, 02:33 PM

The GAs didn't have many kills because I built them a bit too late. One had like 12 and the other had about 4. I think the GA in the previous.

My EE placement felt a lot better and more resourceful to me this run (https://imgur.com/a/i2jSfKo). The benefits of having **separate** stunzones became really apparent. The circled groups are the most important.

In Quadrant 1, each cluster of EEs creates a separate and consistent stunzone. Once they all target the same bat, the bat cannot escape. Each group does this on both sides. With only those 7 EE (could've been 6 even), it guarantees 4 kills assuming that the bats in C1-C2 get killed in time.

In Q2, I made stunzones specifically around C3 and C4 for the first time and it worked out really well. They didn't get kills there because there wasn't enough damage focused on them, but the EEs could always stun bats there for long enough so that they would definitely be killed by the time they escape. They didn't target the same bat the whole time, they would stun one for a bit and then switch to another. Therefore those 4EEs, with the help of other EEs in Q2, would guarantee another 2 kills. Unfortunately, I don't know if the other EEs in the middle of Q2 were very beneficial because they would just stun bats already stunned by the Q1 clusters. I think they did help with other stray bats flying by. Luckily the two touching EEs in Q2 did not mirror a cluster in Q1 so they could focus on different areas.

Not much happens in G3.

In Q4 the downline obvious creates a consistent stunzone at the bottom, and several inconsistent stunzones above. The downline guarantees 1 kill of any bat that makes it to C6 which has been a lifesaver many times.

Overally the 15 circled EEs guaranteed 7/10 kills which is pretty good. The other bats usually get stunned between C2-C3 thanks to the downline. In this run, 4-5 bats would make it past C4 and 1-2 bats would still be stunned somewhere before. 3 of those bats would get stuck in the stunzones and then I relied on luck for the other free EEs to stun the remaining 2.

I don't know what all this means for future runs. Should I build only the circled stunzones first and then build random idle EEs after, like in Q2? If I had 5 extra EEs, I could put them at the very start before C1 and that would guarantee another bat consistently stunned.

216 - 2020-06-13, 02:00 PM

Two 196+ scores was sumbited today (thanks to _mmkay).
Noticed another bug: looks like timer is running after death (look at time: https://imgur.com/WdBUibm), i died 5- 10s after, very bad stun RNG and 5 flyers grouped got past mid (2nd time i mean).But only 1 (probably) was leaking.

This bug can explain very high times in highscores, lement had one 95 lvl with 112k time, with time score as about 20k (if you click "survival" it displays more scores, so it's visible).

I wanted to get good time score, but instead get lowest EE amount succes(20). i got 2 EE after that and hav 1 PE on map, so i trough run will get stable.
Some interesting fact about this run:
- 3 GR, the same as in recent 192 run.
- 4 1-turn jades, 3 early, in 16(?), around 30 and around 50, then 4th in 198.
- 1 turn BO around 70.
- 2 nd GO(and death) in 208.

I think i made mistake with taking GR, blocking good EE postion in zone around malachite, making 2x3 block full of gems.

First 196+ run since months that was ended "normally", due to leak flyer.
Ussually i end stable( at least i think they are stable) runs after lvl 216.

Strange that nobody ever died in lvl 200 and in 204 poeple lose so ofen, in theory should be more 199 lvl scores that 203 on list, but too few people play this game.

216 - 2020-06-13, 09:19 AM

_mmkay - 2020-06-13, 08:53 AM

216 - 2020-06-11, 07:01 AM

Your current run: looks that PE gain is very high, 55k gold suggest not many repicks used, so 14 PE seems much. But you need to build them faster, parts are stealing kills. with 5 PE on map i don't make 6 th FS, even with smaller amount i take fT(in aceptable position ofc) over FS. 
Similar to EGs, building first +-7 fast is important to allow getting lvls on them.

Lack of EG/lvls on them may increase leaks around 100, but this run have high potential, because high PE gain(at first 100 lvls at least).

In your situation i would spam repicks to finnish that 5 EE and 2 EG, accepting only PE and PA(and rare things like Great gems/ 1-t specials). Then go for BO(maybe after getting 6th EG, but you don't have enough POs, even some gems are out of range, definitelly leftmost jades)

After 20 wawes and chance upg. 50 gold left, so enough for repick and silver upg.

How many kills on GT in perfect run? : around 600-700.

I also did after run which was only concentreatd on GT kills, got 1337 in lvl 201(5k dmg,but i already said it in comment few months ago).

lement - 2020-06-11, 01:39 AM

Ah, good point there; indeed you're right - for the time the last 2 live Vivid Malachite will do half damage (so total damage reduction by sixth). Mainly a concern if that reduction would make you leak, I guess.

Haven't looked at ruby damage range before; I've been mainly concerned with asymmetry in perfect opal and perhaps great opal range (though it is kind of hard to get multiple GOs to test that) - in the level 96 GO run freeze that I posted recently, the FA between c1 and c5 was out of range, despite looking like it is inside it. Tbh should put the bit about PO on the wiki: In sandbox it buffs 5,5 up and 0-0,5 to the left, in regular gameplay it is to the right instead. (Would prefer other way around; might also differ between browser and flash versions.)

That was with both gems placed after PO, btw; haven't tested symmetry breaks with placing gems before PO. Greater range for gems placed before might be useful in few cases like first Malachite or first EG vs final POs.

And yeah, I loathe having to go through repicks for the flawless gems; with perfect required it often ends up filling in best gem slots with chaff first, which is imagine is also part of making you restart - by 44 without PO, it's not just the kills being focused on other gems, but also your slots being filled in - making for another reason to go for opal-buffed early malachite, so one can get the increased normal gem chance at cost of wave 12 kills not going entirely to Malachite itself.

However, I find the extra repick after last upgrade chances pretty rare to use - I usually put said gold on upgrading silver or something instead; otherwise it seems like red crystal/etc or wasted kills on FA/NA and such are necessary.

Hm, on this note probably should find a place for unupgraded jade to range fliers without ranging ground; often with early jade I find they steal tons of kills i.e. current non-GO ( https://i.imgur.com/LlGYOCT.png ) run has jade with 96 kills in 103, making it the highest kill gem after initial silver at 183.

Same principle might be useful in general for the early EG (i.e. temporarily maze around it and disable paths of maze that range near other gems). Not workable if you get EG in like wave 38 where you need te extra damage and don't have spare gold, but something to ponder. Though tbh seems lot of your gems range only fliers anyway.

PS: How much kills did that perfect run GT get in total?

216 - 2020-06-10, 06:49 PM

I'm not 100% agree with it. if you have maze that optimal use malachite range, but all 10 enemies is all the time inside, the first 4 will be attacked and killed. Then next 4. After killing first 8, last 2 with full hp(or almost full, ememies sometimes must leave range when go to checkpoint) will be targeted and probably leak( in wawe 10, unless you seperate all ckeckpoints, forcing 7x walk, but its hard to have it at lvl 10).

Notice that leaving range makes each next enemy got 8 hits from Vidid Malachite (no buffed).

So, if you have maze that have several short fragments inside range(long enough to have 4 enemies in range) and outside between( long enough for 6 enemies) it will spread dmg more regulary; when 1st enemy die, all other will have low hp, reducing requied total maze distance inside range almost 3 times.

No matter how long are fragments outside if they are long enough( for 6).

If fragment inside are longer that for 4 enemies, the first (4) will die faster.

Similar is with GT which was my favorite gem to start is sanbox(but 7 / 3 enemies), my "perfect run"(getting lvl 200 with full 10000 lives) was done also from GT.

gems/opal range: i also noticed some strange behaviour, like the same PO sometimes range specific place, sometimes no in the same game ( i think i removed gem for combination, then place next here and it didn't get buff, but don't remember exrtactly).

Also looks like different gems have different center point (while combining 2xPR to GR, you can see range moving left(or right?)).

0,2 EE less: it's not only downside of uprading chances( maybe even not most important).
With GO + over 26 EE (avg is 27 if 190 r.) it almost doesn't matter. But with bad EE luck it may be more imortant. Adding 1 EE in my 192 run would reduce leaks by  0.2 and add about 10 percent succes chance, with it's significant, but after x0.2 not as much. But in GO runs with around 20 EE it matter even more.

However this can be recompensated by making 3 more repicks. after 190 r need 14k+ gold more for 3 next, so 1 extra jade should be enough. This probably will cost a making 1 EG less in late game (one extra no BO-buffed EG in mid may reduce leaks by 0.05, ofc i not talking about 0.08 to 0.03, rather valutes between 0.5 and 1).

The downside that hurt much is less EARLY EGs. This cannot be recompenated by more repicks. Getting first EG in 20 really help, and differences between 20 and 21 and 22 is slighty bigger that you can except. With malachite strat it's only slighty over 1% to get PA in 20( the same in 21 and 22). 
With uprade chances at 20 you can do 1 repick in 20 and 3-4 in 21 and 22, so high chance for EG.

