Printable Version of Topic

Posted by: Paradox Jun 2 2015, 01:26 AM

This doesn't even make sense anymore lol... crazy stuff.

If he gets sub-15 I'm going to play tetris again and marathon it until I get sub-20.

Posted by: MicroBlizz Jun 2 2015, 04:46 AM

According to perfectclear, 15 seconds is the extent of human potential assuming I learn 180 rotations. If someone were to find a key bind that utilizes all 10 fingers apparently it'd be 8 seconds...

Posted by: kastle09 Jun 2 2015, 05:51 AM

When's the documentary coming out talking about how crazy this year has been in the 40 line sprint world record.
I reckon it would be a good story. All the intrigue, east vs west, Romance and Betrayal, Glory and Death...

Posted by: MicroBlizz Jun 2 2015, 06:14 AM

I mean if someone approached me to do it I would but no offers so far lol

Posted by: XaeL Jun 2 2015, 07:09 AM

this already exists...

Posted by: kastle09 Jun 2 2015, 07:32 AM

Faaaark. is this another one of those things were I end up doing it myself... Last time i said "wouldn't it be cool if" I ended up being a tournament organiser. And that shits still going.

Posted by: Ravendarksky Jun 2 2015, 08:10 AM

How long until MicroBlizz ascends from earth to join the gods?

This is getting beyond me. I think I will set my new aim to keep my sprint time less than DOUBLE the WR. If I can place one piece for each of MicroBlizz two then i'll be happy

Posted by: Paradox Jun 2 2015, 06:11 PM

It's hard to set a limit on human potential and I would think sub 15 is possible. Definitely possible for 10-finger inputs but don't waste your time lol.

I mean I got ~19 seconds a long time ago using 10 finger controls and nobody cared so it probably wouldn't be worth it to try lol. I think its because I used the deep drop feature which let you do really weird stuff that would be either impossible or would require soft drop normally.

Posted by: MicroBlizz Jun 2 2015, 06:11 PM

Ya I was just reiterating what was said to me. Even knowing that, I wouldn't want to go further than 15 if that's even possible. At the moment, with a lot of luck and nerves of steel I think the most I'll pull off is a 16.9x

Ya I don't think people care about creativity as much as they should here. A lot of people I've seen do amazing things for Tetris have left because of poor reception. Personally I think this community is a bit broken so I wouldn't too feel discouraged about that. TU PC will fix everything when it comes out! ... I hope... (cries silently in a corner)

Posted by: Raqueloxo Jun 2 2015, 07:11 PM

We all are, Philicia, we all are. ;-;

Posted by: MicroBlizz Jun 2 2015, 09:15 PM

What are the key binds?

Posted by: XaeL Jun 3 2015, 01:02 AM

https://sites.google.com/site/xaelous/finesse

In terms of a more regular setup:

Current:
left rotate, right rotate, hold, 180 rotate
hard drop, left, right, firmdrop

This uses 8 fingers. Discluding firmdrop you have 7 fingers.

If you added an Instant das key (i.e. a button that charges your das instead of waiting 4 frames) that would be a keybind that uses 9 fingers (8 discluding firmdrop)

I plan on someday implementing Instant DAS into nullpomino. I think it should be fairly easy to master and should drop times by a significant amount.


On top of that, nullpo has a fixed frame rate of 60fps.
Ideally, you would want an update rate of infinity (or more realistically, whatever your OS lets you get away with, either 10ms or 1ms). For example, your time was 17.50. Since nullpo only updates once every 1/60, your times are rounded UP to the nearest 1/60. By doing ABSOLUTELY nothing, your time could have actually been (up to) 15.99 milliseconds faster. Also due to the fact that nullpo ends the game the frame AFTER you actually finish, thats actually 31.99 milliseconds faster. That's only 0.03 seconds, but still.

Having variable update rate means you can also have non 60hz das - imagine 3.5 frames (i.e. 0.0583...) of das rather than having to choose between 3 (0.05s) and 4 (0.06...s)

Posted by: tk198 Jun 4 2015, 02:36 AM

Oh my goodness... Grats Micro. I still remember When sub 23 was the hype...

Posted by: MicroBlizz Jun 4 2015, 04:20 AM

Is there a client I can play on that supports instant das for the moment or nah? Mmm I think 3.5 frames would help a lot. 4 is getting slow and 3 is still too fast.

Posted by: simonlc Jun 5 2015, 04:12 AM

I could make a Tetr.js - MicroBlizz variant, with any feature you want.

Posted by: Alexandra Jun 5 2015, 04:21 AM

But Phil is going to get a minimum 15.98 second run.

Posted by: simonlc Jun 5 2015, 01:04 PM

It's also now possible to implement microsecond accuracy. With an event based system, and more accurate time, we can get a more exact reading of time. We can even go to the nano second level, though not on a webpage.

Posted by: caffeine Jun 6 2015, 12:10 AM

Let's say the current world record was accomplished on a centisecond game with a time of 17.00. Then, someone plays on a millisecond game and gets a record of 17.009. Who would have the world record?

Obviously, if the game runs at 60 FPS, then it won't make any difference to count milliseconds.

