Tymon Kolasinski tried the belt method during Cyoubx's stream yesterday ("Sub-6's on Stream EZ. Learning Pseudoslotting and CN"). He modified the method slightly - he forced a parity skip during piece separation and rearranged the steps. He also got a 14 and then called a 15 "bad".
So, I have boarded the dummy train and I have tried to optimize this method the best I can with my limited knowledge. First start with ZZ's EO, this way you don't have to deal with the pain of edge parity later. Then solve the belt with moves only. Which is very easy actually. Then separate edges and corners using bars. Make bars by placing an edge between two adjacent corners on the opposite side using M' U2 M. Place the top bar on the left and bottom bar on the right and do R2 U2 R2. If you have an odd amount of edges or corners, solve using bars until your at one corner or one edge on each side that needs to swap. To swap one corner on each side, place one corner on top of the other and do (U R U' R')3. to swap one edge on each side, place one edge on top of the other and do M' U2 M. From there do OLL and PLL, doing corner twist parity as needed, flip over, OLL, PLL, finish!
Yeah, though only squan corner algs would work, and you’d have to orient the corners first. Edges work a bit different because of the way it slices on squan.
When I “invented” this method, I actually realized this potential flaw, so I actually would solve the white side, similar to F2L, but only the edges for the belt were in the right spot. This way, I did OLL once on the yellow side (solving mismatch, edge and corner parities) and the would only do PLL on both sides. I still think it’s decently viable but CFOP is just better in every way
Very interesting. Before I ever watched a single tutorial on solving, I intuitively created a belt around the E axis. I still do a solve like this occasionally.
You may be able to make separating pieces easier if you don’t preserve the belt, but only require those 4 edges stay in the middle. One algorithm at the end could fix it easily or be combined with PLL to make it one step (I think this would triple the number of PLL’s)
My variation on this was to build the top, and bottom faces, then the belt. That way, you don't have to worry about piece separation. You can also build the first face and the belt at the same time, since it's sorta like F2L. The major flaw in this method came in the PLLS. Originally, I thought I'd just fix the parity with M2 U2 M2, then (I assumed) I could memorise all the algs to solve both faces at once. That was a no, so I figured I'd ignore parity and just memorise all PLLs, including parity. Still a no go (547 algs. Possible, but probably not worth it). I think the best way would just to do the two faces separately, like you did (I have no idea why I didn't think of that at the time lol). Still pretty garbage due to the move count and the rotation at the end, but better than the version you chose ( I think? I'm working on sub 20, so I'm not exactly experienced in this kind of stuff yet).
This is legitimately the first time I've heard about this for 3x3! I like using this for weird non shapeshifting 3x3xn cuboids and I'd also imagine this could work well for other 3x3xn cuboids where n is an odd integer greater than 3.
Here's a way I found to do it Build belt (Fix oll parity, F perm or E perm if only corners) Solve both oll Move edges to correct side Commutate the corners loke for 5x5 centers (Fix pll parity, M2 U2 M2) Solve both pll
this is less efficient but it works i guess . 1. make a cross dosent have to be correct but make a cross. 2. orientate corners 3. make belt boom! no oll parity 4. oll 5. pll ,sometimes pll parity just use 4x4 algorithm for pll parity 6. flip 7. oll 8.pll 9.finished
I've been speedcubing 2 years and have actually never thought of the Belt method. I was expecting this to be about the CFOP-Roux Hybrid method that lots of people come up with.
I remember thinking about a similar way to solve it, and thought it would be good, but for a different reason: 1. Belt 2. OBL (Orientation of Both Layers) 3. Top and bottom face separation (if not color neutral, yellow and white) 4. PBL (Permutation of Both Layer) The goal was to save time in bottom layer and permute everything at the end with only one algorithm, like ortega in 2x2 (which might give better algs than normal PLL for some cases (ex opposite J-Perms (R2 U R2 U' D' R2 D R2)). The problem is that it would need about 400 algs for PBL (no idea for OBL) and I don't think it would even be faster.
I feel like doing EO before the belt could make things more efficient. They maybe you could combine some steps. Something like: 1. EO 2. Belt 3. Separate pieces (edges are easy to recognize when oriented so it might be easier to combine edge and corner separation) 4. Corner parity 5. COLL #1 6. COLL #2 7. Permute edges
Optimized Belt Method: 1) White Cross (doesn't have to be aligned) 2) Belt (leave 1 piece out) 3) White Corners (4th corner paired w/last belt piece) 4) OLL Yellow 5) PLL Yellow (if not real PLL, use M2 U2 M2 to fix it) 6) PLL White (no OLL because step 1 and 3)
Yes, i think we've all tried it. Good practice for OLL/PLL. Cause you have to do it twice. Bad idea when you realize you messed up and have to do much more than you originally planned when even though the pieces are separated, they are still in a configuration that can't be solved without fixing the cube. Really good if you want to do a bunch a u perms.
