HOME WORLD FORUMS WIKI VIDEOS
8310 members and stacking!
     Welcome guest, please login or sign up

2 Pages V  1 2 >  
Reply to this topicStart new topic
> PC after 8 lines (was: C-Spin with first T placed vertically?)
Okey_Dokey
post 4 weeks ago
Post #1


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



tldr: I am looking for a C-Spin / Triple Double Attack setup where the first T piece is placed vertically. Something like this:



I am interested in the concept of getting a Perfect Clear after exactly 8 cleared lines in a T-Spin setup. The nice thing about 8 lines is that you have placed a whole number of bags: 8 lines are 20 placed pieces; with 1 piece on hold it's exactly 3 bags that you use to arrange the placements. Despite of this fact, barely any build exists for 8 lines. TKI 3 opener usually goes for 6 lines PC, DT Cannon goes for 10 lines PC, 26_nameless BT Cannon even goes for 14 lines PC.

In order to achieve a PC the following 2 points must be fulfilled:
  • #L + #J + #vertT is an even number
  • #T + #OddClear is an even number
#L stands for the number of placed L pieces. #vertT stands for the number of vertically placed T pieces. #OddClear stands for the number of line clears that clear an odd number of lines with an odd number of empty cells below. I have tried to explain the reasons behind those theoretical results on the Perfect Clear Opener wiki article.

As illustration, let's have a look at the 26_nameless BT Cannon setup: It clears 14 lines with a DT Cannon into C-Spin into 4 lines PC. 14 lines are 35 pieces which form exactly 5 bags assuming that the first piece of the 8th bag is put on hold at the end of the setup. BT Cannon and C-Spin use 1 vertical T placement each. So #L + #J + #vertT = 5 + 5 + 2 = 12 is an even number. BT Cannon uses no odd clear (the first T-Spin clears 2 lines and the second T-Spin has no empty cells below), C-Spin uses 1 odd clear (namely the first T-Spin which clears 3 lines with 1 empty cell below) and there's no further odd clear during the 4-lines PC build. So #T + #OddClear = 5 + 1 = 6 is an even number.



Let's head back to the 8 lines PC. If you want to have a decent Perfect Clear chance, then you want to use the third bag's T, L and J pieces (all of them together). In this case the theoretical results simplify to:
  • use an even number of vertical T placements
  • use exactly 1 odd line clear (because 3 odd clears will basically never happen)
Again as illustration, let's have a look at achieving a PC after 8 lines in MKO Stacking. The first 2 T pieces are placed horizontally involving no odd clear (horizontal T-Spin Doubles are never odd). So, the following 4 lines must contain 1 odd line clear and the T piece must be placed horizontally - provided the third T piece is not kept on hold (like in 1st, 6th and 9th picture).


IPB Image

In fact, there's a nice 88 % chance to get a PC after 8 lines in MKO Stacking. However, most ways are nearly impossible to spot because pieces are cut in two halves by another line clear. If there's an odd clear in the last 4 lines, then at least 1 piece is always cut in two halves. So it's better to move the odd clear forward to the T-Spin phase. However, T-Spin Doubles with an horizontal T placement are never odd clears. So, in the following we will just look at setups that involve at least one T-Spin Single or T-Spin Triple that doesn't clear the bottom line.

One example would be Hachispin. Actually, you can achieve a Perfect Clear here. While the T-Spin Single is not an odd clear (4 empty cells in bottom line), the T-Spin Triple is. The first T piece is placed horizontally and the second T piece vertically. So the third T piece must be placed vertically (not so nice) but no odd clear is needed (very nice). With 69 % the rate is not as high as for MKO Stacking but the PCs are much easier to spot (at least the first ones).


IPB Image

Another example whould be the pwn_by_numbers variation of Hachispin. However, the setup is not that good for other reasons. Here the T-Spin Single is an odd clear (3 empty cells in bottom line), but the T-Spin Triple isn't (2 empty cells in bottom line). Both T pieces are placed vertically. So the third T piece must be placed horizontally (nice) and no odd clear is needed (very nice). With 83 % the rate is decent and the most likely possibilities are easy to spot.


IPB Image

Other Setups:
  • Mr. T-Spin's STD has one odd clear but places the first T piece horizontally. The overhang prevents any PC (it reaches row 8 with number of empty cells on both sides not divisible by 4). Even if the overhang is built differently, the chance for TSS -> TST -> TSS -> PC is just little (see Fumen for an example).
  • Pokemino's STD has 2 odd clears, thus it's irrelevant (we need exactly one odd clear)
  • Early I STD uses one odd clear but places the first T piece horizontally. The chance for TSS -> TST -> TSS is just little (see Fumen for an example).
  • DT Cannon has no odd clear. While a PC after 8 lines is possible, the chances are very little (see Fumen for an example)
  • BT Cannon has no odd clear. I assume there are some PC opportunities with a proper build, but PC will not be useful enough.
  • Imperial Cross has one odd clear but places the last T piece horizontally. I know too little about Imperial Cross as an opener to make a statement. I assume that you have to use one T piece to build an Imperial Cross. If you place this T piece vertically, then there could be a decent enough chance for TSD -> TSD -> PC.


Now let's come to the setup that is the reason why I made this post: C-Spin aka Triple Double Attack.



There's one T placement used to build the C-Spin which we will ignore for now. The T-Spin Triple is a vertical placement that results in an odd clear. The following T-Spin Double is a horizontal T placement that doesn't result in an odd clear. So we have one odd clear (which is nice) and we have one horizontal T placement (which is not so nice).

This means, if we place the first T piece horizontally, we can only achieve a PC after 8 lines, when substituing the T-Spin Double with a T-Spin Single (TST -> TSS -> PC) or by leaving either L or J piece on hold. Sadly, the first T piece will be placed horizontally in most builds. Example:



However, the Triple Double Attack wiki article also features a way to place the first T piece vertically: TOJ Core aka TKI signature (I guess TKI knew what he was doing). If you stack the second bag like in the following, then there's a 88.6 % chance to get a Perfect Clear with the T placed horizontally on the bottom (TST -> TSD -> PC ; note that T on bottom doesn't necessarily mean T-Spin; the overhang could be already gone when the T piece is placed there).


IPB Image

Any further ideas how to place the first T piece vertically in a C-Spin setup or how to stack the second bag?
User is offlinePM
Go to the top of the page
+Quote Post
iljain
post 4 weeks ago
Post #2


Tetris Apprentice
Group Icon
Posts: 140
Joined: 28-September 10



Grin.png

https://www.youtube.com/watch?v=V949jhwJb30&
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 4 weeks ago
Post #3


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



iljain showed a perfect clear after 10 cleared lines achieved on the standard SZ-start C-Spin. I am currently trying to write a T-Spin Opener page and would be interested in including the SZ-start C-Spin. However, I don't know much about it. But I think a perfect flat 6-rows high stack like in iljain's video would be nice. So, does anybody know how to achieve the following stack with either O or S piece not placed yet? (because this allows to go for T-Spin Triple or Imperial Cross.) The way in iljain's video doesn't count though because I want to think in bags: All pieces of first bag must be placed before the pieces of the second bag (in iljain's video the first O piece was hold until pretty late in the 2nd bag).



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

This post is about 8-rows high PCs with T-Spin Single into 2 T-Spin Doubles (TSS -> TSD -> TSD -> PC). It should send at least 2 + 5 + 5 + 10 = 22 lines which could be enough to take out a center 4-wide player in games without garbage blocking (combo doesn't delay incoming garbage). That said, it's not useful against center 4-wide in games with garbage blocking because the 7 lines from TSS -> TSD are not enough to keep the combo short enough. Again we try to fulfill the 2 points:
  • use exactly 1 odd line clear (because 3 odd clears will basically never happen)
  • use an even number of vertical T placements
The first point is a must-have for any PC. The second point is optional. If the second point is not fulfilled, then you must leave out either L or J piece in the PC.

I mentioned Imperial Cross in my previous post. It has an odd clear and a vertical T placement. If you use a T-Spin Single as prefix, then it needn't be an odd clear. A vertical placed T in the T-Spin Single would be nice to have but I found no way to do so (and I doubt that is possible because the Imperial Cross basically forces you to clear the second line which means that the first T-piece must be placed 1 row above bottom for a vertical T placement). However, this actually works with an S or Z start:



It's a 50% chance to get a PC after TSS -> TSD -> TSS. The last clear is just a T-Spin Single because you will need both L and J. However, if you place the pieces like in the 4th picture of the Fumen, then there are some ways to get a T-Spin Double at the end (either L or J is not placed) additionally to some ways to get a T-Single at the end (both L and J are placed)

IPB Image

Super T-Spin Double (STSD) is another way to get 2 T-Spin Doubles. STSD has no odd clear and one vertical T placement. If you use a T-Spin Single as prefix, then it must be an odd clear. A vertical placed T in the T-Spin Single would be nice to have and is indeed possible:



In the end we have two 3x3 blocks with a missing corner which has quite some ways to achieve a Perfect Clear:

IPB Image

And who says that you have to use a technique at all to get 2 T-Spin Doubles(?). If you just stack 2 seperate T-Spin Doubles, then you have no odd clear and no vertical T placement. If you use a T-Spin Single as prefix, then it must be an odd clear and it should be a horizontal placement. Something like this:



Again we have those 3x3 blocks with a missing corner in the end. The last Fumen is no serious suggestion, I just thought it would look nice because you the left half of the playfield doesn't interact at all with the right side (you are basically playing in 2 seperate 5-wide fields).

Generally speaking, I think 2 seperate T-Spin Doubles won't work out at all. You just don't know where to put the other pieces until finally the next T piece shows up. However, I think vertical T-Spin Single into STSD into 8-rows high PC could be useful, if a proper technique for it exists (my Fumen sucks because the first L piece does block the T-Spin).
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #4


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



This post will be about achieving an 8th-row Perfect Clear after a STD = Single Triple Double attack. The STD will become a Singe Triple Single attack in half of the cases though. Basically, the stack will look like this:



The second row is cleared first and after it's gone you will have a reversed C-Spin / Triple Double Attack stack. Mr. T-Spin's STD, Pokemino's STD and Early I STD are the prime examples of STD attacks. In the fumen above the left half and right half of Mr. T-Spin's STD are swapped.

Again, to achieve the goal of a Perfect Clear,we must make exactly 1 odd line clear. This is done by clearing the T-Spin Triple with an odd number of empty cells below (usually 1 hole). Pokemino's STD will also make the T-Spin Single with an odd number of empty cells below. So, in Pokemino's STD the player is forced to make an odd line clear in the last 2 or 3 lines. However, this is not necessary in almost every other STD setup.

In my last posts I spoke about how an even number of vertical T placements can help to increase the probability of a Perfect Clear. Sadly, this isn't the case for STDs. Major reason is that the left side doesn't interact with the right side at all. The 8 columns on the left are basically a rectangle with a missing I piece. No piece intersects that border, so the result that #L + #J #vertT must be an even number also applies to this area. In the first and second bag, 3 out of the 4 L and J pieces will go there as well as the first T piece. If this T piece is placed horizontally, then in the third bag either L or J piece will be placed there. If the first T piece is placed vertically, then in the third bag both L and J will be placed in that area. In this case a T-Spin Single is not possible (because you have no L left for the right side). Not only does a vertical T placement prevent T-Spin Singles, it doesn't really increase the chance to get a T-Spin Double. More Ls and Js only really increase chances when they can interact with an O piece but the O piece is needed on the right side.

In my first post I included a Fumen of Early I STD. It uses a horizontal T placement in first bag which means that in third bag either L or J are placement on the left 8 columns and both T-Spin Double (O piece on the right) and T-Spin Single (J piece on the right) are possible.



Actually, I had underestimated Early I STD. I came up with it some years ago and thought it would be useless (so that generic name, maybe it even deserves a real name). But by now, I could imagine it's quite useful against 4-wide players in games without garbage blocking. If you get a T-Spin Double in third bag followed by a Perfect Clear, then it sends at least 2 + 7 + 5 +10 = 24 lines within 3 bags which should be enough to kill the 4-wide player. With a T-Spin Single into Perfect Clear, it sends 1 or 2 lines less. Note that no other setup should be able to send more lines in just 3 bags. On the downside, the PC chance is not that high. The initial T-Spin Single into T-Spin Triple is enough to keep the 4-wide short, if you can get them within 2 bags. And that's the real disadvantage in comparison to Mr. T-Spin's STD which guarantees to get the Single Triple in the first 2 bags no matter how the second bag looks like. Plus Mr. T-Spin's STD (and Pokemino's STD) have less dependencies in the first bag.

The 3 mentioned STDs are pretty much the only STDs you can do with the first T piece placed in the bottom row and only Early I STD can really result in a Perfect Clear (unless you change how the overhang is stacked in Mr. T-Spin's STD which takes away its advantage in second bag namely that it is always guaranteed to work because of the multiple ways how the OSZ block can be stacked). However, there are some further STDs where the T piece is placed one row higher. In this case the player can also decide between a horizontal and vertical T placement.



So how do those setups perform in comparison to Early I TSD regarding Perfect Clear chances? Sadly not better as much as I can tell. I only tried those placements where it's possible to get the T-Spin Triple within the second bag and there're barely any ways to make that work. An example where it works:



And for the LOLz, 2 examples for leaving some cells empty in the bottom row (like in Hachispin). Leaving cells empty will make it significantly easier to place pieces in a way that the T-Spin Triple is ready within second bag but it will also make the last T-Spin Double / Single more troublesome. This time the first T piece is placed vertically but that doesn't help.



Summarized: Single Triple Double attacks will never have a good chance for a 8th-row Perfect Clear because the overhang is basically one row too high and thus prevents an exchange of the left stack with the right stack. That exchange is still possible in classic C-Spin though (1 row shorter) where the PC chance can be decent, if the first T piece is placed vertically (in case of STD the orientation of the first T piece doesn't really matter). Thus, getting 6 T-Spin lines can be quite a hurdle for achieving a 8th row Perfect Clear. However, it's not that problematic for 5 T-Spin lines as seen in C-Spin (with first T placed vertically) and Single-Double-Double (that Imperial Cross setup performed well, STSD is also worth researching).
User is offlinePM
Go to the top of the page
+Quote Post
pwn_by_numbers
post 3 weeks ago
Post #5


Tetris Novice
Group Icon
Posts: 43
Joined: 2-October 13



Interesting stuff. I have been thinking about 8 high PC openers as well, my thought process being that if you can attack with the first 2 bags, you could theoretically have a systematic way to place every mino in the 3rd bag for the PC.

QUOTE
If you want to have a decent Perfect Clear chance, then you want to use the third bag's T, L and J pieces (all of them together).


I'm assuming this is just so you aren't forced to hold one the pieces?

QUOTE
However, most ways are nearly impossible to spot because pieces are cut in two halves by another line clear. If there's an odd clear in the last 4 lines, then at least 1 piece is always cut in two halves. So it's better to move the odd clear forward to the T-Spin phase.