If the game runs at 100+ FPS or is event-based, and it counts milliseconds, then crafty players will prefer playing on games that don't count milliseconds. Who wants to get 17.009 when they could've got 17.00 for finishing at the same exact second?

So one option would be, for the sake of comparison, to assume all centisecond records end in xx.xx9. I don't know how popular this idea would be. Actually, it might be possible to find a lower bound by dividing the total number of frames a game lasted by 60. (But now we're entering unsettling territory--if you take too close a look at how some FPS games round down/up their record, you may realize you don't really have the record you think you should.)

Another option would be to allow for milliseconds, but then to round down to the centisecond only for comparing with other game records.

Posted by: MicroBlizz Jun 7 2015, 03:12 AM

How would a game that can run in higher fps impact the gameplay exactly? I wasn't clear on that ._.

Posted by: farter Jun 7 2015, 05:08 AM

more accurate DAS delay timing that may let you use even lower delay value. because 60fps based game always give +-16ms random error, in cases the arrow key being pressed for same amount of time which is near the das delay defined (x * 16.6ms), sometimes it shifts and sometimes it does'nt.
obviously that's not what you want. so using higher frame rate reduces the error range, like 1000fps for 1ms error. or event based system (best on theory but harder to implement)

Posted by: caffeine Jun 7 2015, 02:31 PM

What I was getting at was how each second can contain a maximum of 60 game states in a 60 FPS game. Since centiseconds allow for 100 states a second, centiseconds already capture as much information as milliseconds would in a 60 FPS game. Milliseconds don't reveal anything you didn't already know until once the game has more than 100 states per second.

Posted by: Boingloing Jun 7 2015, 04:58 PM

Was this 40 key? I can't find documentation of this time anywhere, I'd been trying to see how far people had taken 40 key so I know what my goal is now

Posted by: myndzi Jun 8 2015, 08:25 PM

Edit: disregard that, I can't units

And WTF Blizz. You're insane.

P.S. there's nothing hard about an event-based system, at least in Javascript, which is what it should be written in. I'm sorry for not writing code

Posted by: XaeL Jun 9 2015, 01:53 AM

Does javascript have an efficient/fast graphics/input library? Just wondering since we want maximal performance.

I'm thinking of writing a simple one in Unity3D. Wouldn't be too hard.

Also, seems like a lot of libraries die at > 1000fps lol. (just had quick look at unity)

Posted by: simonlc Jun 9 2015, 05:17 AM

JavaScript (in the browser) pretty much has a 60fps cap (in the browser). However, it is an event based language. Depending on the platform you can get microsecond (in browser), to nanosecond (node.js) accuracy, but typically you have millisecond accuracy.

For the graphics, JavaScript (browser and nw.js) has the canvas element which has both 3d and 2d modes which can be GPU accelerated. The new Unreal engine was ported to JavaScript and works on this tech. If you want the best performance in JS you want to build a game in C++ and compile it with something like emscripten into javascript (which unity 5 can do).

As for input libraries, there's nothing that I've seen out there that is any good. I'm currently working on one, but for a frame based input game. I've recently been debating the idea of event based tetris in my head again, though I still don't have a good idea on how to implement it to my standards.

Posted by: MicroBlizz Jun 9 2015, 05:47 AM

I don't know anything about coding and what not but let me know if you need someone to test it simonlc :3

Posted by: XaeL Jun 9 2015, 06:35 AM

i was thinking more along the lines of really fast update loop with separated video/logic threads (this isn't possible in Unity3D, but would be possible in LWJGL/DIY engines). Instead of events, i would do frame-rate independent code (e.g. das measured in ms rather than frames), since in my experience events have a large enough overhead to be highly inaccurate/inconsistent.

Posted by: simonlc Jun 9 2015, 07:55 AM

Well javascript is event based, and single threaded. However it's asynchronous (so it can do what you want), and while the event loop can't be perfectly accurate, you can get around that by cheating a little. All this while being more efficient than running at 1000fps.

Posted by: XaeL Jun 9 2015, 11:38 PM

Ok, LWJGL supports event based input. It deals with 1000hz mouse easily so i might start coding something simple in that.

Any reason why Javascript is the shiz? I don't see how it matters what language is used as long as its event based.

Posted by: simonlc Jun 10 2015, 05:06 AM

It's light weight, fast, supported everywhere, safer, doesn't hassle you for updates (like flash and java), and requires no install.

Posted by: myndzi Jun 11 2015, 10:55 PM

Edit: moved here: http://harddrop.com/forums/index.php?showtopic=7344&st=0&gopid=86260&#entry86260

Posted by: Okey_Dokey Feb 25 2018, 09:21 PM

Meanwhile, MicroBlizz has reduced his record to 17.274 seconds (https://jstris.jezevec10.com/replay/375085). I think it's the first time that the WR is held on Jstris.

Posted by: caffeine Feb 26 2018, 01:25 AM

https://docs.google.com/spreadsheets/d/1ahL3GQo9RaR-zZUrIjoPdPkOG-CIh6J_qDlKPGp0W8c/edit#gid=2052996781

In the sub-30 era, the first WR on a new game are as follows: