Faster in 20G: ARS or SRS?

[Rosti LFC]

I acknowledge SRS has superior speed, but I just can't bring myself to try and overcome the horrific tetlag that comes with the switch.

[mat]

i'm competent in both styles, my previous 40 line record was in SRS. but i just really prefer ARS. TOJ by the way would be 100 times more appealing to me, even with the horrible DAS, if it had rotation selection.

[Cubicz]

I acknowledge SRS has superior speed, but I just can't bring myself to try and overcome the horrific tetlag that comes with the switch.


I don't know about that. I'd say it'd be between 20g ARS and 0g SRS, as top speeds for both of those are 200+ tpm. But there isn't nearly as large a field of 0g srs players compared to japanese TGM guys. So I don't think we know for sure.

[DIGITAL]

I acknowledge SRS has superior speed, but I just can't bring myself to try and overcome the horrific tetlag that comes with the switch.

I don't know about that. I'd say it'd be between 20G ARS and 0G SRS, as top speeds for both of those are 200+ tpm. But there isn't nearly as large a field of 0G srs players compared to japanese TGM guys. So I don't think we know for sure.


ARS wins in 20G hands down. As for 0G, the average number of inputs per piece with SRS is lower than that of ARS due to the initial orientation and the 4 rotational states. TGM doesn't do SRS justice with all the sub 20G delays, and that's the area that SRS is supposed to excel in.

[caffeine]

"ARS wins in 20G hands down." How? Technically.

[DIGITAL]

"ARS wins in 20G hands down." How? Technically.


The downward orientation lends itself to grabbing on to the surface much more easily, especially when DASing. Synchro + DAS Synchro stops piece from getting stuck and opens up a wide range of practical placements. Due to the tetrominoes rotating at an even level, ARS can push off of one row tall walls instead of jumping right over them and it can rotate under overhangs that SRS normally can't rotate under.

[colour_thief]

"ARS wins in 20G hands down." How? Technically.


We've had similar discussions about this in the past... http://www.tetrisconcept.net/forum/images/smilies/icon_rolleyes.gif

It would actually be a lot of work to show, specifically, why TGM's L/J stances are preferable and why the wallkicks are more practical. If you're willing to actually invest some effort to learn to use TGM-specific tricks, then maybe I would make an effort to show you these things as clearly as possible.


In general, DIGITAL's got the right idea though.

[caffeine]

The downward orientation lends itself to grabbing on to the surface much more easily, especially when DASing. I'm not buying it. You're talking about downward J/L/T, because 20G is faster mostly because horizontal stoppage (going left or right, and then something stops you). Then there's the rarer single-hole-in-the-floor-fits-perfectly-for-downward-J/L/T. There's a simple solution for grabbing with these in 20G SRS: rotate twice (or once for double), and then DAS! Now you may say you place J/L/T downward more than upward, and in that sense you save time, and that's fine, but I'd need some sort of math or something to believe that. Now on the other hand, upward orientation is definitely better than downward in the sense that you don't get stuck in those middle-field traps, and have more options. Synchro + DAS Synchro stops piece from getting stuck and opens up a wide range of practical placements.Wa? You mean it makes it so you can do stuff somewhat like SRS can already do? How does it make it faster?Due to the tetrominoes rotating at an even level, ARS can push off of one row tall walls instead of jumping right over them and it can rotate under overhangs that SRS normally can't rotate under.So you're saying ARS has wall-kicks that favor speed more? I see what you're saying with the one row push-off, but I doubt that sort of thing overcomes the movements saved from 1: four orientation I/Z/S and 2: L/J/T rotate twice push offs only present in SRS.

[DIGITAL]

I'm not buying it. You're talking about downward J/L/T. There's a simple solution for grabbing with these in 20G SRS: rotate twice (or once for double), and then DAS! Now you may say you place J/L/T downward more than upward, and in that sense you save time, and that's fine, but I'd need some sort of math or something to believe that. Now on the other hand, upward orientation is definitely better than downward in the sense that you don't get stuck in those middle-field traps, and have more options.
I'm only talking about J/L. I hate the T orientation in ARS actually. DASing after rotating is a slow compensation. As for upward being better for mobility, yes, it opens up your options, but most of those placements are inelegant and slow. You end up tapping more in SRS than in ARS. There are way too many Ti-ARS videos out there I'm not even going to bother providing examples.

Wa? You mean it makes it so you can do stuff somewhat like SRS can already do? How does it make it faster?
It's not about having as many placement possibilties as possible. It's about having as many practical placement possibilties as possible. SRS may give you a lot of options but most of them are garbage. And point me to a game where SRS can perform DAS Synchro.

So you're saying ARS has wall-kicks that favor speed more? I see what you're saying with the one row push-off, but I doubt that sort of thing overcomes the movements saved from four orientation I/Z/S.


Yes, those wallkicks favor speed and they are much more flexible than SRS in regards to utilizing them in combination with DAS. Four orientations make little to no difference when you DAS almost the whole time. Hmm, it might even make it more confusing.

[colour_thief]

Wa? You mean it makes it so you can do stuff somewhat like SRS can already do? How does it make it faster?
It's not about having as many placement possibilties as possible. It's about having as many practical placement possibilties as possible. SRS may give you a lot of options but most of them are garbage. And point me to a game where SRS can perform DAS Synchro.


DIGITAL, this quote is awesome and is basically the end of discussion.

[tepples]

SRS may give you a lot of options but most of them are garbage.
By "garbage" do you refer to moves that are more useful for clearing garbage or sending garbage than they are for single-player stacking?

And point me to a game where SRS can perform DAS Synchro.

As I understand it, DAS synchro needs rotation-movement rather than rotation-gravity-movement, which in turn needs 1G DAS. Games with 1G DAS and SRS include Ti, ACE, Evo, Heboris UE, and Lockjaw. Does synchro work in any of these?

[DIGITAL]

By "garbage" do you refer to moves that are more useful for clearing garbage or sending garbage than they are for single-player stacking?

No, I mean garbage as in worthless for high speed play.

As I understand it, DAS synchro needs rotation-movement rather than rotation-gravity-movement, which in turn needs 1G DAS. Games with 1G DAS and SRS include Ti, ACE, Evo, Heboris UE, and Lockjaw. Does synchro work in any of these?


I don't recall it working in any of these. Of course, I've never played Ti or ACE, but according to Heboris, they don't. As for LJ, I'm sure you would know best. http://www.tetrisconcept.net/forum/images/smilies/icon_razz.gif

[caffeine]

First I'll start by pointing out that you said "hands down." If you still don't think SRS is faster than ARS after reading this lengthy post, then I'm confident at least you'll wipe off that "hands down" attitude. I had to wait until my day off to compile this stuff, so please don't just breeze over and disregard it. Try to keep an open mind.


Let's address why 20G is faster than sub-20G in the first place. Tetrominoes begins on the ground, so you can just run the tetromino into obstructions. This interception prevents you from making unnecessary button presses. With experience, the process can be very fluid and effective. Here's a simple example of what I'll call an "intercept."


http://www.tetrisconcept.net/forum/../theintercept.jpg

Now, I want to explain myself better about downward L/J/T, because I don't think you really grasped what I was talking about earlier. I'll call the manuever in question a "grab."


http://www.tetrisconcept.net/forum/../thegrab.jpg

It pops up every now and then, and you say that because SRS has to rotate twice in order to do this, ARS has an advantage. Let's look at the flip side.


http://www.tetrisconcept.net/forum/../interceptadvantage.jpg

What's this? ARS needs to rotate twice in these cases? Didn't you say "DASing after rotating is a slow compensation"? So you have to look at it both ways. When you do, SRS is equal if not faster than ARS in this sense: unless you can show grabs occur more frequently.


You said, "As for upward being better for mobility, yes, it opens up your options, but most of these placements are inelegant and slow." Upwards is useful for speed because it natually overcomes middle traps. That way you don't have to prevent/correct traps or synchro your way out, you just keep playing "elegantly." The out-of-your-way options that require many taps, I say ignore. Their existence won't prevent you from playing efficiently.


Wa? You mean it makes it so you can do stuff somewhat like SRS can already do? How does it make it faster?
It's not about having as many placement possibilties as possible. It's about having as many practical placement possibilties as possible. SRS may give you a lot of options but most of them are garbage. And point me to a game where SRS can perform DAS Synchro.

DIGITAL, this quote is awesome and is basically the end of discussion.

How does it make it faster?


Wall kicks: you keep saying they're better or faster, but I have no idea why. Really, I wish I could argue with you guys about this, but wall kicks and push offs honestly save about one button most of the time. SRS has wall kicks and push offs, too. I know they're useful because I use them! And they, too, save about one button most of the time. I simply see no way to show one system is faster or more practical in this sense.


I really wanted to test this for myself to make sure I wasn't being crazy. I am wrong sometimes, afterall. You said "As for 0G, the average number of inputs per piece with SRS is lower than that of ARS due to the initial orientation and the 4 rotational states." I want to show you this still holds true for SRS. In the past I've done this little test for many different games. Average input is the best way to judge efficiency, and from efficiency comes speed... comes which rotation system can be more optimal for speed. For ARS, I took Arika's TGM3 GM Master Mode video. I took one hundred tetrominoes starting on level 605. This pace wasn't too fast, so he didn't make many (any) mistakes, but it was still fast enough to be fluid and "in the zone." I counted the number of input per tetromino including IRS rotations, minus hold and lock input (not important). Then I recorded a sample game under TGM3 WORLD rules. I played a fifty-line game, and took the middle one hundred tetrominoes because I didn't want to have an advantage for starting with a flat state. The video (http://www.tetrisconcept.net/forum/../expe.wmv) and the data (http://spreadsheets.google.com/pub?key=plvXjzGAyfVmfyzuo43LeJA&gid=0).


The results: ARS = 1.77; SRS = 1.65.


So there you have it. If you want to account for locking, it would be 2.77 and 2.65. Keep in mind jin8 is a very talanted player, and after analyzing his play frame by frame while making this data, it's easy to see that he is remarkably efficient. Also keep in mind that the input/tetromino stat is surprisingly consistent. Game after game of Lockjaw, my keys/tetromino wouldn't go more than +-.05 . In other words, you can repeat the test many times, but I'm confident that if you play close to optimally, you'll find similar results.


In conclusion, if you want to say ARS is more practical or natural or instinctive... or whatever, that's fine. I might just side with you on some of those points, but I will certainly disagree with you if you say ARS is faster-- especially if you say "hands down."

[mat]

without really deciding on an opinion i would quickly offer up that comparing an ARE game to a no ARE game in terms of elegance is a bit shady, but i find this type of comparison very interesting. i think a lot has to do with the differences in stacking--the typical 20g pyramid ends up getting built quite a bit differently by someone well versed in SRS. more gaps, more terrain in general. but that's still just my impression, i think i'm going to have to conduct a similar study myself before drawing any real conclusions.

[cdsboy]

My question is, does the fact that srs can "kick" out of holes make game play faster or slower? Are we talking about which one is faster in a game where the player plays with perfect form, leaving no holes where a piece can get stuck?

[caffeine]

When you say "kick out of holes," you must be talking about those crazy last priority wall kicks. You see, a good player knows not to bother with that. You memorize the twists that work, and then you add them to your repertoire. I, personally, never see those "kick-outs" when I play.

[cdsboy]

When you say "kick out of holes," you must be talking about those crazy last priority wall kicks. You see, a good player knows not to bother with that. You memorize the twists that work, and then you add them to your repertoire. I, personally, never see those "kick-outs" when I play.


Right. I was halfway asking if they made the game faster or slower.

[caffeine]

How would they make it faster?

[kotetsu213]

They're faster because they make the pieces magically warp to the other side of the screen, perfect for stacking.


To really determine the winner: optimized TAS of two games, one ARS, one SRS, same piece sequence

[tepples]

GBA games can be TAS'd, so why not try it on Lockjaw? It supports both SRS and ARS, and its randomizer is determined by the number of frames from power-on to game start.


Does anyone have the formula for generating the "power-on pattern" of Sega's arcade Tetris game?

[caffeine]

"They're faster because they make the pieces magically warp to the other side of the screen".


Wow, now you've lost me. I thought we were talking about when you put a tetromino in a gap, try to rotate it, and it pops all the way out. What are you talking about?

[cdsboy]

"They're faster because they make the pieces magically warp to the other side of the screen".

Wow, now you've lost me. I thought we were talking about when you put a tetromino in a gap, try to rotate it, and it pops all the way out. What are you talking about?


Thats what i was talking about http://www.tetrisconcept.net/forum/images/smilies/icon_lol.gif

[kotetsu213]

I was joking...­

[caffeine]

Oh sorry. http://www.tetrisconcept.net/forum/images/smilies/icon_redface.gif

[colour_thief]

First, thanks for putting in some effort into that post.


Next, here is why I prefer L and J to point down. TGM wallkicks from the pointing down stances are useful in that they can save direction tapping. With SRS the L and J have no useful wallkicks from the default stance. There's no finesse that opens up new practical moves. Here is an example:


http://img227.imageshack.us/img227/8528/jkickfo7.png

The conditions are high speed 20G, and the desired move here is to stick the J under the overhang. The above image shows the initial spawn. From there the player will go:


Right

Counter Clockwise (piece wallkicks right)

Left


And you're done! It's a very simple and elegant solution that really can't be done in SRS. I'm not an SRS expert... How would you get the piece there?


If a TGM player needs the flat stance to jump over a 2-wide hole, there's a special technique to do it, even at high speed 20G:


IRS-A

Left (held down)

C


The trick is to rotate a second time while the DAS is charging. Everything will happen so fast that players who don't know this particular finesse will be slightly confused and think the player did some god-like synchro move or something.


This move highlights an important point that I have raised in past discussions with you. Because button C is pressed during DAS charging, it is essentially free. However, with your kpt analysys it would count it as an extra button. We can argue stances until we're blue in the face and it won't be very conclusive. Your "kpt" stat tries to speak more broadly, however it has a fatal flaw in that it doesn't acknowledge parallel input. There's a huge difference in DAS'ing an IRS'ed piece to the wall, and DAS'ing a piece to the wall and then rotating. Both are used often as the single fastest move for what you want to do. The first is a 1 step maneuver, and the latter is a 2 step maneuver. Again your methods treat these as equal.


What I'm trying to say is, even if we accept your kpt stats comparing yourself and jin8 in a single game as a representative sample of "perfect" TGM and SRS rules play, it is not unreasonable to think that TGM play has more parallelization which makes up for the difference in kpt and gives it the lead overall. After all, we're only talking about a 0.12 kpt difference here.


Even though I find your methods flawed, I agree with your general approach. Lets revisit the problem:


The humain brain is pretty amazing at thinking and, properly trained, it is not the speed bottleneck. Actually moving the pieces is what limits the speed of play. In a perfect world, we'd measure the speed of a move by counting the frames it would take a player with x frames of timing precision and y frames of double tapping recoil. Then we'd average this out and see how many frames per piece on average is required.


Frames per piece... Hmm... Some of you are probably scratching your heads and saying "but doesn't a standard time attack conveniently give us all this information? and World is slower than Classic?". Well, it does and it doesn't. There's no way to tell how tight different players are. Even if you take the World/Classic double Gm jin8, who likely has a constant x frames of timing precision and y frames of double tapping accross both systems, there's no way of knowing his SRS finesse isn't lacking.


Now, I'm not actually suggesting we do this immensely time consuming frame counting exercise... I have something a little more practical in mind. Instead of using frames per tetromino, instead of using keys per tetromino, I suggest using "steps" per tetromino. A step would combine inputs that can be pressed simultaneously or nearly simultaneously. To specify some precise conditions of a step:


-You can only push a given input once per step.

(So left-left and B-B are 2 steps.)


-A period of mandatory waiting ends the step.

(So DAS'ing to the wall or pressing down to lower a piece.)


-A period of mandatory waiting is a step in itself.

(So non-ARE-charged DAS is 1 step for pressing right and a 2nd for travel time...)


-IRS'ing a piece and ARE-charging DAS do not count as a step.

(...and ARE-charged DAS is a single step for travel time.)


And it follows from the above but I'll make it clear:

-an IRS'ed, unmoved piece is 0 steps

-zangis are one step

-double or even triple rotation are one step


http://jagorochi.free.fr/images/triple_rot_high.gif

I... think that covers everything reasonably well. If you see any issues with the "step" system, please raise them. I've actually been half thinking of making a training program that criticizes a player's finesse by comparing it to the best using a step judgement system. At the very least it's a much better measure of finesse than kpt. Though, it becomes somewhat difficult to judge a "step" unless you are pretty familiar with a given rotation system. But if a representative sample of highly finessed SRS and TGM play were compared using steps, I would concede that one or the other was convincingly faster.


http://img236.imageshack.us/img236/6258/controlszg0.png

Of course, this whole argument is pointless though because we all know my tetris clone that reduces all placements to 1 step is the fastest. http://www.tetrisconcept.net/forum/images/smilies/icon_lol.gif

[Amnesia]

What's this? ARS needs to rotate twice in these cases? Didn't you say "DASing after rotating is a slow compensation"? So you have to look at it both ways. When you do, SRS is equal if not faster than ARS in this sense: unless you can show grabs occur more frequently.