Also less early other specials.

lement - 2020-06-10, 04:17 PM

It doesn't really matter where the other six are for Vivid Malachite though; until you hit the point where you cannot extend maze inside the range there isn't benefit to extending it outside - and you hit the point of Mighty Malachite there way before you fill in the space. (The extending outside the range would be useful when putting off upgrading for later, and when one barely ranges poison gem).

On that note, decided to find the range for Malachite damage and Normal Opal (and put them on wiki). Interestingly, Normal Opal's range is symmetric in normal play, but in sandbox it juts out half a rock on the left side. Malachite damage range is, of course, unsymmetric.

Few realizations from this:

1) Your malachite location in previous image is perhaps the best possible location. Using corner spot as worth twice as much, it covers 42 half-rocks worth of path - meanwhile, diagonally half-rock upward covers 40, and full rock 38. Malachite near mid covers 35. (Land-wise, tried making few mazes and counting; got between 100 and 110 - best I've got was 122 with 39 half-rocks against flyers.)

2) There's fair few possible locations to place the Normal opal so it won't hit anything before c6, at least: (Outlined with border, with c4-c5 having a hook: https://i.imgur.com/JduiQ78.png)

However, doing this shafts the maze; i.e. best placement would mean roughly halving the time ground units keep in range - very much a "lets kill the flyers without upgrade" compromise.

Also that's one nice run, gratz. Improvements in EE placements of course are also nice, but also that time score in 313 was better than the 195 score I submitted :v To exaggerate, what value is 0,2 less PE if you can do twice as many runs?

216 - 2020-06-10, 02:25 PM

Optimal maze for Vivid Malachite is different that Mighty Malachite. because 4 targets limit. If you build lond maze around it, only first 4 enemies will be attacked. To allow it hit also later enemies, need build extra fragments of maze that temporary keep 6+ enemies out of range like that 5-rock line.

Gems near c1: in my maze enemies come to c1 in lvl9( 1-8 are easy) with 90+ hp, so that gems asolutely couldn't steal any kill in wawe 9, 10 and 11.(unfortunetly in 13 and 14 they stole 2, but that runs ended in 56, i spent too much gold, but with 1EE and 2 EG i wanted reset it even eariel).

Early c4 separation: it as rather impossible to aviod stealing kills in wawe 8 and 10, around c4 only fD could be(for wawe 8, still possible kill lost in 10), so better make it later, after 11 ( that blocking rock above c1 was temporary).

Buffing malachite: maybe it's possible to add NO/fO at wawe 10, somewhere left to malachite. but it wasn't needed, so i didn't risk.

Previous malachite run: main EG catch malachite in kills at wawe 313 ( EG: 372, Malachite : 371) ; i decided to end run after that.
Also was strange low EE gain, especially after 196 (23 EE at 196 and 20(?) wawes after), but run was very stable anyway, no very dangerous situations. Previously with silimal EE amount runs was ending around 200, definitelly EEs was located better.

lement - 2020-06-10, 12:23 AM

@216: Huh, well that's not what I expected (is interesting). My old malachite-focused downleft strat focused on making the path around malachite as long as possible i.e. I'd have placed the five rocks on the other side around a corner.

Also helps to have some rocks right above the top of malachite range so they stay in range for longer (those rocks don't need to connect to anything to do their job).

It can be better to start making the sealing line from URQ part of map, as that allows malachite to range c3-c4 or c4-c5 with some temporary help (or both, as WGW's malachite-building doc shows - he also chose to take chipped opal over flawed topaz, interestingly.)

I also try to avoid placing gems where they range c1 because they sometimes steal kills. That your diamond+topaz didn't steal any is pretty neat.

However good point on unable to break wave 12 without upgrade, yeah that's not gonna happen.

Though those are pretty good non-special gems for just twice upgrading chances, you can place the jades/emeralds/sapphires on 1 above c6-exit line and they'll be unable to hit c3-c4 until you upgrade to lucky asian jade (I place some jades there for spreading poison on fliers) ((though I suppose they would hit the c6)).

However, it is a shame that none of the opals buffed the malachite, and might be worth asking how can they be laid so they can do that without stealing kills - the 100/100 shot I posted didn't need to be upgraded by wave 12.

PS: Oh yeah, btw I also play on laptop. Fourth gen i5 :v

@_mmkay: I also do the upgrade sometimes not just because of kills but also because of situations where I get two silvers in ideal places (thus saving me from making SYS; also where I'm willing to accept chipped if I've gotten 2+ 1-turn specials as that takes it just back to 0 effective 1-turn specials), and don't want to restart yet again (but then again, I'm less tolerant of them than 216; I don't think I've ever reset 40+ game from just unluckines as 216 does).

And aw, that sucks. Technical issues, the #1 run-ender.

216 - 2020-06-09, 11:01 AM

Well, i recomend to uprade chances at 20 in EE runs.

It's about 0.2 EE more as well as more early EGs. But malachite strat may give good time scores and some fun, once i managed to get all 190 of first 190 kills, then got EG anyway(from 2 FA) :). 
i got even GO later, i'm now in 175. if i fail it i will  blame that "0.2 EE less" (19EE now, was only 2 in 70. leaked 0 in 100 with 6 EE only).

@_mmkay : there must be some ways to prevent auto-update. if you have Windows 10, you can set some hours(max 18) that it don't happen (i set 10:00-02:00). you can also stop updates for x days.

We all hate losing runs like that :(
I believe you can be 1st that will manage to beat 196 without GO( i also try but didn't even get to 196 yet and ofc when i can i take GO for 196+ run).
Looking at old comments, many people (aznist too) said it's impossible, i said it's exteremelly unlikely. But now i see it can really happen if some luck.

how much kills that Great gems got?

i also like GA early, this gem have really big problem with getting lvls, only way is get it very early.
recently got GA before wawe 28 and it got 32 kills only (GOless run, failed around 100-120, idk if i even sumbited)

_mmkay - 2020-06-09, 09:47 AM

216 - 2020-06-09, 07:32 AM

https://imgur.com/s0CoPUX
Few things about it:
1. still it's not recomended for defeating 196 first time, making stong EG is better.
2. it should worrk even without gems near c1, however almost always i getting something here. fD can't be here at wawe 3. in wawe 9 first enemy come to c1 with around 90 hp, so gems here rather don't steal kills, but with many diamonds here and many crits it can happen.
3. it rather can't beat wawe 12 without upgrading malachite(unless build jade maybe)
4. after upgrading malachite that lne of 5 rocks below can be removed.
5. it's sometiems hard to keep gems away of maze, on this pic i could even put 6 gems extra in upper part of map before wawe 10, but it was lucky. if i put 1 gem away and it's chipped, i often hope i get 2nd, so putting some next gems in maze(in early lvls).
-----------------------------------------------

I when looking at code noticed 
flamestrike = new Flamestrike();  
line in CreateProjectile function, seems it's here acidentally. it may be responsible for some lag, but i'm not sure.
Also found reasons of many bugs, in example i know why star ruby descryption show 0 range(only display bug, used range modifier instead range)

lement - 2020-06-08, 09:52 PM

216 - 2020-06-08, 10:54 AM

lement - 2020-06-08, 10:23 AM

216 - 2020-06-08, 06:22 AM

lement - 2020-06-08, 05:32 AM

Aaaand it froze after I drew into swap space from using too much RAM: https://i.imgur.com/mvjlTWu.png

png, gif, same thing here.

Ugh T.T

Stuff like this is why I said no way I can get 50% win rate with GO. The air enemies stacking is perhaps the smallest threat.

I was expecting some glitch to kill it, and that indeed happened, especially as it lost control of rock placement (staying on place rock button unless replaced and so).

Oddly chrome seems to also leak memory after prolonged usage, too.

...Ah well.

On to other matters.

- - -

Yeah, the /1,14 was because of range. Blood Stone and Malachite both hit all enemies in range, and you're probably stunning 1 or 2 in the range with majority acting as if there's not stun; therefore the damage they do is directly propotional to how much path they cover. 1,14/linear assumption is even generous, tbh, given area covered increases with range².

Repicks at that ending point were 700 gold. Figured I'd first get jades up, then once that was done focus on emeralds, amethysts and some rubies.

I know other flash tower defence games like onslaught and such also lag a lot; it's not unique to GTD. But given the CPU usage goes mostly away minimized, there's some weird coding here.

216 - 2020-06-07, 09:03 PM

That /1.14 was because range? in case of dps calulations it doesn't make sense(but ofc range is important for multi-targets gems). btw i noticed armor table in code and values  was rounded to 3 digits, so it was 0.943 (i didn't found where it's used, so not 100% sure how it works).
do u have link to code(comment in 2015 by a(znist))?

Initial fTs and fDs? 1-4. fS have highest priority(but must be in safe place to not steal kills in this start). I accept also 2x parabia in that runs, since they are time focused.

GO runs: i mean't normaly finished runs, not bugged, accidentally closed, crashed, etc., i know you had at least 3 of them in last 2 months. For me chrome have never reloaded flash, so idk why it happens to you(i don't upload my chore since few months, maybe that?)

3+1 EG and 2EE in 79... very bad (how much repicks?).
But you can get more later, remember about less concentrating EEs on GO runs, especialy around c1, 3 EEs hitting in 1 target may be enough( except stun zones  around c6, where we need to sure that enemy won't escape, 6 is rather safe enough).

i don't believe that 0.001s click interval may be to slow since game runs at around 30(?) fps.

That lag is very good question, it use way too much cpu, also notice that lag increase with wawe time, at start with 10 enemies it's almost no lag, then last few lag very much.

lement - 2020-06-07, 06:33 PM

...Ah, I see I forgot to calc EG's cd, my bad. With, say, average 4 enemies in range would need over 100 kills to match up (then again not sure what is average between all the slow and stun; could use your algo to record damage done by gems). BS is better yeah (relative to Malachite's 115 the initial 135 goes to 89 after ×0,9434×0,8/1,14 which would give it 55% extra assuming splash on average doubles damage), but not so sure about EG.

How many initial flawed topazes and diamonds do you end up with in the wave 12 malachite runs?

Nah, there was also that 2020-06-02 that got go in 83 and in 117? reloaded flash. I guess this is my sixth GO ever (including the one in sandbox, I think?) but zero captures so far so yeah ^^; Only 2 Perfect emeralds by 79 so far so poor luck there. 3 EGs + 1 PA though.

Ah, corrected the wiki. My bad, though can also edit yourself; it's an incomplete wiki as you say. HP of enemies probably should be somewhere, yeah.

Autoclicker is was entirely 0,00 but increased it slightly. I used to keep 1 ms entirely but then decreased it after losing some waves to stacking with clearly separate enemies. Still wonder why the game tends to lag unminimized.

216 - 2020-06-07, 08:09 AM

Sinle target malachite dps is 22% of EG( dps of gems was in my math comment).
My standard malachite gets around 120 of 190 first kills, but it continue also getting kills later, so it's gets over 200 kills, sometimes over 250 in 196.

yesterday strarted next malachite run, now died in 120 (no GO ofc) malachite 330+ kills.
ofc better is farm kills on BS ( or just EG(s)), but almost no wasted kills early is nice.

it's your first GO run(not bugged, etc.) since 2 months?
Lack of EGs and EEs at start may be a problem - more leaks around 100 = more risk.
i hope you finally beat 196, you already play much better that aznist/ others that i found in old comments(using EEs that range only c1-c2 is bad).

if i had that bad EE luck i would probably ingnore game if no GO.

Taking fD+FA is much better than fT+FS, 1 st EG can be in much more places.
Additionally it allows make EG at leat 1 turn eariel wich is important.

autoclick:
So, you have 0ms interval between clicks? it may cause lags, mmkay needed to increase it to 0.02s

I noticed u add Splash range to wiki, but GR dont have slash(also jade crit chance and r vs a/o dmgtable are incorrect on wiki).
Maybe worth add HP of wawes ,
btw hp icrement of flyers is (neglible) more than i said(5999) its 0.3333x HP that ground would have , so 5999.4

lement - 2020-06-06, 10:10 PM

Yeah, that's substantial boost over your standard...160ish kills? Given that each 0-kill Malachite shot is about 30% of EG, that's more than 1 extra EG damage-wise.

I don't go for wave 12 malachite though when I do use it; can't stand the excess flawed topazes/flawed diamonds. Generally I take 110 gold and maybe 140 gold upgrade depending on maze strength; theoretically if one is confident in defeating 16 could go up to wave 19 and there decide whether to take final upgrade or upgrade malachite after killing two fliers (depending on presence or lack of red crystal), but tbh I find by wave 16 if the situation is that strong I have upgraded 2+ chipped specials once, which means that isn't an option.

Admittedly the early game survivability boost is mostly needed if rest of gems don't have much luck - current run took that, but had no extra silvers, malachites or early enough red crystal. However, that combined with early blood stone and buffing Flawless Opal (got over 30 kills so kinda wanted to reset over that before BS) paved way for fastest wave 40 defeat (2019 time) so far. Roughly same luck as you with EE/EG, 'cept for the GO in 41 (hope this doesn't crash/freeze/unload flash/ & last and least: stack, heh. Currently in wave 54: https://i.imgur.com/4PksTT5.png (maze needlessly longer for more gold) ).

Sometimes if I get really early jade I also max that out for a change|nostalgia, but its kind of jank - even if it helps get past 20th wave without as much hole in pocket as malachite, kills on it aren't just as useful. Better than nothing, I guess.

I'll take both fD and FA if it's best move, but generally I try to avoid taking both; I'd much rather try to complete with perfect something less important like blood stone, uranium or tourmaline, though I'll take it over both fT and FS.

Autoclicking:

Perhaps picture of my Actiona setup would be more informative: https://i.imgur.com/Ykxotbl.png

Delays are 0-0 except for first single life click which is 1ms, and for the minimize - 100 extra ms plus currently testing 20 sec down.

Could be improved though; pause on clicking would be useful (I use desktop switching for that atm but not ideal.). There's in-built pause/stop but that's taken out of focus when clicking :v

216 - 2020-06-06, 12:19 PM

lement - 2020-06-06, 11:48 AM

Ouch, that sucks. Yeah, another reason I like early malachite upgrade is the turns don't take as long - in fact, I've noticed the wave duration and malachite kills track each other pretty well in first 40. 6k time in 84 is really good though; that's more like my best wave 60 times. Delayed stack is also why I turn on 1-life purchase in waves 92 and 88.

Though when going for this strat, I reset if, say, jade parts steal 30 kills and I get no 1-turn special to compensate. While most malachites don't get 100/100 kills in first 10 waves, if I'm at 40 and the Malachite doesn't have at least chance of reaching 250 (i.e. about 100 extra or 1 perfect gem) I consider it a bad trade.

Surprised the EG so many kills in that place; good for air waves but it definitely looks like many gems had better position to get kills

Yeah, in that gif you can see about 1-1,5 rock lag, which isn't too bad, but then again that's without autoclicker - also browser behaviour also matters. I recall WGW (with downleft strat) lost a run in 180s with IE due it not accepting clicks that fast. I'm thinking that minimizing air wave for x seconds then unminimizing and starting the autoclicker with bogeys near exit would help ensure a win.

216 - 2020-06-05, 07:12 PM

I can't believe! i lost GO run in 84 because of TRIPLE stack!! leaking only that 3 (or maybe 4). 4 EE, could be better located as first 4, but i didnt excepted that...

run was experimental, upraded malachite early, i didnt get that much kills as i wanted, but GO in 26 forced me to continue. also later got GT in mid and 6-7 EGs. because good aoe time was good, slightly over 6k at 84. i trought it will be 196+ with new time record...
---------------------------------------------------------

@ lement:
main EG was that left to BSs.

The fact that center + 3 line is not optimal don't mean it dont work at all. i think that (near ofc) optilmal stll have 1-2 EEs here(don't remebmer already , i will upgrage my alorithm, then find even better solutions).

Similar to slow on c1-c2, i'm not 100% sure how its important(but almost 100% that slow on c3-c4-c5 is more important).

LAG: you have significant lag in air wawes too? 196 gif that you uplaoded recently don't show big lag that would affect stuns.

also idk how your autocliker affect lag(mmkay said that 0.01 click interval caused lag to him too,so he changed to 0.02)
-------------------------------------------------------

Thanks for link to archieve coment, now i can see all of them!
found link posted by a(znist) to code of the game.

so easilly found splash range of towers:
BS / SYS : 57
silvers : 36
rubies N-P: 30
chipped/ Fd ruby: 25

so SYS/BS have less range than i excepted (i use 80(of enemy posiotion) in my sim to recompenasate splash bug,i will change to 57 of bullet position now).

also need to find extract bullets speeds, already found, bug its strange coded.( i know lag slow it, must be heavy bug in code)

lement - 2020-06-05, 04:47 PM

lement - 2020-06-05, 04:37 PM

Yeah, after posting that it occurred to me that EE that only range C3-C4 and C5-C6 columns both are not the most useful given that the latter is only useful if there's no bat in former. I concluded "more in URQ" from "make stun zones around c3-c4 and c4-c5" and "on 1 verical line between lines c3-c4 and c5-c6" (comment back on 29th february http://web.archive.org/web/20200418234320/http://www.gemtowerdefense.com/comments/) - as you say now, it's not optimal.

The kills in ULQ above downline aren't ideal for that map, yeah. Having five extra jades actually helps here with killing faster to make them useful for second pass, beyond just slowing down c5-c6 (though one my earlier map did overfocus jades in ULQ in hindsight I do think getting slown on c1-c2 is important - it affects how quickly the bat leaves stun range, after all). (And C7 is exit yeah; standarization.)

For URQ, with leaving out the midline between c3c4 and c5c6 and focusing jades more on c3c4 could use the space to build the place into gold gathering zone - just maze the land around all the jades + 1 EG where you weren't going to plaze any EE anyway. Land waves' use is stacking kills and gold, after all.

(Do you mean the one right to Malachite as main EG? If it was 0,5 to left it'd range c1-c2 (then again that's less damage in mid until that dies).)

Back to ULQ, with lack of poison in c1-c2 uranium slow becomes more useful for you. And that's one powerful uranium in wave 39 (must have been bit of a challenge to build that much kills as well as I guess just "7 passes near uranium and no other gems  hitting" wouldn't be enough).

C6-C7 could be replaced by more EE 3 above that line (which can then hit on two passes which is why I was speaking of it), though if you do this I'm suspecting maybe should avoid having EE hit mid-c3 in ULQ, otherwise that spot is overcovered. Jade on 1 above those or 1 below downline can also hit both c3-c4 and c6-c7; latter also allows one to spread the slow around for c4-c5. Only lasts four seconds, but still useful because if you place jade on 2nd column in ULQ it's going to focus on bat in mid if it can reach it before going after bats on c3-c4.

PS: Yeah, looking back that 2nd Feb run had a bit different maze than now, 2 instead of 1 pass through ULQ). Nice palindrome :)

LAG: There's several ways lag affects GO run success rate:

A: Lag weakens EE (amongst other effect/splash gems). EE's stun strength comes from being able to stun enemies already stunned, thus compounding upon itself. An EE straight on path covers about 7 rocks. But if there's 2 rock delay, there's only 5/7 coverage where stunned bat can be stunned again by same EE. And when it's hitting 4 rock delay, it's only 3/7 - often I see a bat stunned out of range of all EEs near exit. Obviously, this is stronger effect the smaller the EE group is, i.e. under lag 2EE+2EE stun rate is reduced more than 4 EE stun rate. Based on what WGW mentioned earlier, even pure damage gems are affected.

Empirically, that minimizing game can turn a loss into "all die before c5" is pretty strong demonstration of lag's effect on gem damage (I often do this near 39 if the maze is leaky or nearly so).

B: Lag+autoclicker = more lag. I've had emerald runs frozen, to the point of having to just close browser window few times, in higher waves because the game just stops past certain point. On that 196 w/o GO run, I used just regular clicking in many of the later waves, alongside only briefly targeting autoclicker at the game in some (though this is risky as I can't see whats happening). Can't buy 49 lives in 192 if trying to do so locks the game :v

C: Obviously, it takes longer to complete, which means less runs and more mistakes made due lack of focus. Depending on how much is going in background, also affects earlier waves - for ex., had a 20th wave completed in 1800 time today; slow when I've gotten to that in under 800 many times before.

216 - 2020-06-05, 08:18 AM

well, that run wasn't perfect, i'm also doing mistakes. But avg leaks(simulated) was 0.95, 30% succes chance, and adding next EE (you can see i had PE and FsS already) would improve it. 184 was only wawe with 0 leaks ,and it was even before 2nd BO, which i build between 184 and 188( in 188 1 leak). I got many wawes with 1 leak(180 also 1 if i remember correctly) ,but more with 2 (something with 3 too).

leaking 2 in 192 is really dangerous, need 13 clicks with moving mouse from 5 to 1 life button. I died with 36 lives, so i moved mouse too early (maybe even miss 1 live button),but i think i had not enough time for 13 auto-clicks anyway.

I did some mistakes by rushing with gems(and spam repicks, i wanted to get PE), like that GT taking 2 places(let's pretend you don't see that) and righmost jade wich was 1-turn suprise out side od opals, but was help slowing.
EG on dowline was forced early, i was out of gold and taking FdD here was only reasonable choice.
EE closest to mid was 1st(90+ kills), i had FdT and FsS before.

Main EG was near that 2 BSs, lvl 22(maybe 21), BSs were 3 and 8.

Opals: i tihnk 5 PO is need in no=GO runs , 3 for buff, 2 for BO.
that 4 EE and GR in ULQ was out of BO range(but EG was in range) and they had some lvls(at least two had lvl 4), builing 3rd BO in upper part would reduce leaks from 0.95 to 0.76 about.

i rarely have lvl-ed EEs here, on this run dowline was flooded by PEs and other gems early, so i needed build EE in other locations. I couldn't range all gems that i wanted by BO, making 3rd would require next PO.

actually before 150 i had 2 POs which range all my gems, but i needed make another 2 to be ready for removing PO from dowline (also wanted EE in this place, near forced EG, it would reduce leaks as much as 3 rd BO, so my intuition was correct).

Jades: slow on c1-c2 is less important than around c4, which was good in my run. Maybe SYS could be good to slow on c1, i didnt make 2nd, because PE and PA flood early, this also caused having 3 jades only around lvl 80, so i didnt have much time for building SYS.

EEs: When i said to build more on URQ? i said to make 2 stun zones: on c3-c4 and c4-c5.
Also i was hoping for more EEs, so i used absolute minimum for that(2 for c3-c4, 2 for c4-c5), which is good for GO runs, but with no-GO it's not always enough. center + 3 line in not nessesary(its not bad, but it's also not optimal, sometimes single gems are good).

wait what is c7?  On c6 - exit sometimes 1-2 EE are needed to stop last enemy. Remeber some EEs can hit other unit before c6 so only last 3(righmost) on downline was ready to stun escaping enemy.

wawe 39 - it's generally hard, sometimes i have few EGs and EEs and still leak. but 1 BS ussualy solve problem, SYS is also helpfull and... uranium, which i rarely have at this point( but need maze around).

with special maze U235 with 350+ kills can solo wawe 39(tested in sandbox few months ago).

i was using this maze since few days, i often change maze, but they are all similar.

This is my 1st succes : https://imgur.com/rBzAZPm

(it was 2 feb, not 3,so 02.02.20 2:53 AM, shame that not 2:20  :)
I finished that run few days later btw.
I used many other mazes.

i think _mmkay is doing it better already.

i dont see reason why lags can affect GO run succes rate (possible to have much over 50%).

lement - 2020-06-04, 07:52 PM

@_mmkay:  Yeah, with C1-C5 stun strats I've also noticed that killing speed is needed; otherwise can't do the 2nd pass.

A: That's another reason why I like first BO in upper half, then second as near to top as possible; in addition to allowing placing first BO (that ensures ~all gems get the buff) sooner it also boosts the c1-c5 towers. (Though I think main damage towers should be bit lower so they can reach downline at least.)

B: Current run got early GT (wave 21-23 I think), in wave 73 I have one near mid (not as good location as before): 78 kills, 425 damage to up to seven targets; making it the gem with highest kills at this point (got lucky early on with 3 extra 1-turn chipped specials). Problem with GT is that the kills mainly accumulate on highest damage gems that hit the weakest enemy the longest, so my previously better placed GT didn't really get any kills due its entire range covered by about-same-time EE on one side and EG on other side, plus also bloodstone that does more damage for earlier stacked-together ground units.

*tabs back*

....And the run with my strongest GT yet lost somehow (Damn, I was hoping to build thousand damage per hit on that thing....). Did I tab away during flying wave while distracted or missed a rock and left a hole to land? What the hellfire and damnation >:@ D:<

@216: Yeah, this and lag is why I have doubts about hitting 196+ :P. 50% win rate of GO run? Naw, not reachable¦.

❶ I use (ESR) firefox for main browsing but always chrome for GTD. This allows me to restart the former (if needed due memory leaks; imgur browsing in particular hits this due their coding) without losing run in latter. There's some slight differences where you play, i.e. the gems PO covers in sandbox and normal game are slightly different (one doesn't cover five and half up and half left, other half right instead).

❷ I do wonder about shorter mazes, but tbh the maze I last posted already had like ⅓ deaths in first 40 due leaking ground so its really one of my shorter ones, especially early on (I can't built it as deep as in SS typically due choosing to pick both flawless Amethyst and flawed diamond typically as "best of bad pick"/earlier EG). Though Malachite would also be stronger there, not just because of raw value but also because with silver when the hitpoint is four-five gems behind the dwarf the slow is never applied.

One thing that helps is tabbing away/minimizing the browser window. This can take maze from "leaking 2 ground" to "last ground dies around c5". Would also help with ¦, but requires kill rate high enough to be secure that no more than 1 leaks even with bad luck (on pre-196 waves).

Typically I take wave 39 as test how strong my maze is. If it leaks ground, it is weak - though I only cancel runs in about first dozen waves when forced to take a chipped, and sometimes not even then; i.e. I don't do the "EG/EE by 44 or bust" - which goes with your 1-2-3 EE locations, though as aforementioned c1/c5 and GT comments show there's downside to early high mid kill building. With the 196 without GO run, I aimed (and succeeded, I think) building damage/kills around the lower right quadrant, as to grab and kill first enemies above on c3-c4 and then around c6. (The main EG did hit mid ofc). (In fact, it looks like your 192 run also built damage with EG mainly around there).

Still, while your and aznist mazes are both very short I'm don't like to reuse mazes too much :P No way I could stick with sameish maze as long as mmkay has, let alone someone like WGW (I don't know about you as you don't really post mazes so can't see how you change them over time). The better EE/EG/GO locations are already constraining!

❸ That doesn't mean I can't find them interesting, though. The 192 run picture has many interesting things. You focus on EE locations a lot, but my attention is drawn more by

O➀ High amount of BO/PO. 5 PO, 1st BO on wave 60.... This is a choice that decreases your odds of getting past 196 in favour of increasing your odds of reaching 196 in the first place. You reduced your BO rate later on though ofc. Albeit in that shot both BOs are centrally placed - I try to save those spots for BS for gold, but it'd make sense if you weren't even trying to cover all gems with BO - I guess the topleft EG is barely in range, but nothing past that point.

➁ I'm also suprised at you getting 0 flyers leaks at 184 without GO. I think I haven't managed that yet (then again I've never taken 2 BOs by then).

➂ With my recent mazes, there's exactly single place I can place the PO to cover everything needed so I can cover everything in lower half with 2nd PO and not impign the locations of jades/crystal/etc (in your 192 run the upper PO could have been higher to cover c4-c5). This means I prefer that they go third (in case I'm forced to attempt 196 without GO), with second covering the lower - there's six places (3x2 above c6 and 1 to the right) PO can cover 3 below 1 before c3), and first is acceptable anywhere that covers my main damage gems.

J➀ Jade locations. 13 jade isn't low amount, but that's definitely not how I place my jades.

➁ I do sometimes accept central jades just to build gold (after all, they're going to pay off in like six waves), especially when low, even I'd prefer them to be elsewhere. Generally, I aim to take advantage of their extra range to reach places EE and EG cannot, though.

➂ But you have no slow nor lines of jades in upper left quadrant; it looks like maybe 2-3 flyers get slowed and rest go through the c1-c2 at maximum speed!

(Reminds me of earlier talk of me wanting them to cross c3-c4 faster so it's open for the ones coming around for a second pass.)

➃ Also, doesn't look like you practice "downlining" them either; closest is the 2nd colum in URQ. It looks like jades are placed not to maximize how many flyers they hit but by how much gold they get.

E ➀That's not to say EE locations don't have few surprises.  In particular, lack of them in upper right quadrant after telling _mmkay to use more than you have with same amount of EE is quite surprising XP

➁ Also bit surprised that six EE on downline weren't enough to hold flyers. Though rather than putting some EE on C7-exit I feel better placement would be catching one or two earlier; the 3 gems above mid has comparatively low EE coverage in that maze.

216 - 2020-06-04, 08:15 AM

216 - 2020-06-04, 07:11 AM

216 - 2020-06-03, 10:18 AM

Yes. with GO and BO i leak 0 or 1 usually (2 and rarely 3 also happens). 
I'm trying to make downline much eariel, often since my 1st or 2nd EE. That 15 don't look as bad as leaking 3, but:

- 2 and 8 looks like redundant, 1 of then should be enough

- 9 alone rather can't stun flyer between c3 and c4, need 2 nd EE to help (worth puting single EEs closer to c3-c4 path for that).
maybe 4 could help , but zone that can be ranged by 4 and 9 is very short, so it's unlikely that flyer get stun here(if 4 was closer to path problem would'n exist).

- 5 might be redundant (only in GO runs) , at least in c1-c2, after c5 it helps ; i think if 5 or 7 was higher it would work better( actually not 100% sure).

- you can have more on dowline at this point( maybe 1 more somewhere between 11 and 14

- on later EEs also worth make stun zone on c4-c5 fragment.
maybe need only add 1 EE to help the "4" (than new EE can't range c3-c4 stun zone).

However lack of dmg gems/their lvls (mainly EGs) or maybe bad locations of them also can affect leaks. Worth to have many EGs between mid and downline.
EEs also have good dmg, but lvls on EGs are better.

I copy leak table from my eariel comment(over 80 comment ago), when i had ONLY 3 EEs in lvl 97, then started getting o lot of them:

44	0
48	2
52	3
56	4
60	0 (1st BO (1-t))
64	1
68	1
72	2
76	3
80	2
84	4
88	4
92	3
96	3	
100	3
104	2
108	2
112	2
116	2
120	2
124	1
128	2
132	0(2 nd BO)
136	0
140	3 (all with low hp)
144	1
148	2*
152	0*
156	0
160	2 too close

so yes, with very few EEs or bad luck(look at 140) i leak 3, but leaking even 2 is dangerous for me.

_mmkay - 2020-06-03, 08:24 AM

216 - 2020-06-02, 07:17 PM

216 - 2020-06-02, 07:09 PM

@_mmkay  About great gems:
Generally they worth exept GD and GE, because only 1-turn,
GAq: slighly higher dps than even EG, worth to have it close to mid, carefull with very low range.

GA: above EG(and GAq) dps, good range, what else to say(shame that it gets kills really slowly, only can kill air)

GR: low dps, but x3 targets it's slithly higher than EG, not bad range. U may take them.

GT: quite good dps as multi taget gem (55% of EG in 1 target), higher than GR because x1.2 instead x0.8 multi, and can hit 7 targets, so if good lacated this gem can be helpfull. I very like GTs.

GS: IMO sucks, low dps (36,6% of EG), but sometimes i keep it, it can slow from long distance.

GE: not good dps as single-target gem (worse that EE), good slowing poison, but taking it meant 1 less EE.

GD: absolutely usseles, but you can try get it on wawe 1 on sanbox and have fun with 20k+ crits in lvl39 (easy mode, on survivel about 13k crits)

GO:(no need comment)

i can repost my math comment if you want.

I'm still impresed how good you are doing no-GO runs, 
but leaking 1-3 with GO is a lot, especially with 14EE at 128, i sometimes leaking 1-3 without GO at this point(but always eventually got stack). 
I have 2 pics from my no-go 140+ fails , maybe i will upload them soon as well some 196 pics since after your 196+ run they are not big spoiler.

really 200k+ gold at 128?

Chance for 1-turn paraiba is about 1:12k (unlike other specials), the same as getting 4 the same flawless, so very rare.
How many kills it steal? i know that paraiba can got about 100 kills.

ofc if some gem got x kills it doesn't mean main EG would have x more, only part of it, so it matters les than you think.

"bats were dying too slowly in c1 stun zone", did you have a BO and rage RC here? actually upper EEs dont have to switch on 1 st flyer than come on c5, can on nect ones.

I got 3 the same perfects several times, once i got 3x PT + 1 PE, created GT and regret later (one of my 203 scores).

_mmkay - 2020-06-02, 05:37 PM

@lement Aw man that sucks :(

I also just lost a GO run. I know I said I wouldn't try anymore but I had one open in the background from a while ago and got a GO so I decided to keep it. Reallllly unlucky stack on 128. It's so depressing to watch a normal bat catch up to a slowed bat at the very end. Had 14EE, 6EG, 3Blood, 2YSY, 11Jade, 1Tourm. Repicks were only at 1500 and I had 200,000+ gold. Was only leaking 1-3 in the previous waves. I got a 1-turn Tourm early on and that soaked up too many of my main well-placed EG's kills to the point where a different EG was getting more. I don't know if that would've mattered in late game, however I definitely noticed that bats in my first stunzone below C1 were dying far too slowly for the EEs to switch to a bat below C5. I would've made more EGs there to fix it probably.

Also I see you guys talking about GTs a lot. What's so special about them? I don't really know anything about them. The only great gems I keep are GO, GA and GS but that's just because I don't remember ever getting much of anything else (except in my 200+ run where I got 3 perfect emeralds in one turn :o)

lement - 2020-06-02, 04:46 PM

216 - 2020-05-28, 09:02 PM

_mmkay - 2020-05-28, 06:35 PM

216 - 2020-05-27, 06:11 PM

lement - 2020-05-27, 04:47 PM

Thanks for running the tests (and bothering to add in the gems necessary for that, at that). That 21,77% is way better chances to beat 196 without GO than I expected (hopeless). I guess in late game it might be worth taking a fourth perfect opal to have 2nd black opal to have some chances of salvaging the run.

And interesting to see that not making the BO was better choice - still, 5,16% chance is not much to write home about.

Also, that BOs+GO in mid seems way too good - that's 93,32% success chance, but I did have some quite amazing luck with emeralds(26!) and perfect amethysts(25), plus five great gems isn't bad either. Didn't your algo max out at 90% with several better placements of emeralds shown?

But, timing wise, heh...No way I can do 0-196 run in 1 day. Maybe the last 40 waves?

Also, in retrospect I should record more gifs; it allows me to spot details I wouldn't otherwise due laser-focusing on "buy lives" and "will it stack".

Regarding last enemy or so, I'm thinking that maybe there needs to be line of emeralds on the centerline. Main thing stopping it is needing GO locations and wanting blood stones there, but....Maybe a staggered line is possible - emeralds right after each other are only so-so anyway due 1 gem lag guaranteeing that flyer will be in range of both when the hit from first lands (at least it is better than with land mobs, where I get 4-5 rocks worth of hitting lag in late game.

