[Help Thread] OLL and PLL Discussion

[mark49152]

You finish your 2 look OLL, then scan the LL for a PLL you know, after 3sec of inspection you realize it's actually a N perm you didn't learn yet, so you run your 2 look PLL that you could have done 3sec earlier.

If it takes you 3 sec to recognise an N perm, knowing the alg isn't going to help much. Actually everyone goes through a phase of mixing 1- and 2-look while learning PLL. In my experience, it works the other way around. You learn to recognise the cases you know more quickly, and default to 2-look if you don't recognise the case quickly enough. In other words, don't spend 3 sec recognising it's an N perm, just spend 1 second checking if it's a case you know, and if you don't recognise it quickly enough, move on with 2-look.

 

[brian724080]

If it takes you 3 sec to recognise an N perm, knowing the alg isn't going to help much. Actually everyone goes through a phase of mixing 1- and 2-look while learning PLL. In my experience, it works the other way around. You learn to recognise the cases you know more quickly, and default to 2-look if you don't recognise the case quickly enough. In other words, don't spend 3 sec recognising it's an N perm, just spend 1 second checking if it's a case you know, and if you don't recognise it quickly enough, move on with 2-look.

That means you're two-looking the recognition. Just learn the case, and learn how to recognize it, it's not difficult.

 

[mark49152]

That means you're two-looking the recognition.

What does that mean?

 

[aboeglin]

The problem, I find, is that if you don't recognize it within say 1 second as you say, it doesn't mean it's not a case you know. Maybe you knew the algorithm and you'll still run 2 algs where 1 was needed. But I suppose everyone is different. While learning the algs I just don't time myself on a whole solve anymore since the LL is "under construction", I'll wait until I know all the algs to do this. For now, I rather focus more on my F2L ( which I time ) when I do full solves, and then take the time to learn how to recognize PLL cases and run algs slowly in a first place to find the most appropriate fingertricks. Even if I don't know the alg, I still look it up on the list and run it, it makes the learning later easier since you've seen the algorithm before. But obviously it works for me and everyone should do it the way he thinks it'll be the most efficient for himself.

 

[mark49152]

The problem, I find, is that if you don't recognize it within say 1 second as you say, it doesn't mean it's not a case you know. Maybe you knew the algorithm and you'll still run 2 algs where 1 was needed.

Yes, in a timed solve, I would just get on with 2-look rather than waste more time trying to recognize the case. I'd rather spend 1.5 sec on an A perm than 3 sec concluding that the case is a nasty G perm. In untimed solve, obviously I would practice recognition. But that was just my experience. Nowadays I know full PLL and can recognize quickly from any two sides anyway.

4x4 is a good example. Parity messes with PLL recognition. If I don't immediately recognize it as a normal PLL, I will do a quick A or V perm to fix the corners, reducing it to a simple and easily recognizable (parity) EPLL. Quite often I will end up with a regular EPLL, meaning there was no parity, but that's OK. Before I started doing this, I would fix parity if I didn't recognize normal PLL, but that meant I spent longer on recognition trying to be absolutely sure before committing to the fix.

 

[OneMinuteMan]

Any Advice on Last Layer

Hey, this is my first time on the site... I just picked up a rubiks cube a month ago from Walmart, and learned how to do it in around 3 minutes. Now that I've been practicing a lot, I bought a Zhanchi and Guyhong (Love my Zhanchi) and my best time now is 56 seconds. I am wondering if someone who is experienced, could tell me how I could tweak my last layer method and not have to learn OLL and PLL, I will learn this over time im sure, but I want to know any shortcuts to get under 45 seconds. Right now:

- I get the cross on the bottom
- F2L
- Then I look for a Dot/Backwards L and do (F,U,R,Ui,Ri,Fi) if its a straight line ill do (F,R,U,Ri,Ui,Fi)
- Then I do (R,U,Ri,U,R,U,U,Ri) until the top is all yellow... I usually have to do this algorithm 3/4 times until the top is completed.
- I match up a corner piece and then with that piece in the back left or right corner... Id do (Ri,F,Ri,B2,R,Fi,Ri,B2,Ri2) If there was already a 2 matching corners this would put all 4 corners in the right spot, but if there was only 1 matching corner then id have to do this algorithm 2x to get all 4 corners matching.
- I then match put all 4 corners in their right spot, and if 1 color on the top layer is solved I'd but that in the back then determine if the colors need to go left or right and do the algorithm for whichever it is... Right (F,F,U,Ri,L,F,F,R,Li,U,F2) Left would be Ui instead of U. Id have to do this 2x if the no edge was in its right position.

Im sorry if this is hard to understand, im sure its not though with how many experience cubers are on this site. Im a newbie, so I am just trying to learn; and I have become obsessed with finding new ways. Again, please let me know any adjustments I could make to my way of doing it now to get a faster time, I can usually finish the Cross and F2L in 20-25 seconds... then the last layer takes me awhile. Which Im sure you already know by looking at how I do it. I already know that learning OLL and PLL would be the best bet, but I want to keep making a few adjustments at a time instead of learning all of them. I am in the process of learning PLL and I have been keeping track of which algorithms come up most often in my solves, and out of 104 solves... here is what I came up with (instead of learning the most common PLL algorithms i see, could you maybe see a connection with how I do the last layer and any shortcuts I could do?):

Ub 9
Aa 9
Ua 8
Gd 8
T 7
Rb 7
Y 6
F 6
Ra 5
Jb 5
Ab 5
Ja 5
Gb 5
Gc 4
Na 3
V 3
E 2
H 2
Ga 2
Z 2
Nb 1

Thank you in advance, again Im sorry for if this is confusing! I just am loving learning new ways, and figuring out how to make adjustments without having to learn all 21 PLL algorithms/the crazy amount of OLL cases. I also dont want to learn 2 look, I just want something for the time being until I can learn all PLL and OLL cases.

Video of me doing LL:

 

[aboeglin]

Here you'll find the probability of any PLL case to happen : http://badmephisto.com/pll.php

Also, have you heard about 2 look PLL and 2 look OLL ? It is the same as PLL and OLL, but with intermediate states. The first look OLL gets you the cross, the second one completes the orientation. And then you have 2 look PLL, one to permute corners and one to permute edges. So I would definitely go for this. You actually already know the algs to get the cross on the last layer I believe, so you just need to learn the 7 OLL cases that contain a cross ( out of 57 for full PLL ), and for 2 look PLL I think you need about 4 or 5 algs.

What's your average at the moment ? About 1:15, 1:20 ? And what means you can "usually" finish the F2L in 20-25sec ? When you give times it's better to speak about averages since best times are usually due to good F2L cases that you can run very quick and smooth and/or skip LL ( partially or totally ). My best time at the moment is about 12sec bellow my average which is in no way representing my "usual" solves.

Because if you do the F2L in 25sec, and you average 1:15, you effectively need to start worrying about LL. What you can do to see better where you stand is to time yourself in two times, you do a few solves and you stop the time after F2L completion. Then you time your LL. It should be rather balanced but LL should take less time than F2L since it is supposedly only consisting of performing algorithms, and need fewer moves than for the first two layers.

 

[OneMinuteMan]

I greatly appreciate your response! Yes I have heard of 2 look PLL and OLL, my average is around 1:09-1:14, and about the 20-25 seconds, I am new to this again so im sorry for the confusion, my average is 25-30 seconds to get F2L. The reason why LL takes me so long is because I am doing a lot of the same algorithm multiple times, the LL is definitely taking me way to long to complete the way im doing it now. Like I said before, I am actively learning the PLL cases, and its taking some time to say the least lol. But after looking at my original post, is there anything else I could do to decrease my time on the LL without having to learn the PLL/OLL cases. I understand I could learn the 7 oll cases that involve the cross, which is great advice and is what I intended on doing. But maybe look at how I do it with the algorithms I put on the orginial post, and let me know where I am wasting too much time on LL? I know since im doing the same algorithm 2-4 times on LL is where its happening, but maybe some solutions as well?

Here you'll find the probability of any PLL case to happen : http://badmephisto.com/pll.php

Also, have you heard about 2 look PLL and 2 look OLL ? It is the same as PLL and OLL, but with intermediate states. The first look OLL gets you the cross, the second one completes the orientation. And then you have 2 look PLL, one to permute corners and one to permute edges. So I would definitely go for this. You actually already know the algs to get the cross on the last layer I believe, so you just need to learn the 7 OLL cases that contain a cross ( out of 57 for full PLL ), and for 2 look PLL I think you need about 4 or 5 algs.

What's your average at the moment ? About 1:15, 1:20 ? And what means you can "usually" finish the F2L in 20-25sec ? When you give times it's better to speak about averages since best times are usually due to good F2L cases that you can run very quick and smooth and/or skip LL ( partially or totally ). My best time at the moment is about 12sec bellow my average which is in no way representing my "usual" solves.

Because if you do the F2L in 25sec, and you average 1:15, you effectively need to start worrying about LL. What you can do to see better where you stand is to time yourself in two times, you do a few solves and you stop the time after F2L completion. Then you time your LL. It should be rather balanced but LL should take less time than F2L since it is supposedly only consisting of performing algorithms, and need fewer moves than for the first two layers.

 

[Zeotor]

Hey, this is my first time on the site...

Welcome!

I am wondering if someone who is experienced, could tell me how I could tweak my last layer method and not have to learn OLL and PLL, I will learn this over time im sure, but I want to know any shortcuts to get under 45 seconds.

In short, 2-Look OLL and 2-Look PLL are what you should use for your last layer. I'll go into more detail below.

- Then I look for a Dot/Backwards L and do (F,U,R,Ui,Ri,Fi) if its a straight line ill do (F,R,U,Ri,Ui,Fi)
- Then I do (R,U,Ri,U,R,U,U,Ri) until the top is all yellow... I usually have to do this algorithm 3/4 times until the top is completed.

Up until this point, 2-Look OLL (which comes before 2-Look PLL) is similar. With 2-Look OLL, you use one algorithm for the dot, backwards L, or straight line. You use a different algorithm for each case. There are seven cases left after that. Each one has an algorithm that solves it.

- I match up a corner piece and then with that piece in the back left or right corner... Id do (Ri,F,Ri,B2,R,Fi,Ri,B2,Ri2) If there was already a 2 matching corners this would put all 4 corners in the right spot, but if there was only 1 matching corner then id have to do this algorithm 2x to get all 4 corners matching.
- I then match put all 4 corners in their right spot, and if 1 color on the top layer is solved I'd but that in the back then determine if the colors need to go left or right and do the algorithm for whichever it is... Right (F,F,U,Ri,L,F,F,R,Li,U,F2) Left would be Ui instead of U. Id have to do this 2x if the no edge was in its right position.

2-Look PLL would be the final thing you'd do to solve the cube. It is different than the approach that you use. With 2-Look PLL, you permute (put in the correct place) the edges and corners in separate steps.

So, with those two last layer methods, you use sixteen or seventeen algorithms instead of seventy-eight.

so I am just trying to learn; and I have become obsessed with finding new ways.

This is good. Looking for new and better ways to do things will help you continuously improve.

(Ninja'd by aboeglin, but I'll post this anyway because I provide links to the Wiki.)


EDIT:
How many algorithms are used in 2-Look PLL? The Wiki's page states seven, but only lists six.

 

[Jokern]

Also this might help: http://www.cubewhiz.com/pllrecognition.php

 

[aboeglin]

Actually I think you can do it with 3, and one is a reverse of the other, so 2 and a half. But with this version you only permute 2 corners or 3 edges at a time, which means you might need to perform up to 4 algs. Even though they are easier and shorter than the ones of the beginner method it still takes a bit of time. But I did this for a while before jumping to full 2-look PLL, since I was learning 2-look OLL at the same time. You can find it here : http://solve-the-cube.com/intermediate.html#step2. Then later I learned other algorithms to get to real 2-look PLL. So, I learned H and Z perms to avoid having to do 2 U perms to permute edges. And finally I learned E perm. So yes, I think 6 algorithms are actually enough to do any case in 2 looks.

 

[aboeglin]

I now know all PLL algs, I still struggle with recognition though. A, R and G perms are a mess. I find R and G very similar, only difference is that R perms have 2 corners solved, but I always need to AUF to find it out which makes me loose precious second, still I believe it'd be better than 2look PLL and I'm certain it'll get better.

Also for recognition there's another link from badmephisto which collects all the algs and diagrams in one single pdf : http://badmephisto.com/badmephisto-speedcubing-method.pdf. There we can see that for G perms if you hold the headlight on the left, each case had the 2x1x1 block positioned differently. So now everytime I discover I face a G perm I directly but headlights on the left and then it's easy to find the case, and it's already setup to run the alg. But still, I need to first make sure I'm not having a R perm.

 

[mark49152]

http://www.speedsolving.com/forum/showthread.php?41108-Two-sided-PLL-recognition-guide

 

[mark49152]

I recently learned the L-shape OLLs and am having trouble recognising them quickly. I can recognise and execute them fine, but in a timed solve, because there are six similar cases, I have to look at multiple sides to distinguish them and it just creates a pause. Does anyone have any good tips on recognising these quickly?

 

[aboeglin]

OLL and PLL recognition is definitely nothing easy to get to a good level. I cannot help you since I still do 2-look OLL. But I learned full PLL last week and I now start to see some improvement on my recognition even though it's definitely not there yet. There's apparently no magic trick to get it better, you need to practice them, a lot. It can be frustrating at first to loose 3sec on just recognizing the case, but it still isn't that slower than having to run a 2-look OLL I suppose. What I do for my PLL, I still time myself, I barely got back to where I was with 2-look PLL, but I try to have a look at my F2L times, which are the ones that matter the most to me. Sometimes I would also just time myself until F2L only on a few solves ( to measure the progress of my F2L ), take my time to work on recognition for LL algs, run a few algs I'm not that great at performing, and scramble again.

 

[Phillip1847]

I recently learned the L-shape OLLs and am having trouble recognising them quickly. I can recognise and execute them fine, but in a timed solve, because there are six similar cases, I have to look at multiple sides to distinguish them and it just creates a pause. Does anyone have any good tips on recognising these quickly?

I'm assuming you are referring to the Little L ones.
The first ones - the F (double sexy) F' and its mirror are easy - there are headlights and everything else is broken up. If you just see the bar and no headlights, they are on the opposite side. The headlights are placed on the left or right side. If its on the left, do F 2xsexy F', if its on the right, do the other one.
The next ones - the frying pans - have a 3 long bar and headlights opposite. Put the headlights on L/R. If its on the Right, do the right alg, if on the left, do the left alg.
The last ones - The squeezys- have the 2 bar and the 3 bar. I had trouble distinguishing between them and always ended up doing the wrong alg until I sat down and figured which one is which. There are no headlights, which made it more difficult for me.
Put the L on the left side. AUF until the secluded corner faces B or F. If it faces B, Do R' F...... if it faces F do R B'...
I don't know what algs you use, so the squeezies might be wrong.

 

[mark49152]

Thanks. I have no problem doing any of that. Recognition is easy enough; it's doing it quickly I struggle with.

The 3-bar is the most obvious feature, so that's what I look for first. That means checking two sides. If I find it, great. However, if there's no 3-bar I then have to look at each of the legs of the L to see which has the headlights around it. So in the worst case, I have to look at all four sides of the cube to recognize the case. I can't think of any other case, OLL or PLL, where I have to look at all four sides.

Anybody know any good tricks for faster recognition?

Alternatively, any good algs for OLL 47 with headlights on left? That would make it easier because I would always AUF the Pi corner case the same way and choose the alg according to whether the L is at front or back.

 

[Phillip1847]

Well, if you look at the F face, the U face and the L face you can technically recognize every OLL. The only thing you cant see from those are with way the BRU corner is facing if it is unoriented, but the last corner depends on the rest so you can infer that with 100% accuracy.
I would recommend going through each case and looking at those faces for characteristics, for the AUFs as well.
Other than that, a more experienced cuber might be able to help you.

 

[mark49152]

Well, if you look at the F face, the U face and the L face you can technically recognize every OLL.

Technically that's true, but in practice, as for two-sided PLL recognition, you need to apply some simple logic to the sides you see to reach a conclusion.

I think I have worked it out and it's quite simple. I look on the two sides away from the legs for a 3-bar. If there is one, AUF it to the left, and case depends on H or Pi corners and whether the L is front or back.

If there isn't a 3-bar, there must be a 2-bar, so AUF that to the front. Case then depends on whether the L is left or right.

Will practice it for a while and see if it saves me the 2-second pauses.

 

[aboeglin]

Coming back to PLL briefly. My question concerns E perm, I use this algorithm at the moment : x' (R U' R') D (R U R') D' (R U R') D (R U' R') D' , even though I might try to perform it with R U and L faces at some point, I have difficulties with the D' in the middle. I can recognize the case easily, but I always struggle to AUF correctly to perform it. The way I found now is to place the edges correctly, and then check what colors I have on the F side of the UFR cubie, if it's the same color as the R face, then I'm good, otherwise I do U or U' and perform the alg. But I loose a lot of time doing this.