HOME WORLD FORUMS WIKI VIDEOS
9580 members and stacking!
     Welcome guest, please login or sign up

 
Reply to this topicStart new topic
> Insights into perfect clearing
myndzi
post Oct 23 2013, 05:47 PM
Post #1


Tetris Grand Master
Group Icon
Posts: 1,932
Joined: 26-June 09



:::WARNING:::

words ahead

:::WARNING:::


If you're still reading, you probably have an idea what you're in for, so let's get down to business.

Some months ago I had an idea about perfect clear stacking. I mulled it over and talked to a few people about it, but it wasn't immediately apparent how it could be put to any practical use. Well, I spent about 4 hours this morning when I couldn't sleep stacking perfect clears, and I think the relevant bits finally sunk in. So, I'm gonna tell you about them Smile.png Before I do that, though, I'd like to cover a few related concepts.

Stacking is the art of not screwing up
Most of us who've played Tetris for any length of time have developed a feel for when things are bad and when they're good. We've identified a number of principles to guide us in our decision making which, in combination with our instincts, serves to help us stack neatly. In general, our goal is to play while avoiding the pitfalls we've learned about. Sometimes this is phrased in the form of things you should "do", as with connecting surfaces, but that can also be states as "don't create a rough surface." We juggle all of the concepts we have learned while playing, trying to find the placements that best fit all the things we want at some specific moment.

More tools in your toolbox
The more skills you've developed as a player, the easier it is for you to avoid pitfalls. Whether it's platforming, spins, downstacking, upstacking, setups, or whatever, the more you know, the more you've got to work with.

Parity
I am going to make a distinction here. What we've talked about as 'parity' (T-piece parity) I will refer to as 'grid parity' -- the idea that the grid of Tetris pieces is not in balance. Row parity refers to whether a row contains either an even or odd number of minos. There's also column parity, which applies to perfect clear stacking but not so much else.

-

Okay. All these things above apply to stacking in general, but they apply even more to stacking for perfect clears. Since you are stacking within a limited space, there are many more mistakes to make -- each of which can make it impossible to complete your goal. Here are a few principles I've used to aid in perfect clearing:

The box
The box is the area you are trying to fill. It is important to know when it is impossible to fill the box, because the sooner you know that, the sooner you can extend it -- and the sooner you can extend it, the easier it is to fill. Typically you want to aim for 10x4 and extend to 10x6. You want to go right up to the edges, don't shy away from them, and fill outside in.

Empty spaces
If the count any group of connected empty spaces within your box is not divisible by 4, you cannot complete a perfect clear within those boundaries. This rules out quite a number of placements; if you stack a piece through an empty space and cut it in two, both remaining spaces must meet this principle still.

Edit: As UJS3 pointed out, this is not entirely true -- when separate spaces can become connected again, as in a forecast play of some kind.

Forcing shapes
Visualizing the boundaries of the box, you will frequently find that there are places that can only be filled neatly by a specific piece or pieces. Sometimes the required piece will itself create a shape that can only be filled by a different piece. As you add pieces, this will become more likely until the shapes that you must use are readily discernible -- and then you'll know if you can succeed or not.

Early extension
If at any point you discern that you're unable to complete a perfect clear, you should immediately increase the size of your box by as little as possible. Remember that this will go in two-row increments. Usually, if you've stacked a large portion of a 10x4 box, it will be hard to fill in a 10x6 box because you'll have wide stretches that require pieces almost as restrictive as a 10x2 box. You should immediately prioritize placing O, L, J, and I pieces along the top to fill this space. If you see that you cannot (usually by the principle of forcing shapes), then your best bet is 10x8.

Bag balance
In the bag randomizer, you will receive a balanced mix of pieces. This means both that you can count on some pieces and also that you can count on not receiving some pieces. If you create a shape that can only be filled by a specific piece, you had better be sure that you will receive that piece at an appropriate moment.

Bag balance and forcing shapes work together in a significant way. For example, if you place an L that creates a shape that can only be filled by another L, that's not going to work much of the time. Similarly, stacking too shallow may leave you with a requirement for two flat I pieces on the top of either side of your matrix. O and T pieces are common culprits, they both have many arrangements that require either awkward stacking or another O/T piece.

Grouping
It is mentally convenient to visualize chunks of the box being filled. This can be either a sequence of forcing shapes or a common pattern you've learned. In combination with bag balance, you can mentally "check off" pieces from your previews as a means of determining which pieces you still have available.

Order
Grouping, bag balance, and forcing shapes all work together to do a large chunk of the work for you, however the order of the pieces is obviously important too. If you've found a potential solution, be sure that the required pieces are going to arrive in an order you can execute!

-

For the most part, the principles above are what I've used when stacking perfect clears. I've learned by rote memory certain patterns that are easier to complete than other alternatives. You usually don't know the next 11 pieces before you place your first piece, however, and so there's been guesswork involved.

Which pieces do you stack first, and where? What will give you the highest chance of completing a perfect clear with the remaining pieces when you know them? If you've tried perfect clearing for any significant amount of time, you've probably gotten a feel for when a PC is going to work or not. How soon can you determine this? I believe I can help answer these questions objectively.

A standard Tetris matrix is 10 wide. Standard pieces are composed 4 minos (blocks). Necessarily, then, any perfect clear must be accomplished in a multiple of two rows. You would have to use 2 and a half pieces to fill only a single row. More generally, both the x and the y dimensions of the box you are trying to fill are known to be even numbers.

Even numbers wide, even numbers tall, and all my blocks have even numbers of minos... should be easy, right? Well, there's a problem. Because of the unique shapes of each piece, they have differing effects on the row and column parity of your matrix when you use them! If you could stack everything in nice square blocks, perfect clearing would be easy... but you have a little more to deal with than that.

Like sudoku, you have to place each piece carefully, with awareness of how it affects all the important factors. In sudoku you require one and only one digit for each row, column, and diagonal. In Tetris, you require even column parity and even row parity over the entire area you are trying to fill. The principles mentioned above help you to rule out many placements, but none of them address this most important thing directly!

Consider the S piece. In its spawn orientation, the top row contains 2 minos and the bottom row contains 2 minos. When placed, this piece will not alter the parity of the rows it occupies. (Odd + even = odd; even + even = even). If rotated, however, the rows become 1-2-1; the parity of two rows will be reversed. (Odd + odd = even; even + odd = odd). Same for the Z piece.

The O piece never alters parity. The I piece, flat, won't alter parity but it will frequently cause forcing shapes in this orientation. The I piece, vertical, will reverse the parity of every row in a 10x4 area.

The J and L pieces will be either 1-3/3-1, or 1-1-2/2-1-1. Note that no matter which way you put them, the two odd rows will be consecutive.

The T piece, finally, is the oddball. In spawn orientation it is 1-3, and when rotated 90 degrees it is 1-2-1. This gives it the characteristics of both the S/Z and the J/L when it comes to altering row parity.

(Excised)

I need to study grid vs row parity more. I no longer think that row parity supersedes grid parity. They refer to different properties, but I can't think of a decent name for the concept I am discussing that isn't "parity." Whether or not row parity supersedes grid parity, though, we can still use the concept to our advantage and derive a couple of useful principles from it.

Principle of row parity
We have a 10x4 box that we want to fill. If any row parity is odd, we will be unable to succeed; we must end with all of the rows (and columns) having even parity. Since we know how various pieces are able to affect the parity, we can therefore rule out quite a few placements because no pieces can correct the parity pattern, or the pieces which could correct it won't fit, or the pieces which could correct it and fit violate another principle, such as the principle of empty spaces.

By keeping this in mind, we can look at our next previews and see far ahead what we're going to have to deal with. If we get L's or J's we know that we'll need to offset odd-odd pairs. L and J always alter row parity. If we get Z's or S's, we know that we may need to offset odd-even-odd patterns. This can lead to a need to skim the line between the odd parity sections so that an L, J, or T can then counter the now-adjacent odd parity rows. A T, of course, is useful and flexible in that it can take the parity characteristics of either of these four pieces. An O piece, on the other hand, is completely neutral - it doesn't affect anything parity-wise, but it does get in the way physically, restricting what can be stacked around it and therefore has a secondhand influence on parity.

If we get an I piece, we know that we'll need to make everything odd or find a way to place it flat without breaking any of our other principles.

Principle of neutral placement
A neutral placement will either not affect parity, or affect it in a way that can be immediately corrected. It is best to begin a perfect clear setup with neutral placements, so that you can see as many previews as possible before committing to a specific parity pattern. In the standard perfect clear setup, the parity you wind up with is alternating even and odd because of the T-piece placement. Vertical S and Z shapes violate the empty spaces principle, therefore you must either skim the even line between the two odd lines and use a J/L/flat T piece to correct the odd parts, or use a vertical T piece to directly counter it. If you add in a vertical I piece, the row you must skim changes.

-

Neutral placements can help you abort a perfect clear early when you see that the upcoming pieces won't work out, and it can help you decide what to do first at the same time. Once you can see all your previews, you can make certain assumptions about how you must place a given piece.

Awareness of row parity can also lead you to more exotic perfect clear solutions, such as situations where you must create or wind up with something strange like odd-even-even-odd. Or avoid such a situation.

Awareness of row parity can also help you to stack pieces in two separate groups on opposite sides of the field confidently; if you maintain the parity that you will need to complete the perfect clear, you won't fill in 5 more pieces to find that nothing seems to work and it's a big mess.

-

Miscellaneous thoughts:

T, Z, and S cannot be used to fill a 2-high area; they force a violation of the principle of empty spaces. This is useful in ruling out a lot of placements in combination with forcing shapes.

The O piece needs to be wrapped into a block in most cases. Make it a priority.

The T piece often determines the orientation of the later 'corrective' pieces. Put off positioning the T for as long as possible (this usually means avoiding S/Z too), and pay attention to the orientation of your final pieces when deciding how to use it.

Using an S without a Z is dangerous, but there are some convenient shapes for using up two Z's and an S, for example:



You can increase your chances of a 2-line perfect clear by attempting to complete perfect clears with T/Z/S pieces. In the standard setup, that may look like this:



That shape can also be useful to dispose of O pieces if you're willing to give up an I. If you can abort the standard setup for an alternate early enough, you can use up your S/Z pieces when you might not otherwise have been able to.

A couple other random things:



Another thread discussing perfect clears in depth

-

To go back to the beginning of this post, stacking is the art of not screwing up. Perfect clear stacking is the same, but taken to the extreme. It's easier to screw up a perfect clear than a general game of Tetris, and harder or impossible to recover from. Hopefully I've provided you with another tool to make this process a little easier.


--------------------
User is offlinePM
Go to the top of the page
+Quote Post
UJS3
post Oct 23 2013, 06:42 PM
Post #2


Tetris Apprentice
Group Icon
Posts: 147
Joined: 21-November 12



Cool, I'll have to take some more time to think about row and column parity.

QUOTE(myndzi @ Oct 23 2013, 05:47 PM) *

If the count any group of connected empty spaces within your box is not divisible by 4, you cannot complete a perfect clear within those boundaries.

I know what you mean, but the wording isn't quite right. There are counterexamples where line clears connect empty spaces so that they become multiples of 4:
User is offlinePM
Go to the top of the page
+Quote Post
Shuey
post Oct 23 2013, 10:34 PM
Post #3


Tetris Expert
Group Icon
Posts: 464
Joined: 12-November 10



Great info myndzi! This should be directly associated with the Perfect Clear study (here!), and I'd love to see people get the study going again and take it further than we went before Smile.png.


--------------------
Subscribe to me on YouTube
User is offlinePM
Go to the top of the page
+Quote Post
myndzi
post Oct 23 2013, 10:45 PM
Post #4


Tetris Grand Master
Group Icon
Posts: 1,932
Joined: 26-June 09



QUOTE(UJS3 @ Oct 23 2013, 06:42 PM) *

Cool, I'll have to take some more time to think about row and column parity.
I know what you mean, but the wording isn't quite right. There are counterexamples where line clears connect empty spaces so that they become multiples of 4:




Good call. Not sure how to word that succinctly, for now I'll just edit in a caveat or something.

QUOTE(Shuey @ Oct 23 2013, 10:34 PM) *

Great info myndzi! This should be directly associated with the Perfect Clear study (here!), and I'd love to see people get the study going again and take it further than we went before Smile.png.


It was a bit too long to just post in that thread, and I wanted to cover some of the basics I've developed so that I don't have to keep explaining them when I talk to people Naughty.png


--------------------
User is offlinePM
Go to the top of the page
+Quote Post
myndzi
post Oct 24 2013, 09:37 PM
Post #5


Tetris Grand Master
Group Icon
Posts: 1,932
Joined: 26-June 09



I may reverse my opinion on whether you should make neutral placements first Sticking Out Tongue.png That part (pretty much only that part) was more speculation than experience, but it seems to me that no matter what, you're going to have to break parity... so if you use the pieces that force you to break it first, you may have more options for a solution later. This after I just got done examining an LOJI start into something like ZSTOSLZ


--------------------
User is offlinePM
Go to the top of the page
+Quote Post
zunixaani
post Feb 20 2015, 12:36 PM
Post #6


Tetris Novice
Group Icon
Posts: 1
Joined: 20-February 15



TF is definitely a good start. One thing I would recommend doing is going through all the missions. They'll force you to think differently about Tetris than you've been previously used to. I think you'll gain good "new" experience that way. I would also recommend seeing if there are some people on HD who you can play against in Nullpomino netplay; particularly people who have matching skills so you can gain experience from playing while also helping them grow.
__________________
Sign up with Testking ccnp routing and switching pdf ccna for getting incredible online testking courses a+ 220-801notes and lynn We also provide best
User is offlinePM
Go to the top of the page
+Quote Post
morningpee
post Feb 28 2015, 08:38 PM
Post #7


Tetris Expert
Group Icon
Posts: 319
Joined: 1-October 13



QUOTE(myndzi @ Oct 23 2013, 05:47 PM) *

For the most part, the principles above are what I've used when stacking perfect clears. I've learned by rote memory certain patterns that are easier to complete than other alternatives.

Just to emphasize this: Once you have learned the necessary twists and other techniques, perfect clears are all about memorization. Here is a video of me doing perfect clears using invisible blocks, which arguably requires less thinking because it doesn't involve trying to calculate a solution from what you see. Simple memorization.


User is offlinePM
Go to the top of the page
+Quote Post
caffeine
post Feb 28 2015, 09:38 PM
Post #8


Tetris Grand Master
Group Icon
Posts: 1,752
Joined: 27-June 09



Wow
User is offlinePM
Go to the top of the page
+Quote Post
Rosti_LFC
post Mar 1 2015, 12:28 AM
Post #9


Tetris Grand Master
Group Icon
Posts: 1,237
Joined: 26-June 09



That's a really impressive video.

I find the invisible stuff fairly intriguing - under the right conditions it's incredibly punishing and difficult to play. But I feel that video and the recent MasterK by Question_Mark in Phantom Mania (video) go to show just how broken and solved small-bag randomisers are from a puzzle perspective. It's easy to detail it in long wordy posts and fumens, but the fact that it makes even invisible play demonstrably easy (or if not quite "easy", then "more about rote memorising patterns than figuring stuff out on the fly") is the real kicker imo Sticking Out Tongue.png
User is offlinePM
Go to the top of the page
+Quote Post
Shuey
post Mar 1 2015, 02:27 AM
Post #10


Tetris Expert
Group Icon
Posts: 464
Joined: 12-November 10



I'm always amazed at how quickly people jump to talk about how broken bag is, but yet there's only one person in the world so far that I know of who's completely dominated bag in Phantom Mania...


--------------------
Subscribe to me on YouTube
User is offlinePM
Go to the top of the page
+Quote Post
Rosti_LFC
post Mar 1 2015, 10:19 AM
Post #11


Tetris Grand Master
Group Icon
Posts: 1,237
Joined: 26-June 09



QUOTE(Shuey @ Mar 1 2015, 02:27 AM) *

I'm always amazed at how quickly people jump to talk about how broken bag is, but yet there's only one person in the world so far that I know of who's completely dominated bag in Phantom Mania...

Sure, but that doesn't really mean anything. For one, the whole playing forever aspect is a theoretical side of Tetris that a lot of players don't really dabble in, and for two, actually executing it requires quite a reasonable amount of dry memorisation that isn't really found in many other aspects of Tetris and probably doesn't interest players. And for three, most of the people who play games with a bag randomiser don't go for modes like Phantom Mania, and most of the people who do train invis tend to play with different randomisers.

I'm not trivialising it - it's definitely an achievement (especially given that Phantom Mania is hardly kind with the speed requirements), but the fact that hardly anyone has done it doesn't mean it's challenging or difficult if hardly anyone has actually attempted it. It's the same reason there are so few people who have cleared TGM+ mode on TAP, or have World Gm on Ti - it's difficult enough to require quite a bit of work towards it and need multiple attempts, and very, very few people have actually bothered to do that.

My understanding of Q_M's Phantom Mania videos (and morningpee's above to a lesser extent) is that they're to demonstrate how powerful his playing forever techniques are, and that 7-bag is so broken/solved that you don't even need to be able to see the pieces to play indefinitely. It's comparable to solving a rubik's cube by figuring it out, and by learning and memorising the correct sequences and algorithms to do it. Fundamentally they're different approaches, and once you've done the groundwork to memorise the algorithms it makes solving it straightforward, as opposed to incredibly difficult.
User is offlinePM
Go to the top of the page
+Quote Post
morningpee
post Mar 1 2015, 09:05 PM
Post #12


Tetris Expert
Group Icon
Posts: 319
Joined: 1-October 13



QUOTE(Rosti_LFC @ Mar 1 2015, 12:28 AM) *

It's easy to detail it in long wordy posts and fumens, but the fact that it makes even invisible play demonstrably easy (or if not quite "easy", then "more about rote memorising patterns than figuring stuff out on the fly") is the real kicker imo Sticking Out Tongue.png

Unless you're doing a metagame thing like playing forever, I would say that using a bag randomizer is only super helpful for about the first two bags. I would imagine it's pretty common that people can place a TKI using invisible pieces. But bag players can still be very skilled, and that skill does carry over to memoryless play.

For example, I was able to apply the same principles that I used in my first video but with a memoryless randomizer. It was more difficult but still possible. Video:




I think that all of Tetris is about memorizing the right things, not just bag play.
User is offlinePM
Go to the top of the page
+Quote Post
Rosti_LFC
post Mar 1 2015, 11:53 PM
Post #13


Tetris Grand Master
Group Icon
Posts: 1,237
Joined: 26-June 09



Nice, that's a really impressive video! The difficulty is clearly a lot higher (it takes you twice as long to get fewer PC's) but that's fair enough given the fairly gnarly sequences you get at times.

And point taken, though my original comment was more aimed at the playing forever Phantom Mania stuff, with just the general point that playing invisible adds an extra element to the demonstration of how some theoretical simplification can make a huge difference.

I don't entirely feel that Tetris is about memorisation, or at least not the conventional sense of memorisation. There's definitely a lot of pattern recognition and muscle memory involved, but I feel for the majority of Tetris games the skill is in decision making on the fly, and that decision making largely comes from experience (which is a more subconscious form of memory) rather than someone bulk-memorising the "best" move in any given situation for any given piece.

Given enough previews and a predictable enough randomiser, I think it's possible to go towards pure memorisation (as with the Playing Forever stuff), and there are obviously little patterns in things like T-spins that are basically memorisation, but in general I don't think even the top players play that way. And people who program Tetris AIs always seem to go down the path of logical decision-making to get the best results, rather than just a massive look up table - the AI equivalent of memorising the best moves every time.
User is offlinePM
Go to the top of the page
+Quote Post
Question_Mark
post Mar 2 2015, 09:22 AM
Post #14


Tetris Apprentice
Group Icon
Posts: 108
Joined: 14-January 11



Excellent job, morningpee! That really is impressive. You should give it a go in Phantom Mania - see if you can stack PCs in 20G. Grin.png It puts a lot of emphasis on visualising the stack since you need to be aware of floor kicks and when certain moves are possible.

Now... this whole Phantom Mania / Breaking Bag debacle is getting to be a problem. Courtesy spoilers here; I don't want to hijack this thread - I honestly think it deserves its own - so I'm making this all one post.
QUOTE(Rosti_LFC @ Mar 1 2015, 12:28 AM) *

I find the invisible stuff fairly intriguing - under the right conditions it's incredibly punishing and difficult to play. But I feel that video and the recent MasterK by Question_Mark in Phantom Mania (video) go to show just how broken and solved small-bag randomisers are from a puzzle perspective. It's easy to detail it in long wordy posts and fumens, but the fact that it makes even invisible play demonstrably easy (or if not quite "easy", then "more about rote memorising patterns than figuring stuff out on the fly") is the real kicker imo Sticking Out Tongue.png

Sorry, I could be misreading this, but are you insinuating I obfuscated PF-V3.1 in the post unveiling it? If so that makes a lot more sense, as you also suggest below that Playing Forever is "simple" and straightforward. Let me clear that right up - the 20-bag loop was; the 5-bag loop is anything but. You don't see all the crazy cases I tackle in my solution, but rest assured they come up in practice when you use it to clear tens of thousands of lines.

Those many fumens were a lot of work - granted, it's not like anyone could've helped me - but they're not there to obfuscate either. The whole point was to make it clearer, preferably as clear as possible. Did I succeed? Probably not, because to my knowledge only Shuey has learned it, though maybe they're just not interested rather than confused. But only the two of us known to use it... hmm, maybe "broken" doesn't capture that so well.


QUOTE(Rosti_LFC @ Mar 1 2015, 10:19 AM) *

QUOTE(Shuey @ Mar 1 2015, 02:27 AM) *

I'm always amazed at how quickly people jump to talk about how broken bag is, but yet there's only one person in the world so far that I know of who's completely dominated bag in Phantom Mania...

Sure, but that doesn't really mean anything. For one, the whole playing forever aspect is a theoretical side of Tetris that a lot of players don't really dabble in, and for two, actually executing it requires quite a reasonable amount of dry memorisation that isn't really found in many other aspects of Tetris and probably doesn't interest players. And for three, most of the people who play games with a bag randomiser don't go for modes like Phantom Mania, and most of the people who do train invis tend to play with different randomisers.
[...]
My understanding of Q_M's Phantom Mania videos (and morningpee's above to a lesser extent) is that they're to demonstrate how powerful his playing forever techniques are, and that 7-bag is so broken/solved that you don't even need to be able to see the pieces to play indefinitely. It's comparable to solving a rubik's cube by figuring it out, and by learning and memorising the correct sequences and algorithms to do it. Fundamentally they're different approaches, and once you've done the groundwork to memorise the algorithms it makes solving it straightforward, as opposed to incredibly difficult.

A very good point. Yes, me being the first to do something in Phantom is not very significant. I knew that from the beginning; for starters, there aren't many sadistic enough to play it. And for some reason, there's always been an aversion to the bag randomiser before I ""broke"" it. So, for the most part, people who play bag don't play invis and vice versa. I'm in a thin intersection because I play bag, a lot, and I watch TGM3 players like KAN pwn invisible Tetris.

But more importantly, I really, really need to address this line in red here. It suggests that if you know this method well enough, you can execute it without seeing the pieces, and play indefinitely - which is by absolutely no means true! Executing this method in 20G depends on knowing the state of most or all of the contour of the playing field while placing every piece. You can't pre-memorise a bunch of patterns and then just expect you can build them. You'll try to put an O half-supported on a J, not realise you've slid off it, and now you're stuck because you can't put it back at 20G. Or put a J and L upside-down and have no way of getting the O in. And (my favourite) you'll MD because you forgot the sideways offset during floor kicks, which aren't a thing in ARS but a real pain in the ARSe when playing Guideline. L upside-down on the sheer edge of your stack? Easy, slide it and rota - oh, pants, it's fallen into the well. Stupid floor-kick.

Now, somehow you're able to place J, L, O in a rectangle without memorising anything. What was the state of the STZ heap? What pieces does it need to be completed? And, the best part, which placements did you use? Remember (as many critics forget) that PF-V3.1 is not a static setup - everything is situational. You know this method really well, so you know what shape that substack needs to form. But what does it look like now?

Hopefully you get the idea. Maybe knowing the patterns is enough to understand the PF-V3.1 method, but definitely not to execute the method - you really need to know how the playfield looks. In other game modes that's done by seeing, but obviously this is invisible Tetris. And obviously, you will screw up the execution of any 20G invisible Tetris unless you memorise. Frankly, you'd have to be silly to expect you can play invisible Tetris without memorising and/or visualising your stack.

That skill is very much disjoint from the skill of knowing PF-V3.1. You simply don't become awesome at memorising PF-V3.1 intermediate states just by learning the method or memorising the patterns. I very consciously use snapshot memorisation while playing - the kind you get when you quickly shut your eyes and things are still visible for a split second. The lock flash frame I use in my current Rule makes the snapshot easier to cling to for the 0.5 seconds I need it. I don't know the mechanics of how the TGM3 GMs do it, much less KAN in his crazy manual locking, but that's how I do it. I'm visualising the playfield using short-term memory while I stack PF-V3.1 with long-term memory. If anything, I'm using memory more than most decent invisible players. And it works.

Well, except when there's an error between the chair and the keyboard. (That's me, by the way. Hello.) In invisible play, people usually mess up. It's pretty bad for a run and few people can recover from an error so they just stack over it rather than trying to fix invisible holes in their stack. I have no such luxury - playing PF-V3.1 means pieces need to be perfectly placed. One error means it's all over. Why? Because I mess up around 300-500 when the game is already pretty fast (I imagine you're familiar with T.A. Death), I panic and misdrop, and that mistake just piles up. A messy stack is very hard to visualise. That being said, I've given up on being a machine and so many of my newer runs feature mistakes and "recoveries," which are where I restore the playfield to a state where I can continue with PF-V3.1. Can't very well do that if I don't have the playfield memorised, now.

You might think I'm faking it. That I'm pretending to memorise when I'm really just relying on patterns. Well, I invite anyone to give bag Phantom a try. PF-V3.1 is hard - start easier. Try Playing Forever, the old 20-bag loop. (Phantom is 20G, so you need to switch the middle and right substacks to make it work.) Can you break 100? If you don't care, just give it a half-decent attempt and you'll have to agree, knowing the method is not enough (but is necessary), you need memory. Even if you can pull it off, I doubt you'll be saying it's easy when you're the other side of 100.


QUOTE(Rosti_LFC @ Mar 1 2015, 11:53 PM) *

[...] my original comment was more aimed at the playing forever Phantom Mania stuff, with just the general point that playing invisible adds an extra element to the demonstration of how some theoretical simplification can make a huge difference.


Again, not entirely sure what you mean by this. Is this in reference to your claims that the Playing Forever approach to bag Phantom is straightforward, easy, and/or built on rote memorisation instead of incredibly difficult? I promise you - it's neither straightforward nor easy (much to my disappointment). It's built on rote memorisation in addition to on-the-fly memory/visualisation needed for a Phantom Mania execution, and remains incredibly difficult, as invisible Tetris is and should always be. You're significantly overstating the advantage I get from knowing my own method - I still have to make decisions for where those pieces go, which depend on what piece is active, what my playfield looks like, and what my next pieces are. That sounds a lot like all of Tetris to me. The difference is I've rigourously worked out how to PC in 35 pieces, and that's fun to do in Phantom Mania. I make it look easy by playing very fast - not because it is easy.

I think that Tetris is Tetris. How you choose to play it is your choice. And I chose to get very good at getting 10+ perfect clears in invisible instant-gravity Tetris because that's bloody fun. I hate to bring this up now, having plugged my PF-V3.1 method so much. But I don't need to play invisible this way. You, like many others, seem to think I use this as a crutch to get me through Phantom Mania, but I chose to play this way. GM requires a bunch of Tetrises, which (if you glanced at my weeks of work) you may notice PF-V3.1 does not give you at all. If I wanted an easy way, for example, I'd just Play Forever the way people used to do, wouldn't I? It's a guaranteed loop in 20 bags and not my open-ended WCS stuff. I'm doing this because I like stacking perfect clears and just general 20G pattern building.

But here's the real kicker: I don't need to use a Playing Forever method to play invisible at all - I can stack normally, no problem. It's just not as fun, so I don't do it as often. And since I rely on one generous person who is not me to make all of my videos, I never use them. I still use bag and SRS, of course - Phantom Mania, unlike Speed Mania 1-2 and Grade Mania 1-3, is not isomorphic to a TGM game and I see no reason to treat it as such.

For those of you who think my approach is fundamentally different from most - obviously. But if all I did was stack for Tetrises, occasionally filling in overhangs or skimming to fix holes, I'd probably even get bored of invisible. The challenge of PF-V3.1 in Phantom is it makes my short-term memory work hand-in-hand with my long-term memory while speed mounts against my hand-eye coordination. I do it because it's extremely hard, not straightforward. If it were straightforward, it wouldn't have taken me two whole months to get MK. And if you're thinking that maybe I just suck, remember, I invented the method in two weeks and learned to execute it in 20G before moving to invisible the week after. It's not unreasonable I might be MV now if I'd gone with Playing Forever.

This is how I play invisible - and I'm not changing.
User is offlinePM
Go to the top of the page
+Quote Post
Rosti_LFC
post Mar 2 2015, 02:16 PM
Post #15


Tetris Grand Master
Group Icon
Posts: 1,237
Joined: 26-June 09



I like your use of the spoiler tags, makes things a lot easier to follow
EDIT: though they broke for me when I tried so fuck it Frown.png
EDIT2: and for some reason that I can't fathom, quotes aren't working for me either so I've used bold instead until they do (╯□)╯︵ ┻━┻

[quote]Sorry, I could be misreading this, but are you insinuating I obfuscated PF-V3.1 in the post unveiling it?[/quote]

I don't think I was insinuating that you obfuscated it, more that from the long explanation and accompanying fumen, there's a bit of a sense of "well this is theoretically interesting, but so what?". I feel the invisible demo completely breaks through that aspect - it makes it less of an ivory tower concept and more something that's a genuine tool to break through sections of Tetris which would be difficult otherwise. Or at least that's how I saw it.

Plus in terms of wider reach, a explanation on playing forever will always be a bit tl;dr to a lot of people, whilst clearing 500 levels of invisible Tetris has a lot more bite to it.



[quote]A very good point. Yes, me being the first to do something in Phantom is not very significant. [/quote]

I would say in this case it is significant, I was just making the point that the rarity of an achievement is not on its own a particularly good judge of how difficult or revolutionary something is. To rag on a rare achievement that I have myself, there are very few players in the world who have gotten an exact 99,999,999 max-out of Tetris DS Endless Marathon (so that it shows up on the records page in-game). That's not because it's hard, it's actually incredibly easy, it just takes bloody ages and requires some gimmicky play to get the 999 part of the score precisely. On the other hand there are quite a few people in the world who can run a marathon in under 3 hours - something that personally I'd say is really fucking difficult. There's some correlation between difficulty and how many people can do something, but it's not the whole story.


[quote]
But more importantly, I really, really need to address this line in red here. It suggests that if you know this method well enough, you can execute it without seeing the pieces, and play indefinitely - which is by absolutely no means true! [/quote]

Apologies if I came across this way, I appreciate that there's still a lot of skill in the execution - I've played plenty of 20G ST-stacking trying to get high scores on Tetris DS and I know that even when you're building the same pattern and can see the playing field, it's still far from trivial to not fuck things up with SRS or back yourself into a corner.


[quote]Frankly, you'd have to be silly to expect you can play invisible Tetris without memorising and/or visualising your stack.
[/quote]

And again, I've played enough invisible to know that even at slower speeds and even when trying to keep the stack as clean and simple as possible there's still substantial challenge. I'm not knocking the fact that you've definitely got a decent handle on the invisible play, after all you've been practising it for quite a bit, but I feel that PF-V3.1 goes a long way to de-skilling a few aspects, and I think that de-skilling does make a pretty massive difference in terms of how things work. Sure, you've still got to memorise and visualise your current stack, but you have a far better idea of where you're going, and there are far fewer possible permutations that your stack can be in (as I understand it anyway), and that takes things out of the realm of superhuman and into an area where it's very impressive but possible for mortals to actually achieve Sticking Out Tongue.png




[quote]Well, except when there's an error between the chair and the keyboard. (That's me, by the way. Hello.) In invisible play, people usually mess up. It's pretty bad for a run and few people can recover from an error so they just stack over it rather than trying to fix invisible holes in their stack. I have no such luxury - playing PF-V3.1 means pieces need to be perfectly placed. One error means it's all over. Why? Because I mess up around 300-500 when the game is already pretty fast (I imagine you're familiar with T.A. Death), I panic and misdrop, and that mistake just piles up. A messy stack is very hard to visualise. That being said, I've given up on being a machine and so many of my newer runs feature mistakes and "recoveries," which are where I restore the playfield to a state where I can continue with PF-V3.1. Can't very well do that if I don't have the playfield memorised, now.

You might think I'm faking it. That I'm pretending to memorise when I'm really just relying on patterns.
[/quote]

I don't think you're faking it - even with a playing forever technique there's still too much situational variation and other things that can happen to be able to rely purely on the pattern without any playing field visualisation. I'd be interested to see some of the mistake-riddled runs, because from personal experience with invisible playing it's the mistakes that are the biggest pain - keeping a rough idea of the top two or three rows of the stack isn't too difficult, but when you've got holes lower down and you lose track of where they are, that's when the invisible aspect seriously fucks you over.

And I might try it myself at some point just out of interest, though my 20G SRS is incredibly rusty these days. I am curious to what kind of a difference it makes.



[quote]But here's the real kicker: I don't need to use a Playing Forever method to play invisible at all - I can stack normally, no problem.
[/quote]

This would be the biggest thing for me - if you have no issue getting MK on PM by just freestyling with SRS and 7-bag (or a more memoryless randomiser), and can get it as often as you do using your Playing Forever techniques, then I guess that renders the PF aspect redundant and means it's not a factor beyond personal preference. Which means I'm wrong in assuming it was.



[quote]The challenge of PF-V3.1 in Phantom is it makes my short-term memory work hand-in-hand with my long-term memory while speed mounts against my hand-eye coordination. I do it because it's extremely hard, not straightforward. If it were straightforward, it wouldn't have taken me two whole months to get MK.[/quote]



I never meant to suggest or imply that it wasn't difficult or impressive - I don't think there are really any feats in invisible Tetris that aren't. Obviously it's substantially less difficult than the constraints TAP/TGM3 invisible rolls, but lower speed and higher lock delay will always do that, and it definitely doesn't negate the hardest aspect of the challenge (after all, there are players who are incredibly comfortably with fast ARS 20G who still can't play the invisible stuff for shit).

That said, two months isn't a long time by the scale of a lot of the people on here - there are plenty that are on the longer side of five years and are still straining for better TGM records or 40L times or whatever.

My initial reaction was that two months is very impressive for MK on PM. The fact that you didn't do it with TGM2 or TGM3 rule kind of negates a chunk of the difficulty - if you'd done it with step-reset lock delay then I'd have been seriously impressed (and most likely incredibly sceptical) that in two months someone would be able to beat out the likes of Kitaru and KevinDDR in the invisible domain. It also breaks my familiarity of the difficulty, because I know how hard ARS 20G invisible is all too well, but I've never really played much Phantom Mania with SRS.

So then, from the context of "which aspect makes someone beat scores of guys like Kitaru (who has been playing for years) in just two months?" and on the assumption that you're not some crazy invisible god the likes of which we've never seen before, it was either due to the SRS, or due to the Playing Forever techniques. I, perhaps naively, figured that it was the PF stuff, because to be honest that was pretty novel to me, and I feel that if it was *merely* SRS that someone else would have smashed Phantom Mania long ago. If you say that actually the Playing Forever aspect isn't that big a deal, or in particular if it makes it harder, then I'm happy to admit I massively misunderstood things and I apologise (though suspect I'm not the only one in this position). But in that case I feel you should make the PF aspect the focus and not the grades you're getting in PM, because as it currently stands it comes across (to people like me who haven't tried it much) like the PF technique is a tool that can be used to break and solve invisible Tetris. Not that the PF technique is the challenge and accomplishment in itself.
User is offlinePM
Go to the top of the page
+Quote Post

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

 

©2009-2014 Hard Drop Community & Forum
harddrop.com is not sponsored or endorsed by The Tetris Company or its subsidiaries.