And ah, now I can see it. @_mmkay: As you might guess, I'd place more of the emeralds near start

However I must say your line of emeralds is amazingly regular - ten in a row, nice. Also pretty long - I leave out the space directly below c3 because then GO can't cover both c1-c5 and those space - it looks like your rightmost EE is uncovered by BO.

That's an interesting idea you have there with ABS in lower right - they're mostly not covering the land troops, so they'll not stack kills. Not bad idea! (especially considering the primary killbox focus on lower right I did on that run)

216 - 2020-05-27, 09:17 AM

I think "failue" is not right word for 196 no-GO run.
Very interesting run, stun zone on c3-c4 could help.
Simulation if you could do that BO without removing PO:
0:1004
1:2242
2:1207
3+:159

,but removing that PO hurts:
0:1002 (yes, i was waiting unltil 1000 succeses)
1:7893
2:11473
3+:4214

not building 2nd BO :
0:1001
1:6546
2:9093
3+:2778

BOs and GO in mid:
0:1034
1:72
2:2
3+:0

mmkay's link is broken , but i figured out how to fix it, just remove last 2 letters.
Both of you are playing better this game than me in february, but have worse luck.
My first 196+ was on 03.02, so over 110 days while got 23 GO runs.Mine GO luck was good and runs have better time,so was possible to get from 0 to 196 in 1 day , even with huge breaks(ussualy 2 days avyway).
also i was losing no-GO runs fast, in 110-130(many not sumbited), so they last 3-5 hours, was possible to make more than 1 attempt in 1 day( if i really wanted to).
Aslo i'm doing other things while runs(plaing other games too), just every 3-5 min start new wawe, but i'm rather not plaiyng on laptop (this game take all CPU for you? ).

That 2x2 EE sqare wasn't ussually enough to stunlock last enemy.

Also one of EEs was outside of all POs range, unbuffed EEs are very inefficient, maybe need more than 2 PO on map.

With few more EEs( and POs) succes chance would be quite high.

lement - 2020-05-27, 04:20 AM

Well, that run came to an end; another failure.

I decided to go for silver run this time - to save myself from building a SYS for more gems. It hurt kills-wise, but did end up needing the space.

I also decided to use Perfect Amethysts for additional damage rather than making Golds. But ironically, lack of gold ended up smarting.

---		95 https://i.imgur.com/JpLURLH.png

Been a while since I've done "circle around the outer edge to reach c1" maze. The idea is fun, but ended up wasting kills and leaking one land in starting ground waves due the time it took to complete the seven rings; the enemies don't stick around the silver just not as long as they do with typical emerald. However start is more convenient.

The ABS was stupid mistake - I forget that I had normal Amethyst already taken twice when picking gems.

The central Uranium was gotten off single pick. Then I decided to make the other one in c4 corners since I already had taken two parts.

The paraibe I  hoped to hit c6-c7, but it was out of range. A more central location would have been better.

----		131 https://i.imgur.com/5v9Y3zi.png

At this point I had concluded that placing those central blood stones might have been a mistake due lack of space; as perfect ametyst strat needs up ending more space. It'd have been fine without needing to cross back into center. Though on the other hand, very good places for perfect opals right next to mid.

speaking of which, the upper PO next to jade was intended to have location swapped with said jade; the upper-right emerald ended up uncovered by opal due that.

159 https://i.imgur.com/oa5qlqD.png

The  emeralds on downline got complete, also got exit covered by EG (though I wish it was 1 to southwest).

Got a second not ideally placed red crystal off single pick along with perfect amethyst; decided to go for first due the damage difference being small enough that extra armor reduction would be better even assuming I'd get third with all that gold (but that ended up not happening).

Also since I got third perfect opal, went for 1st black opal.

----		179 https://i.imgur.com/jvYemhm.png

The Amethyst location was bit unfortunate since I wanted to build a line of emeralds hitting center and rightmost columns, but I figured I couldn't affort to be picky.

Also decided to add an EE on c3-c4 line as I figured the flyers would be dead and by the time the next ones came down and I was missing a line of emeralds in the five groups between c5 and c6.

187 https://i.imgur.com/z7mRvRh.png

Went heavy on repicks since little but PE, PA or GO would do at this point.

191 https://i.imgur.com/lhgY4TR.png

Lost fair bit of gold. Decided to take the wasted turn with perfect ruby(though i wanted an EE two gems to the left), which was unfortunate but repicks were over 4k at this point. Also wanted one EE covering c0-c1 though that wouldn't be covered by GO so placed it last in priority.

At this point and on previous wave it leaked two.

Even with all shut down, autoclicker tends to freeze the game way before this point, but thankfully with so few leaks can just pick gold manually.

196 (5,6 MB gif) https://i.imgur.com/Dz3M3OH.gif

With 4800 gold per repick, I was finally out of cash. Hoped that by taking but not upgrading the BO, the perfect opal buff would remain as it often does with combining jades so giving low odds there, but no luck. Maybe taking the BO special was a mistake; not upgrading promptly definitely but oh well.

Well, that's that. New 195 score on leaderboard in last place, at literally 25% worse than next one.

No GO, but then again its been months since I last saw GO (and I think the total nr of them I've gotten out of sandbox can be counted on one hand) and not like am going to be doing better than this without it.

I guess I shall go play some other games :v

@_mmkay: It seems your imgur link is not displaying for me

Regarding jade, a thought: might be worth doing a line of jade above c6-c7 line, as to ensure they briefly hit each flyer as to spread poison around.

I also used dozen jades, but I think more jade in general for gold might be wise to get better GO chance.

@216: Honestly I think the main reason why the time score is so bad is playing on laptop. WGW did mention time scores getting 30% better just by using a better computer; at this speed doing 10000 runs would require 192 years assuming a relatively high pace of run per week XD

Also seriously 23 GO runs in 3 months? Assuming 50% GO rate, that's 46 runs in 83 days. That's some specs to do a run in only 2 days!

PS: Btw I think the rest of aznist shots might be available in wayback machine; I know I have 160 pic of his run saved (differene being mostly less edge EE and less jade on c3-c4/c6-c7)

216 - 2020-05-26, 04:58 PM

_mmkay - 2020-05-26, 03:44 PM

216 - 2020-05-26, 12:23 PM

_mmkay - 2020-05-26, 09:53 AM

216 - 2020-05-23, 06:57 PM

_mmkay - 2020-05-23, 11:04 AM

216 - 2020-05-22, 08:34 AM

Even if u have most dmg zone between c1 c5, ingoring c2-c3 is not recomended. When too many enemies is coming in group around c5, some of them will leak. So need to stop some of them eariel (no even need to stun lock and kill them here, they can come later to your dmg zone).

So your maze was even longer before, maybe it's reason of bad time score.

GE: i meant that GE can't slow all units, because one ememy get stunned and it keep atacking it. Anyway GE sucks IMO, lower dps than EE( i can repost my math post if u want, i copied it eariel), so i prefer build 1 EE more.

SYS: Even with speaded jades kepping slow on all is sometimes impossible, because of relatively high range too many jades can focus on one emeny, so SYS often is heplfull.
single target dmg of SYS is 3x weaker than EG, so its not important dmg gem in late game ( but i don't mind if it gets some lvls early).

Aznist's run (i saw only that 1 pic) wasn't so optimal, but some of his ideas was quite good. Limiting EEs range often worth, 
imagine all EEs have unlimited range -no reason to build more than 6-8, rather impossible to beat 196( unless very good dmg gems, like uberstone ?)

I will tell more about c3-c4/c4-c5 stunning after you finally get past 196 (I wish you more luck with GO, since february i had 22 GO runs (23 if u count than closed browser run, ),many runs had more than 1 GO, so i have good GO luck).

I was also thinking about local optimums, actually was starting sim again each time was changing something in algorithm, ofc solutions wasn't extractly the same, but i was looking rather for general rules of locating EE, coying postions is sometimes impossible.

You used GA as short of genetic algorithm? - fro me GA = Great Ametyst (GAq = Great Aqua) , so i didn't get what u mean.

lement - 2020-05-21, 11:06 PM

As far as downline got to make second 2x2 box there (with 3 EE 1 EG), though the concept of c2-c3 slowing feels bit flawed, as I want the center to be empty when the fliers arrive there 2nd time - often I've had a moment where I wanted them to pass by faster just so they'll not take up damage that should be focused on leading bat.

Thus I think using just emeralds is not entirely witout merit (but would still prefer EG), given that, as each emerald on target bit more than doubles expected stun duration - so adding an overlapping one may make the enemy go from "darn, they're both in mid" to "start moving after the main ones have passed"

(Though I mainly built those EEs above downline first because at the time my maze extended under them (which meant that the selected EG actually covered more land ground than the one above 2 ABS). Which is fine at last third of waves, but may make getting to last quarter harder).

Anyway, you're right about being able to place the EE above GT (and half to the right so it hits both c3-c4 and c5-c6); good point. GT point may be optimal for area coverage, but it doesn't hit corners which means the EE would switch around. aznist even deliberately chose to put them closer to path and avoided building on GT location or under it despite being able to.

Regarding slow: I've been thinking, while starting slow splash may be useful, SYS is mainly more of a damage gem there due jade coverage (stackish enemies on c5, and c1 is only point where fliers go backwards); plus with spread-out jades the enemies are generally slowed anyway.

And as far as damage from SYS goes, bit of time ago got SYS and GS at around same time relatively early - and GS racked kills up twice as fast as SYS as game progressed. Which led me to thinking 'hey, perchance perfect amethyst outdamages both on fliers'. aznist only focused silvers in center, too.

And yeah sadly GE placement was not ideal here (it should be closest to one or other flier arrival direction due how poison works, but for best effect should put the upper jades elsewhere), but I think it was okish since got on track to 23 at 196 and it racked up acceptable 60 kills, putting it in top six towers or so. I think the strongest emerald was at like 79 kills (also got early EE below the selected EG, so I was quite surprised to see how much more kills the EG got)

I generally try to spam BS around center to conserve gold, otherwise yeah would put EGs there.

Red Crystal: Perhaps waiting till 180+ may be flawed given my GOless run ended at 180 =P Though it also depends on whether I happen to get it early in the lower part of map; in which case it doesn't cover quite as much.

Yeah, not blocking other gem locations is good idea; just was wondering what you intend to do with the leftover spots. Though there's that singular emerald and singular SYS slows are, on the baseline, roughly equal, so may get timing issues with them not included. Also I'll note that your results are based on the high-damage EGs locations; i.e. the results with best EG in exact center, covering c1 and c5, or covering L-exit area + mid would be different. Maybe varying the locations or kills on further test may be worth considering.

Last comment was about genetic algoritm; to elaborate it might have found local optimum for spread-out slow, where moving any one emerald makes things worse, which may prevent it from finding actual optimum that could involve, say, moving all emeralds by half a rock.

There might be some math method to compress the calcs with same RNG; might be worth pondering for speedup.

As for now, got another run going. Though no idea if I can capture when the lag gets downright 'frozen still' (but well, that's pretty standard in flash based GTD), but I plan on spreading jades around more and putting more damage on exit as well as I previously did - focusing closest to 96 BO is not worth the risk from excessive clustering, I guess. Also maybe using some EE only hitting the fliers once, but for long; hebe also demonstrated it (though they used 14 ABS with just 7 jades).

216 - 2020-05-21, 06:12 PM

Jades: i like your jades locations, but having all here is not best idea, since it's main slowing gem(also remember it help stunning). Slows around c4 is important.
I would say that 4 EEs in 2x2 block are a bit overconcentrated, but with that jades its definitelly unoptimal.

GE: making GE mens sacrifice of 1 EE, having 12 EE at 147 is not much, so i would downgrade it( or just keep perfect, unless u got it from 4 flawless, like i yesterday, then downgrade needed). 
GE here could be good for slow all enemies, but because of stunners it didn't work i think.

EGs: they get most kills from ground units, so shape of maze have most impact on kills. Having EG near many EEs makes it lose kills to EEs (which is not very bad, EE dmg is not much lower than EG).
that your EG above ABS was well located to farm kills - range main part of your maze and have only few other gems near that would take kills.
So i'm not suprised it got many kills.
Also having massive dmg in mid is good, so spamming EGs around center is often good idea. 
Having parts of map not in EG range have bad imnpact on time, but often don't have bad impact on leaks - low dmg zones actually dont need armor reduction, however if an enemy is stunned in low dmg zone for long time, it's frustrating.

In that aznist's pic than mmkay linked i also saw zones without EGs - they cause long time waiting for enemy kill, but it's not major mistake.

EEs: looks like u ingore downline competly. Ofc is no needed build full downline, I understand your strat by stunning around c1/c5 - its not bad, but you can also put something on donwline.
I dont see other way for efectivelly stun few enemies between c2-c3 (u made 2 EEs 1 tile above line,so i think they often was hitting the same enemy).

You said that many enemies was stacking around c5 - i often experience it, to prevent it need make some stun zones on c3-c4 and c4-c5 parts(also need slow here). Not much EEs in needed here, they 
just can stun only for some time to separate enemies (and delay their reaching c5 moment).
You could put EE in your GT location, but it wasn't nessesary - other (more or less nearby) locations work also well
(i don't want spoil extract positions/amount here now, actually i don't know optimal solution yet and its different in GO and no-GO runs).

Waiting for RC : i don't take 1st part of 2nd RC before 180, often 190 ( unless forced by lack of gold)

19 EE simulations: i used also 2 RC(for armor reduction, armor mechanic is fully implemented), and ofc GO - in constant positions(also 5 EG with high lvls had constant position + 1 SYS , i think i should make 2)
Rest was lvl 0 in (initialy) random generated positions and was testing many times, avg leak was a score, then best solutions were evolving.

i didn't made other gems to prevent bloking potential good positions - making BOs was poitless, neglible dmg and i harcoded BO bonus to all towers(for this testing), paraiba also not very big dmg.
in normal game u make some BOs, Paraiba and other gems, so i think 19 EE and 21 EG was good replacement of all dmg gems that u would have at 196, actually is possible to make 21 or EGs with low EE gain.
Why extract 21? - i started with 20EE, 20EG,14 jades, then increased "difficulty" its not a special number , just replaced EE with EG.
Scoring very good solutions was hard, leaks are always random, need very much amount of runs to score gems. If avg leaks is around 0.1 random factor is too significant, ever with 500 runs, need more but algoritm work too slow then.

i will do more testing like that - notice that even if i find super optimal poitions, copying it extacly may be hard/impossible. You know sometimes u can't get last EE part extractly where u want.
so need to find many good solutions.

I didn't got what u mean with GA testing (at end of your comment).

lement - 2020-05-21, 11:18 AM

Alright, this time got score posted of non-GO stack fail at 180 (shot of maze at 147 https://i.imgur.com/wu58b5p.png)

Frustrating, since I'm sure I'd have survived without lag from the way the fliers were moving in stuttering motion. Even shut down things.

First new experiment was getting three well-placed perfect opals, and then sacrificing the first for edge-case BO to cover all gem locations in upper half that second BO in mid wouldn't cover. I guess it helped a little, reduced flyers leak from 5 to 3-4. (Maybe should have stopped hoping for 1-turn red crystal). This allows BO earlier than going with low BO then upper one, as there's more gems out of range of best-placed BO that hits all 3 gems between C1 and C5.

{Though the location may be bit dubious, considering Aznist doesn't use it at all, which allowed him to place GO lower and thus fill center with killzone - and of course, his shots kinda show the concept of two lines of EE in upper left lung may be flawed, as he deliberately chose to place extra EE directly on the C1-C2 path rather than somewhere they could hit both. Loooking at shot, tbh having at least 1 jade in lower right might be good for poison as well to minimize non-stunned movement.}

Second was having even more 121+ range gem locations in upper left quadrant between the double lines for emeralds, with land entering in the middle. The line tips by half a stone so that the EGs have the fliers between C2 and C3 barely in range and long distance to hit them when coming down from C5. It seems like the one EG above two ABS turned out to be supremely good, starting at disadvantage relative to the early EG in shot but almost catching up in kills at the time of death despite far less stun in area. Will definitely be focusing more on having that location for EG in future.

Additionally, I'm thinking the GT location should have EE; lots of times I saw multiple enemies go through c5-c6 relatively simultaneously; having something that would slow down one as much as possible would be useful to stagger out the fliers.

Of course, the early first EG was helped around by already having flawless opal and flawed diamond, and not going for maxing malachite early this time (the malachite ended up about 170ish kills; about 80 kills behind what I'd expect, though, so idk if it was worth it).

Overall, I'd want to use it again, though am kind of bothered by last bit of hitting C6 not touching it much.

However, downside of strat was force concentration in the lungs, which goes against spreading out ones' emeralds.

When I started counting "what gems will I pick for final rush" I also ran into the situation where I realized I had picked one too many Perfect Amethysts.

Though tbh damage-wise it might be better to pick many of those, once all bits of path have been covered by EGs. As WGW once states, the initial early combination actually tends to dip your damage, and even without that 2 and definitely 3 Perfect Amethysts will outdamage EGs.

On another note, gratz on putting together AI for it. It's been something I've been wanting myself, but dismissed as too much effort. Sounds awesome, tbh; I'm surprised to hear that you can reasonably possibly get 196+ run without GO, though that must be some good placement. The GO one also validates aznist's running around with large chunk of jades. I do wonder if any of the gems would have been out of range of GO

Also, 19+21+14=54 gems out of 64 3*specials. Were the rest left empty? I guess you could fill it by 2 red crystals, Tourmaline, GO, initial malachite, 3 black opals maybe, 3 lucky specials though this leaves still one unaccounted for.

Oh wait, Yellow Stars and Uranium in centre; also ABS though tbh part of the reason I pick them is gold. I do wonder how regular BS would stack up; though both it and Yellow Star might need to consider lag for splash damage. In any case, sounds pretty awesome; I'll see what results from it. You might want to try to run the GA from nothing several times (or with some seeded start locations) to see if you get same results each time btw; it might be that it hit upon local optimum.

216 - 2020-05-21, 06:39 AM

216 - 2020-05-20, 08:19 PM

_mmkay - 2020-05-20, 03:19 PM

216 - 2020-05-12, 11:02 AM

I failied a GO run in 168, becuse i once forgot to turn on autocliker, i had 2-3 s between enemies, but i lost some time to realize that autoclick no work(clicked mouse button 5 than starts autoclick then realized it no work and i started clicking manually, but did 1-2 clicks less than needed).

I couldn't believe it can even leak 2 with 20 EE, i reconstucted gems in my sim (and set hp of enemies to 206967, but forgot about GAq with some lvls that i had and didn't know extract lvls of EG's, i used less lvls probably) - about 7% chance to leak 2+

i didn't have 2nd BO yet, but most important gems was affected with lvl 96 BO (i had only 6-7 EEs in 96, after that i made few jades mostly EEs).

The game still prove i'm a noob,
i used genetic algorithm (in my sim) to found optimal positions of EE(and some EGs and jades), i realized that i can speed up sim by about 10000 times, making it possible(why that game lags so much with normal speed...).

Maybe i didn't found really optimal solution, but better than i could found manually (i won't spoil it yet), 
but 19 EE was enough to get 90% lvl 196 succes chance(with 14 jades and 21 EGs with 95 lvls total, all gems with BO buff).

also try to found solution for no GO, used 28 EE ,15 EG and 11 jades(that gems seems likely to have), but max around 60% succes chance found.

@lement: correct on wiki red vs o/aq dmg(0.8), idk how i did that much mistakes (4)

I understant that i colud think D vs R is 1.25 instead 1.2, but rest mistakes are bigger, i tested that topaz 0.55 many times and still made mistake...

lement - 2020-04-20, 11:34 AM

216 - 2020-04-20, 10:31 AM

_mmkay - 2020-04-20, 09:04 AM

Reading the code shows me that there's loads about this game that I knew nothing about. What's mana?

Also up until now I didn't know exactly how damage worked for gems. They have a range but I wasn't sure exactly how the range worked. Here's how it works:
let base damage = B, dice (randomly generated number starting from 1) = d, number on dice = n

Lower number = (B + n) * (1 + 0.1towerLevel) * (1 + damageAuraValue)

Highest number = (B + nd) * (1 + 0.1towerLevel) * (1 + damageAuraValue)
So if the tower is level 7, that means it'll do 1.7 times damage.

For EE: B = 98, d = 51, n=2
Therefore the minimum damage is (98 +2) = 100 and the maximum damage is (98 + 2(51)) = 200

I'm trying to read about aura (speed and damage) but I'm having a hard time understanding the code. Does anybody know if the damage boost aura from BO (0.4) and SYS (0.05) stack? I see a variable called "bonusAura" but I don't know how it behaves and all the towers with an aura seem to have bonusAura = true so I don't think they can stack or else you can just get loads of opals. I don't see any mention in the code of auras adding together.

216 - 2020-04-19, 08:00 PM

lement - 2020-04-19, 07:21 PM

_mmkay - 2020-04-19, 01:32 PM

216 - 2020-04-19, 01:14 PM

_mmkay - 2020-04-19, 12:53 PM

216 - 2020-04-19, 09:20 AM

Normal is easiest to get with max chances, becuse it can be combined from 2+ flawed (and it's the same chance for downgrading flawless as normal) . With 20/30/30/20/0 or lower chances flawed is easier ofc.

Imagine how much harder getting special gem was if no "downgrade" button.

I realized someone score today, so scores work and finally submited over 500 score. I think i'm not going to do it again, way from 200 to 500 is long and boring.

Fun fact: 1st my attemp to lose failied, i closed main road, removed few gems to allow enemies walk around, but anyway all got killed...
So i removed rock around c6 and forced enemies to walk away from gems +added normal opal to slow few EE to be sure. More than 1 was going to leak.

Defeating 196 without GO is very interesting, but i sugest to do it with GO first :)  , 
it's MUCH easier, GO is 1.3x more dps AND 1.3x more "stun power" on your EEs and jades. you can say also more gold proc from jades, but enemies also die faster, so i'm not sure.

The fact that your gems actually GAVE you change to defeat 196 is really interesting, maybe it was small chance (around 1%), but was real.
Now situation that someone get to 196 without GO and get luck in 196 (and die in 200 probably) sound very likely, especially with more than your 23 EE (with 190 repicks avg is 27 EE). 
To have more than 50% chance to succes rather over 30EE is requied.

You build 10 jades , i'm not sure if it was enough, but u got many EE early, so i think u didn't had time for jades(and u didn't know that u will get so bad EE luck later).
If not above avg EE gain i recommend 12 jades, if really bad EE gain maybe even more.

I wish you more luck on next runs and wait for your 196+ scores!

I want that "defeating 196 is almost impossible/reqiues unreal luck" myth to be completly busted.

_mmaky - 2020-04-19, 08:31 AM

lement - 2020-04-19, 12:58 AM

216 - 2020-04-18, 09:41 PM

@_mmkay: why you said you accept normal ruby, but no flawed ametyst?
The order from easiest to hardest(with max chances)to get is normal, flawed, flawless, chipped, perfect, great.

I accept 1 flawed opal - for uranium, and flawed topazes - ofc i want have it at last part of EE, but i make an exception for first few (use shapphire).

there is more than 1 (4-6)sqares on mid when i accept uranium.

Simutalion results:
your gems in 195:
0 leaks: 2 (!!)
1:28
2:81
3+:55 (there was some with more than 3, but all of them are counted as 3+).

after adding GO in mid:
0:132
1:29
2:3

i'm still using lvl 0 EG instead ABS(not programmed jet), and didn't know extract lvls of all towers, but even big dmg increase have quite small impact for amount leaked.

So your gems aren't as bad as u think, even with that concentration of EEs (which probably saved you).

@lement: i'm stating only 2 games at once, and around 20- 40 i choose more interesting run, i don't close that 2nd, but do nothing with them, often open new tab(s) in this window.(sometimes i come back to this runs after 1st is finished).

In GO emeralds runs autoclicked is rarely needed afted building BO, leaking 0-2 ussually, around 180 rather no leaks(unless bad EE gain).
In my current run i have no leaks since 140(but good EE gain), so if u can get to 180(as u show, u can get 196+.
Sadly u lost 2 GO runs by bugs/chrome :(

_mmkay - 2020-04-18, 09:02 PM

_mmkay - 2020-04-18, 08:48 PM

lement - 2020-04-18, 08:39 PM

Ugh, lost another GO emerald run at 140...To chrome deciding to reload flash plugin >_> (hate losing runs like that). However, your replies are appreciated, 216!

I have definite tendency to go "what do I pick out of those meh gems, I guess jade will do", though the larger number of jades was partly driven by no GO for 171 waves on that one.

I kinda want to wait with building 1st BO until there will be no future specials in its range that would be outside the range of 2nd BO (would have to go 3 BO with 96 BO otherwise). You're right about ground units not leaking at that point, though in post-40ish ground sometimes could leak sometimes even with full maze, if I lack in opals or other boosts. Though really, I only waited with 1st BO so long in that run due loss of perfect opal with no GO replacement.

Flyers...I'm afraid I've forgotten at this point ^^;; Though I see you did your own tests :o (2 might be bit low)

And yeah I've been doing multiple runs at once after your tip of starting multiple at same time :3 It can cause other bugs as well, sometimes in early waves time keeps running after last enemy dies or such.

Hm, seems first 20 waves we have similar times, but at 40 you seem to get ahead (I've gotten 2k 40th, but they're outnumbered by 4k 40th). Shortening ground entrance surrounded by EG followed by emeralds is an interesting idea. I find I get fastest early game times when I get best kills on emerald; recently managed to get 100 kills on emerald in first 10 waves: https://i.imgur.com/woMdgOF.png (though this was downleft strat maze)

Yeah I pressed the submit button but it didn't go through. Ah well.

Mazes are neat to look at though I don't really consider them spoilers - it's not like one can just go to 196 with merely knowing a maze layout+that you have to focus on EG, EE, jade; heck I've already changed my layout. Tbh this discussion is a bigger spoiler, though while I thank you for the compliment not sure I can do 196+ run - I've lost a 195 run with no slowers due autoclicker being too slow before, after all. Though interesting find about my maze stacking.

@__mmkay: Interesting to see someone else attempting runs; that's very good time - if you look around, you can see 15k at same time and even 24k time by me, and even 36k time on 103.

It seems you focus your gems quite tightly around center. Good for fast killing, but bad for stunning flyers.

I use Actiona for autoclicker; it has function to find image which I use to detect flying waves. Though given my experience, it may not be something to recommend....I do have another tool to use maybe.

Your second maze seems quite good....Though the second chipped special is in bad place. Gratz on 196 score!!!