I personally don't find it much harder to find PCs that involve a line clear... Isn't avoiding this limiting your options quite a bit? (For example in the standard 4-high system there are quite a few ways to clear it that involve odd-line clears.)

---

Going back to 3-bag/8-high PCs, I was thinking they would be useful for designing a better opener for playing vs Puyo in PPT. I noticed how a lot of these methods involve TSS, which is more powerful vs Puyo. However, comboing attacks is much more important, so the methods with split up attacks might not be as effective.

Going off the garbage table on the wiki, here's some potential 3 bag openers:

4 doubles = 1,1,1,1 -> 4+4+4+4 = 16
8 combo (singles) = 7 -> 16
tsd,tsd,tsd = 3,4,4 -> 6+6+8 = 20
tsd,tst,tsd = 3,5,4, -> 6+10+8 = 24 (alb)
tsd,tsd+tsd = 3,8-> 6+20 = 26
tsd+tsd,tsd = 7,4-> 16+8 = 24
tsd+tst = 7 = 16 (DT, cspin)
tsd,tst = 3,4 -> 6+8 = 14 (broken up dt)
tsd,tsd,tetris = 3,4,5 -> 6+8+10 = 24
tsd+tsd,tetris = 7,5 -> 16+10 = 26
tsd,tsd+tetris = 3,9 -> 6+24 = 30
tss,tst,tsd = 2,5,4 -> 5+10+8 = 23
tss,tss,tss+tetris = 2,3,8 -> 5+6+20 = 31
tetris,tetris = 4,5 -> 8+10 = 18
tetris+tetris = 9 -> 24
pc,pc (4highs) = 6,6 -> 13+13 =26 (can be better if you get 2 combos for 7, could be 16+16)

Notice how bad DT cannon or C-spin are vs other openers. Also 2 PCs is barely better than 2 Tetrises. Can we do better? Imagine tss,tss,tss+tetris+single for a PC - this would send 2,3,15 for 55 puyo! Granted this isn't very practical, but it shows the power of adding a PC into a combo. Tetris into a 4 combo PC sends 4+3+6 = 13 -> 46 puyo, this is still way better than any other potential opener, and probably not too hard to setup.

For example, the following setup has a 98% PC rate with I in hold!



... Too bad it's impossible to stack. Something basic like the fumen below only gets a 76% clear rate. I can't find a good way to add extra blocks on the right hand side without taking away either the Tetris or the easy combo...



What about Tetris+TSS->PC all in one combo?



A setup like this seems really good in some ways. 100% success rate if you start with TI. Finding the PCs is really easy too. But setting up the Tetris with a TSS requires the T to come first, and it's hard to not break the combo when PCing. Still, that example sends 16+13, for 29 total, so even breaking the combo leads to a decent opener. (Note: if you could build this setup with T in hold, it has an 88% PC rate, so not terrible.)

What about TSS mini 4w -> PC?



This has a 100% PC rate with T in hold! If you can combo the whole thing it will send 43!



If you can't clear without breaking the combo it's still decent - this one sends 29:



Still working on a consistent way to actually set any of these up. Any thoughts?
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #6


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



QUOTE(pwn_by_numbers @ Mar 28 2018, 12:43 PM) *
QUOTE
If you want to have a decent Perfect Clear chance, then you want to use the third bag's T, L and J pieces (all of them together).

I'm assuming this is just so you aren't forced to hold one the pieces?

Yes, T, L and J are the pieces with 4 rotation states, so you have more options to place them as O, S, Z and I pieces. I was majorly referring to setups where you try to make a T-Spin in the third bag. In this case you have just the L and J piece to make the PC happen. For example, to fill a perfect rectangle you will always need both L and J.

QUOTE(pwn_by_numbers @ Mar 28 2018, 12:43 PM) *
QUOTE
However, most ways are nearly impossible to spot because pieces are cut in two halves by another line clear. If there's an odd clear in the last 4 lines, then at least 1 piece is always cut in two halves. So it's better to move the odd clear forward to the T-Spin phase.

I personally don't find it much harder to find PCs that involve a line clear... Isn't avoiding this limiting your options quite a bit? (For example in the standard 4-high system there are quite a few ways to clear it that involve odd-line clears.)

I think the necessary of the odd clear (if you want to use 3rd bag's T piece) is the reason why you don't see 8th-row PCs often after T-Spin openers. If you want to have an outstanding PC chance in practice (and not just on paper), then you have to make the odd clear during the first 2 bags. I am not saying that making the odd clear during the first 2 bags is always better though (I am just saying that forcing an odd clear in 3rd bag is suboptimal).

QUOTE(pwn_by_numbers @ Mar 28 2018, 12:43 PM) *
Still working on a consistent way to actually set any of these up. Any thoughts?

So you are basically trying to find a way to beat a Puyo player with the 2 things that got nerfed in the last patch: Perfect Clears and combos. I don't know if this is the right way. You try to optimize your garbage/trash output but at the same time you give the Puyo player more time. And a Puyo player profits more from having time than a Tetris player. The Tetris' player garbage output per time behaves like a quadratic function at best (or maybe cubic with a long 4-wide) whereas it's almost exponentially for a Puyo player.