Do you consider we use the double rotation for the both? SRS and ARS?

I don't understand in this kind of discussion (because my knowledge is very low in SRS) if you combine the motion of the 2 systems, for lockjaw as example..You talk about the IRS in SRS ? I learn SRS has IRS http://www.tetrisconcept.net/forum/images/smilies/icon_confused.gif .. http://www.tetrisconcept.net/forum/images/smilies/icon_rolleyes.gif
And can you explain me by the way why there is such a big gap between the torikan of ARS and SRS in TI ?

[tepples]

Do you consider we use the double rotation for the both? SRS and ARS?
I don't understand in this kind of discussion (because my knowledge is very low in SRS) if you combine the motion of the 2 systems, for lockjaw as example..You talk about the IRS in SRS ? I learn SRS has IRS http://www.tetrisconcept.net/forum/images/smilies/icon_confused.gif .. http://www.tetrisconcept.net/forum/images/smilies/icon_rolleyes.gif
IRS is present in Ti regardless of the rotation system.

And can you explain me by the way why there is such a big gap between the torikan of ARS and SRS in TI ?

In Ti, rotation system and lockdown system are conflated. "World" has SRS, hard drop, and Infinity, while "classic" has Ti ARS, firm drop, and step reset. A soft drop (to make up for the lack of firm drop) takes time, and because (AFAIK) lock delays are set the same on both lockdown systems, automatic locking will be slower on Infinity. Just be glad that Mihara resisted some of the recommendations of the guideline, such as 1/4G or slower DAS.

[caffeine]

The wiki says Zangi-moves are made possible by sonic drops. Am I right in thinking we wouldn't be counting any of your one-step Zangi-moves in 20G?

[colour_thief]

Right. I was just generalizing.

[mat]

And can you explain me by the way why there is such a big gap between the torikan of ARS and SRS in TI ?
In Ti, rotation system and lockdown system are conflated. "World" has SRS, hard drop, and Infinity, while "classic" has Ti ARS, firm drop, and step reset. A soft drop (to make up for the lack of firm drop) takes time, and because (AFAIK) lock delays are set the same on both lockdown systems, automatic locking will be slower on Infinity. Just be glad that Mihara resisted some of the recommendations of the guideline, such as 1/4G or slower DAS.


the soft drop speed doesn't affect shirase, and no one who would make either torikan (be it shirase or master) is regularly allowing pieces to lock on their own.

[tepples]

But do you need to lock all the way up to 500 to make the torikan?

[mat]

not every piece, but you really do need to play aggressively.

[DIGITAL]

In Heboris DOOM, I have to play extremely agressively at 0-400 to make the Torikan at 2:28, with a few tetrises here and there. My best time is 2:19 and that's due to making a lot of tetrises.

[K]

In TGM3 (20G), does anyone have already seen a non official and reliable video demonstration of SRS high speed ?


this fact talk by himself.


But i must admit it open more possibility than ARS. imo, most of those possibilities are totally useless in high speed since "top players" have to keep an optimized structure for a reduced set of high speed of move (as caffeine also noticed for wall-kick repertory). And ARS show is superiority here. Putting aside player old school preference, people rapidly feel and choose what is good for them...


ps : SRS is used by the fastest japanese player in SAKURA mode because of instant lock.

[herc]

i cannot say much to this thread, because i am neither highspeed in SRS nor ARS http://www.tetrisconcept.net/forum/images/smilies/icon_wink.gif

but my 50cent:


at least the missing floorkick in TGM2 for the I piece is a BIG problem for most players except the true experts. how many times i fucked up my tgm death gameplay just because of not IRS the I-piece fast enough.. and having the I piece fat and long blocking several holes.


but yeah, thats fixed in TGM3! wish i could play it sometimes.

[jujube]

In TGM3 (20G), does anyone have already seen a non official and reliable video demonstration of SRS high speed ?

http://youtube.com/watch?v=PuySIYh9fyY

shirase mode Ti world S13 in 5:19:25 and very non-official http://www.tetrisconcept.net/forum/images/smilies/icon_lol.gif pretty fast, but i think a few are faster.

[K]

so a kind of exception that confirm the rules ? http://www.tetrisconcept.net/forum/images/smilies/icon_biggrin.gif

[Amnesia]

In my opinion, SRS 20G and ARS 20G have the same potentiel of max speed..But the strongest japan players are a bit more focused on ARS..So the best times are ARS..