Not a bad idea, I'd do it like this: 1) Put all the white pieces (randomly - non-permuted and non-oriented) on the top layer 2) Build the belt 3) Now we should understand if we can fix the eventual parity 4) oll and pll white pieces (top layer) 5) oll and pll yellow pieces (lower layer)
There's a WAY better method than CFOP called ZB. You first do F2L up to your last pair but then use an algorithm to orient all of the last layer edges. After that, you use a ZBLL algorithm to solve the rest. If there wasn't as many algorithms, would you recommend this method or is it still not worth learning? Also, why do the few cubers who use this method get slower times than any top speedcubers (like Feliks)? btw amazing content
@@abstractcuber236 I wasn't thinking of actually learning it but I was just curious about why it was slightly slower. But anyway thanks for the explanation : )
While you make the belt you can try to do piece/edge separation at the same time to save time, like I think it could be possible to get fast at this method if you blend some of the steps together like the first 2 Edit: If you use Roux like me, you can use on of the triggers M' U2 M to swap edges, I think the Belt method could work with roux solvers
Step 1: solve the belt(7 or less moves max) step 2. Orient all the edges(max of like 10 moves) step 3: solve the f2l pairs(rotationless) step 4: solve the 4 cross edges(rotationless) step 5:ZBLL(or last layer)
I also have created a method: the Minus2Y method Step 1: Make a line just as you make the cross on CPOF/Beggibers method, but instead of a cross it is a line or a minus Step 2: Make the 2 blocks next to the line's centers like in the Roux method matching with the center's colour Step 3: OLL and PLL Conclusion: this mixes roux and CPOF (And maybe ZZ?)
I actually solved my old 5x5 for years using a sort of layer by layer method. Do the first 3 layers and then the bottom ring of edges and then the final 4th layer edges. Finally, do the remaining center pieces. It was slow, but can be adjusted to work with any size cube
Lol, I really thought about this method as well and initially thought the only extra step would be the white/yellow sorting. Luckily while solving 2x2 with Ortegra I noticed that I mostly get invalid OLL cases for the top face when the bottom face is not oriented. So I didn't bother trying "Belt" ;)
What about the instead of the cross, you do an X? Corners first, if advanced, do 2nd layer edges, then do 1st layer edges, or do 1st layer edges, then 2nd layer edges, then do last layer however you want?
Just suggestion, how about doing "oll" on both top and bottom and use 5x5 center building to complete the faces. Then do E perm if both top and bottom have adj. corner swaps by moving the corners to be swapped from the bottom to the top. Then do EPLL and parity (if any).
I have a possibly sightly better way of doing the method: 1: EO-Belt. (Do edge orientation and the belt) 2: Piece seperation. (to keep EO use only R2, U & D moves. My method is to first have a 1x1x2 block (1 edge and center) of the same colour on top, then insert one 'pair' to make it a 1x2x2 then another 'pair' to make a 1x2x3. Finally you pair up the last two corners and edge then insert to finish separation. (if needed use R2 U2 R2 U2 R2 to flip R2 layer).) 3: OLL on top with one corner ignored. 4: PLL parity. (M2 U2 M2) 5: PLL on top 6: Flip cube, Then use EJLS to twist the last corner and finish OLL at the same time. (If the corner isn't twisted you could use ZBLL to skip step 7) 7: PLL to finish the cube.
Hi JPerm i have just started watching your videos about 2 weeks ago and your videos have really insiperd me to get a runik's cube myself! I just wanna say thank you for the amazing content you give us
Actually we can make this method more efficient by reducing OLL paritys. Bassicly, when we put wrong for example white edge on the bottom layer, we can determine if it should be white in front or bottom. Same with corners, we can use "edge on place" F2L (only 3 algs, not that much) to insert pieces in correct rotation. Then the only parity left is PLL parity which can be done in 3 moves.
i personally like to make a pll case on the bottom by solving just the side, then i do the belt with no problem, i do the oll and pll at the top, and if there is a parity then i keep moving anyways, i do the pll case at the bottom, and then i solve the parity there (if there is), then i do the top, and it is solved
@@neezduts6933 oh yeah that's actually kinda efficient now that I think about it, you just do a pll on top of belt and flip the cube 180 to do pll on the layer on bottom since the bottom isn't aligned into a cross?
I came up with a method that is the same as belt. 1. white face (no need for correct orientation 2. Belt 3. permute white 5. parity (If needed and there is only on case and can be solved by M2 D2 M2) 6. just do the yellow normally.