I don't think you will have much success in finding a setup that starts with a Tetris or T-Spin, that continues with a combo and ends with a PC. Is the Tetris really worth the trouble? It's maybe 2 lines that you send more compared to a longer combo.

I don't really know about Tetris into PC builds, so I will just speak about classic combo into 8th-row PC. Classic combos can have a 100% PC chance. However, most often the 100% just orginate from the sheer amount of solutions that end in a PC. A 6-pieces PC will have about one thousand of different PC solutions but each solution only works for a little amount of bags. Also remember that you will most likely have to make an odd clear during the execution (unless you clear a line somewhere in the middle before you start the PC).

There's an easy way to build a 3-wide:



This 6-pieces PC has a 100% PC chance. Solution Finder fails to tell me the best solutions because it runs out of memory (too many possible ways to place the pieces). So I dropped an O piece in it in which case you still have a 100% PC chance (with the now remaining 6 pieces) and the best solutions look like this:

IPB Image

I also tried this out myself in NullpoMino Practice mode (you can define Maps there and disable certain pieces). My PC success rate was under 50% and half of the time I left the T piece on hold (this can avoid the odd clear). Moreover, I also broke the combo in almost 50% the cases in which I succeeded. 4-wide is better to keep the combo and still achieve the PC in my opinion.

The only way to build a 4-wide combo that I know is like this:



I just included the last frames because of those ridiculous S-Spin and Z-Spin Triples. I think it's better to have 5 or 9 residuals on bottom than having that Z piece overhang on top and 3 residuals. If I remember correctly, the PC chance wasn't 100% in both cases.

The rest of my post will be just of theoretical nature. I want to compare 4-wide with and without an odd clear beforehand. I consider the 2 following builds pretty much optimal for 7-high builds (I am assuming a line was already cleared because they can't be stacked without a line clear). I don't think there's an actual way to stack them. Both setups have a 100% success rate on paper.



As said before, you can try it out yourself in NullpoMino Practice Mode (with maps). In the first setup, I could achieve a PC in roughly 80% of the cases. I think I could increase it to 95% with some practice. It's pretty much straight forward: you try to place the T piece horizontally (so you can use both L and J), you try to avoid an odd clear (unless you don't want to place the T piece), or make 2 odd clears in quick succession (e.g. laying J piece flat followed by L piece). Moreover, I basically never broke the combo.

Here are 30 of the 31 solutions without any split pieces. Those solutions alone have a 66% success rate.

IPB Image

And here are 30 most frequent solutions with at least 1 split piece:

IPB Image

Now towards the second setup: Like in the 3-wide example, my success rate was under 50% and half of the time I succeeded I left the T piece on hold (in which case one can sometimes avoid the odd clear). Sometimes, it took me until 2 remaining pieces to still spot a PC opportunity. On the plus side, I basically never broke the combo.

And here are 30 most frequent solutions. Note that I removed a line and assumed a horizontal I piece placement or otherwise the solution finder would have run out of memory again. Also note the odd clear if T was placed.

IPB Image
User is offlinePM
Go to the top of the page
+Quote Post
pwn_by_numbers
post 3 weeks ago
Post #7


Tetris Novice
Group Icon
Posts: 43
Joined: 2-October 13



QUOTE(Okey_Dokey @ Mar 28 2018, 02:46 PM) *

Yes, T, L and J are the pieces with 4 rotation states, so you have more options to place them as O, S, Z and I pieces. I was majorly referring to setups where you try to make a T-Spin in the third bag. In this case you have just the L and J piece to make the PC happen. For example, to fill a perfect rectangle you will always need both L and J.


Hmm this logic makes sense, but can you back it up with any data? From messing around with sfinder, I've noticed certain patterns will have much higher success rates with certain pieces, but I haven't noticed a bias towards L and J. (I have however noticed that T seems to be favored significantly.)

QUOTE

I think the necessary of the odd clear (if you want to use 3rd bag's T piece) is the reason why you don't see 8th-row PCs often after T-Spin openers. If you want to have an outstanding PC chance in practice (and not just on paper), then you have to make the odd clear during the first 2 bags. I am not saying that making the odd clear during the first 2 bags is always better though (I am just saying that forcing an odd clear in 3rd bag is suboptimal).


I think the reason you don't see people doing 8-high PCs isn't because they can't see odd clears, it's just that no one has bothered to learn them yet. Just from my personal experience learning PC openers (pure PC opener -> multiple PCs and DT PC), it's more a matter of just doing each pattern enough times to get an intuitive sense of what works and doesn't work. Sometimes it just makes sense to do an odd line clear, and it's just as easy for me to see as any other pattern.

When you say "in practice and not just on paper" I'd argue that one should first optimize on paper and then practice all the possibilities. To get a high PC percentage, you really only need to memorize a small number of patterns - for the normal PC opener there's what, 10 or 20 to cover every possibility? (Sounds like a lot, but its combinations of 4-5 pieces, with a lot of mirroring and similarities, I learned them all in ~1hr.) In trying to come up with a system for DT PC I found that with just 3 systems, I can get >90% success from the 3rd bag.

QUOTE

So you are basically trying to find a way to beat a Puyo player with the 2 things that got nerfed in the last patch: Perfect Clears and combos. I don't know if this is the right way. You try to optimize your garbage/trash output but at the same time you give the Puyo player more time. And a Puyo player profits more from having time than a Tetris player. The Tetris' player garbage output per time behaves like a quadratic function at best (or maybe cubic with a long 4-wide) whereas it's almost exponentially for a Puyo player.


The thing is both PC and combos on their own are useless and worse than simple T-Spin or Tetris spam, but the garbage table heavily favors attacks combined in combo, in that once your attack is at 5, every additional +1 from a combo adds 3 puyo - as much as a TSD. Like I showed in my last post, an 8 combo on it's own would send just 16 puyo, but an 7 combo -> PC would send 43 (7 combo because you'll almost always end with a double). Even with the extra line clear delay, you're easily sending as much garbage as you could sending a fast opener and then an extra bag worth of pieces, so I think it'd be worth it.

I actually stopped doing DT cannon vs Puyo because it's so slow. The best I've found so far is things like albatross or TKI where I can just send a TSD/TSS or Tetris every bag to keep the pressure on. But against the best players it's not enough garbage to force them to do anything and they just build up a chain and win. Spikes are necessary to actually win vs a competent puyo. I noticed in the midgame that a T-Spin/Tetris spike -> a decent downstack combo sends a huge amount, so why couldn't I just do this from the opener? If you're not building up a giant combo there's no reason it has to be slow enough for the puyo to build a huge chain - I'm guessing there's a point where it becomes not worth it to stack up anymore, and ~3bags seems to be around where that is.

QUOTE


I don't think you will have much success in finding a setup that starts with a Tetris or T-Spin, that continues with a combo and ends with a PC. Is the Tetris really worth the trouble? It's maybe 2 lines that you send more compared to a longer combo.



You may be right here, I hadn't considered that a pure combo->PC could be as good as mixing in a T-Spin or Tetris. If my math is right, it's probably only better to mix them in if you can do both, unless there's an easy way to add a TSS to the builds you demonstrated without compromising PC chance or combo-ability. Thanks for those diagrams, I'll have to test some of these out and see how effective they are.

---

Switching gears a little bit, I think you mentioned iron cross as a potential 8-high PC earlier. Here's something I accidentally did in a game today - maybe has potential to become a good 8-high PC system?


User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #8


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



QUOTE(pwn_by_numbers @ Mar 29 2018, 10:12 AM) *
Hmm this logic makes sense, but can you back it up with any data? From messing around with sfinder, I've noticed certain patterns will have much higher success rates with certain pieces, but I haven't noticed a bias towards L and J.

No, I cannot back that up with data.

QUOTE(pwn_by_numbers @ Mar 29 2018, 10:12 AM) *

Thanks for those diagrams, I'll have to test some of these out and see how effective they are.

I think I have found a good way to stack the 7 rows-high 4-wide with 4 residuals with an odd clear beforehand. However, it should be only a 5 combo most of the times whereas it's likely that the last line clear is the Double (which doesn't send an extra line if PC). So it won't send many lines to the Puyo player. You use one L piece from first bag and one J piece from second bag to make that odd clear. With some luck that (odd) line clear itself increases the length of the combo by 1. Setup will basically look like this:



The grey area can be built like this:

IPB Image
IPB Image

Note that there are way too many possibilities to fill a 6-wide grey area, so I left out the I pieces and in setup #1 also the O pieces.

QUOTE(pwn_by_numbers @ Mar 29 2018, 10:12 AM) *

Switching gears a little bit, I think you mentioned iron cross as a potential 8-high PC earlier. Here's something I accidentally did in a game today - maybe has potential to become a good 8-high PC system?

Iron Cross must be a synonym for Imperial Cross. The nice thing is that Imperial Cross has an odd clear which is the important part for me (it's the only T-Spin setup with an odd clear except for Triple Double Attacks and second row T-Spin Singles with an odd number of holes below). I think it's best to combine it with an initial T-Spin (Single or Double). You can do so with the TKI3 Opener.

TKI Castle top variation into Imperial Cross into 8-row-PC:



The chance for a PC with the T piece placed on bottom are astounding 84 % (despite the fact that an odd number of T's was placed vertically and despite the fact that you sometimes clear 6 lines through T-Spins). I think every solution forces a T-Spin (or the placement can be made a T-Spin like in the last one). You will only need 12 of them, most later ones are redundant.

IPB Image

TKI Snail variation into Imperial Cross into 8-row-PC:



This one is not as good as the last one, majorly because the chances are lower that you can prepare the Imperial Cross with the second bag. If O piece is dropped on the wall, then success rate for the T on bottom solutions are 68 %. If O piece is placed like in the last frame of the Fumen, then chances increase to 89%. Again, I think every solution forces a T-Spin. You will only need 10 of the solutions in the later case.

IPB Image

IPB Image

Anybody else knows a setup for T-Spin Double into Imperial Cross?
User is offlinePM
Go to the top of the page
+Quote Post
pwn_by_numbers
post 3 weeks ago
Post #9


Tetris Novice
Group Icon
Posts: 43
Joined: 2-October 13



Here's an interesting possibility, if you offset the normal Albatross and Pelican builds by moving the I piece, you can setup a TST the opposite way. The interesting part is you can setup the same shape with both offset Albatross and offset Pelican, so the same PCs should work on both.



Both are odd line clears! Note O isn't place in offset Albatross and S isn't placed in offset Pelican in the above diagrams. With sfinder, both setups give a PC rate of 85%, which lowers to 81% if you ignore builds that clear the TST shape without a T.

If you place your 2nd bag T on the left side and clear the TST with the 3rd bag, unfortunately the PC rate drops a lot - somewhere around 30%. However if you get a lucky enough 3rd bag, it's possible to get the TST instead of a mini-TSD and still have good PC chances.

This has an 85.83% clear rate with S in 2nd bag not yet placed, 78.89% with O:



I'm curious, do you think TSD, TSD, 85% PC is "good enough"? What clear rates should we be aiming for here?
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #10


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



I also thought about about dropping the I piece on the other side in Pelican but then I saw your Wiki edit with that nice Albatross-like 10-lines PC continuation. Somehow, I never had that idea for Albatross (dropping I piece on other side).

QUOTE(pwn_by_numbers @ Mar 30 2018, 12:12 AM) *
The interesting part is you can setup the same shape with both offset Albatross and offset Pelican, so the same PCs should work on both.

It's the same shape but one time you haven't placed O and the other time you haven't placed S. So it's not the same situation and the solutions will look differently.

QUOTE(pwn_by_numbers @ Mar 30 2018, 12:12 AM) *
Both are odd line clears!

Those are not odd clears by my definition. Sorry, I hadn't fully explained what I meant with odd clear. For each cleared line, look how many empty cells are below that line, and sum up those numbers (assuming you cleared multiple lines at once). If the result is odd, then it's an odd clear. So for an odd clear you will need an odd number of "odd" cleared lines.

In first bag of Albatross and Pelican we have 5 empty cells below the 2nd row (first cleared line) and 5 empty cells below the 3rd row (second cleared line). 5 + 5 = 10 is an even number. Note that you will never get an odd clear with a T-Spin Double into horizontal position. Same is true for a Double with an O piece or Tetris with an I piece. In second bag of Albatross and Pelican we clear the bottom 3 lines. Those lines have no empty cells below, so it isn't an odd clear either.

The T-Spin Triple can become a T-Spin Double with an odd clear if you don't fill the 2nd row completely. This is impractical because you will clear less lines and still have the overhang (except if Imperial Cross). I also don't see how this is applicable to any of those single bag platform T-Spin Double setups.



Regarding those 80%: that's a similar situation to MKO stacking (see my first post). Note that you still have 7 pieces to place instead of 6 and each remaining piece increases the number of different path. But I find it more difficult to see each path. For example in case of that Albatross variant you will have 35 different solutions. Note the odd clear in every solution with a placed T (odd clear is illustrated by a single mino that is seperated from the rest of the piece by an odd number of lines).

IPB Image

QUOTE(pwn_by_numbers @ Mar 30 2018, 12:12 AM) *
What clear rates should we be aiming for here?

This depends on how many pieces are still to be placed, how many lines are still to be cleared and if you want to make a T-Spin in last bag or not.

In case of an 8th row PC with one bag still to place (and no odd clear to go):

I would say 80 % or more if you don't want to make a T-Spin in last bag. And at least 50 % if you want to make a T-Spin.
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #11


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



So far I had only taken a look at T-Spin setups where the player tries to make another T-Spin in the third bag. In this case it's best to make one odd clear within the first 2 bags (via 2nd row T-Spin Single, C-Spin or Imperial Cross), and if you cleared only 3 lines during the first 2 bags, then those T placements should be placed in the same way (either both horizontal or both vertical).

However, if you just want to start with 2 T-Spins and use the 4 remaining lines to make a PC, then setups without an odd clear (usually 2 x T-Spin Doubles) also perform decently. For example, this TKI build has a 99.9 % success rate on paper:



IPB Image

I would consider those PC chances in practice at least on par with the one at Hachispin pwn_by_numbers variation (which uses an odd clear by contrast). In the first 2 solutions, pieces are not split apart. This is only possible because the T piece was not placed. Those 2 easy-to-spot solutions will cover 50 % of all bags. Most other solutions are also pretty easy to spot. Only the first 15 solutions are needed in theory.

However, you will barely ever be able to stack the first 2 bags like this. I also tried TKI to Imperial Cross and that also barely ever worked.

So, just 2 T-Spin Doubles into 8th-row Perfect Clear can also work out, provided there are enough no-split solutions. However, I am personally only interested in solutions where you still try to T-Spin in the last bag (because PC might fail or opponent sends lines in mean time, you also lose the b2b bonus and I personally must slow down a lot to process the next piece queue).
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #12


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



Somebody going by the name Riviclia on Puyo Puyo Tetris invented a new opener. I am mentioning it in this thread because there's a PC chance after 8 cleared lines in total. Video & Fumen:





It's a very decent setup besides the fact that the chances are rather slim to stack the first bag like this. There're holes after placing the first bag but it can be fixed with a J- or L-Spin.

If you can place the first 2 bags like this, then there's a 90 % Perfect Clear probability. There's one PC solution that can result in a T-Spin Double (picture 4) and two further solutions that can result in a T-Spin Single (pictures 1 & 2). Those 3 solutions have a probability of 69 % together, although I guess it will only be a T-Spin in half of the times.

IPB Image

Note that the second T-Spin Double is an odd clear. Since the first 2 T pieces were placed vertically, a T-Spin Double can only occur in third bag, if you use both L and J. And a T-Spin Single into vertical position can only occur, if you keep either L or J on hold.
User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post 3 weeks ago
Post #13


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



This post is about exploring further options for the "Number One" setup mentioned in last post. As a reminder first bag is stacked like this:



J piece can be spun twice into the following position. This enables pretty much the same continuation as in the build from Riviclia but with a slightly different right side:



PC success rate will be 98 % in this case, with 2 options for a T-Spin Double (pictures 1 & 2) and a few further options for a T-Spin Single (among others pictures 3 & 7).

IPB Image

There are also ways to stack Imperial Cross setups:

If you spin L piece, then you can stack an Imperial Cross like this. PC success rate is about 80 % with the options for a T-Spin Single like shown in the last frames of this Fumen:



If you spin Z piece, then you can stack an Imperial Cross like this. PC success rate is again about 80 % with an option for a T-Spin Double as well as 4 further options for a T-Spin Single like shown in the last frames of this Fumen:

User is offlinePM
Go to the top of the page
+Quote Post
pwn_by_numbers
post A week ago
Post #14


Tetris Novice
Group Icon
Posts: 43
Joined: 2-October 13



Here's an opener I've been having fun with. In lieu of a better name, I'm going to call this Pwn's STPC. Basically, it's a Fin TSS -> TST -> PC. It works with almost every bag and has an exceptionally high PC-rate.

Overview:



This variation has a 90.44% chance for a PC in the 3rd bag, with an added bonus that the PC variations are pretty easy to spot.

First bag:

Choose the mirror based on whether L or J comes first, as you would in Albatross. The only bags that don't work are when both O and T come before I (and some of those can even be made to work as I'll show later on).

Normal build order:


Note that you can do the Fin TSS without the O piece, but soft-dropping to the right height is a lot harder.

If S/Z comes early you can slide the L/J in:



And if S/Z come in the wrong order you can kick the middle piece into place:



If you get an early O, but can place both L and J, you can use this variation:



Second bag:

The best variation is the one I showed in the overview, with an early S/Z, then O before L/J. Some of the other variations will force you to go for a TSD instead of TST if you still want to go for the PC.

Early Z, L before O:



If the I piece comes too late in the second bag, you should place the O on top (instead of the O before L build) so you can do this and spin it in:



(Note: I placed the L piece that way so I'd have a place to put the O piece for the PC.)

PC rates:



90.44%


82.18%


28.85% (with O first 53.89%)

Bonus variation:

If you delay the O from the first bag, you can go for a DT instead of an ST. In practice I find it rare that I get a DT and am still able to get the PC, so I often go for DD like in this example:



Note: In PPT vs Puyo, ST sends 1 more nuisance puyo than DD.

For those late I piece 2nd bags you can also try something like this. Not as good as an ST vs Tetris, but vs Puyo it is very good, especially if you can combo the Tetris into the PC. This example sends 43 puyo! (Normal STPC sends 28, and if just TSS, Tetris, PC without comboing it also sends 28.)



In practice I find it difficult to setup the Tetris so that I can combo it into the PC - I'm not sure how to use solution-finder to check the probabilities here (since I only want to count ones that can be cleared with Tetris), but if this could be consistent I think it has the potential to be one of the best openers vs Puyo. A quick TSS to get B2B followed by Tetris->PC in 3 bags is about as good as it gets.

Bailing out:

If for some reason you can't PC or you get sent lines you can do something like this: (vs puyo it might even be better to forgo the PC and go for this anyways if you want to spend one more bag's worth of time)



Or you could try grim grotto:



Edit:

Here's a video I made showing a few different attempts (with a DT version at the end):

User is offlinePM
Go to the top of the page
+Quote Post
Okey_Dokey
post A week ago
Post #15


Tetris Expert
Group Icon
Posts: 467
Joined: 13-May 15



Oh, so by now people even try to invent openers with Fin T-Spins. Well, it's better than those Fin / Neo / Iso T-Spin setups you find on the Tetris Wiki for sure.

Here's a Single-Triple-Double with a small PC chance after 8 lines and where the T-Spin Single is a Fin T-Spin.

User is offlinePM
Go to the top of the page
+Quote Post

2 Pages V  1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 

©2009-2018 Hard Drop Community & Forum
harddrop.com is not sponsored or endorsed by The Tetris Company or its subsidiaries.