I wonder if you could do it this way: 1. EO + Belt + Corner separation, and the corners at the bottom are all white and are a valid CLL case. To simplify it, make adjacent corner pairs that are either "lizard eyes" (as in the T case), solved or headlights. 2. Edge separation. If the corners aren't solved then make an invalid edge permutation at the bottom and if the corners are solved then make a valid edge permutation on the bottom (U, Z, H perm etc) 3. ZBLL both sides. There should be no parities of any kind This should work but at this point just solve with CFOP or even with ZZ
Lmao I also thought I invented something similar🤣. Though my pattern is a bit different since im not a speed cuber and not for the purpose of speedcubing. My pattern is to build a white cross then just put white pieces randomly in that cross. Build the second layer, then third layer, adjust things that needed to adjust then go back to the first layer and do the third layer thing🤣. Its fun making your own formula🤣.
There's an alternate method you could use. The pseudocuboid method 1. Solve belt. 2. Orient edges on both sides. The edge parity is solved by turning one of the belt centers with an unsolved edge 180 degrees. 3. Orient corners on both layers. Same as edge parity. 4. Solve corners like you would on a 2x2x3. Hope you know your cuboids. 5. Put white edges on white side and yellow edges on yellow side. 6. Permute edges.
I actually got my PB with the belt method by trying it right now for the first time after watching your video haha. I use layer by layer with algs invented by me and my bestie and average around 40 seconds, and the belt gave me a 25 seconds time smashing my old PB by 10 seconds
I actually tried to improve on this method, and here is my current revision. Step 1: Belt You know how this goes. Step 2: Parity Swap First, you need to learn the swapping algorithms. Edge Swap - M U2 M’ This swaps the UF and DF edges. Corner Swap - Sledgehammer x3 This swaps the URF and DRF corners. Back to step 2. Ignore the fact that yellow and white are different colors. Are there an odd number of bad edges on the top layer? Just swap one of the good edges with a bad one from the bottom. Similar process for corners. Step 3: OFLL Do OLL on the top and bottom, still ignoring the fact that yellow and white are different. Step 4: Color Swap Unfortunately we can’t ignore the colors forever, so this is the step we solve the colors. It should be obvious how to do this step. Step 5: PFLL Do PLL on the top and bottom. If you have parity, do M2 U2 M2.
I made a belt method for beginner method. 1. Midldle layer. 2. white cross. 3. yellow cross. 4. yellow edges 5. white and yellow corner seperation. 6. permutate first layer 7. permutate last layer 8. orient first layer 9. orient last layer 10. orientation parity
OMG J PERM!! Its been sooooooooooooooo----- long since u uploaded your video, I was waiting for this. The only entertainment on youtube for me is your channel. I have a GAN 356 RS and I want to magnetize it, but its too hard.
Huh not me though i have invented 2 methods of my own. Method-1: solve all the edges and then all the corners. Method-2: solve all the corners and then all the edges
Jperm, I have a really weird way of solving the cube, first make the first two blocks in roux, then do cmll, then position the edge pieces that matches the colors of the first two blocks, then solve it with cfop
Here is a similar method that might work better: 1) belt 2) oll on both sides. treat yellow and white like the same color, and just do single side roatations for the parities 3) now you effectively have a square-1. Solve one of the layers (yellow or white) intuitively, but only rotate one side 4) if you performed an odd number of rotations in the last step (two belt pieces swapped), fix it 5) oll and pll on the top layer
1) So far fine 2) I guess we can ignore parity by treating edge parity as solved and corner parity as a solvable orientation. Then we can have 2 flipped edges and 2 twisted corners which can be solved 3) This is called domino reduction and kociemba's algorithm and is used for computers to solve cubes. We should solve the bottom side 4) R2 U2 R2 U2 R2 U2 5) Since yellow and white are all facing up, no need for OLL Working but quite impractical
J Perm, I know that you said that in this method you can get 9 things that make this method bad, but there is actually a way of doing this method and only getting 1 PARITY, this is how it works: 1st: You do a face on the cube, not a layer, just a face. 2nd: You do the belt, not necessarily with F2l, just with the algorithm of putting edges on the beginners method. 3rd: You do Oll, and there is no parity here because the opposite face is solved, and all of the edges and corners are orientated correctly so that you can do Oll with no worries. 4th: Do Pll on the two opposite faces. 5th: If there is parity on Pll, just do the algorithm: M2, U2, M2, that will fix the two pll's left. ( Do this step only if you get an Pll case that seems impossible. ) I hope that this was helpful for anyone using this method. Bye!
I invented the belt method too, although slightly different. I first made the cross without worrying about orientation, then I paired up the belt edge with any bottom layer corner piece in any orientation and inserted it. Then I dealt with parity and then oll/pll. I then improved it by making the cross with 0, 2 or 4 edges oriented correctly which saved the parity later. It was still bad but the steps of putting in cross pieces first in a good orientation and putting corners on the correct layer when doing the belt pieces makes in immensly better. You don't need to worry about edge flip or corner twist parity with my improved version.
For a challenge vid, can you speedsolve with an FMC solution. So basically, you get like 45 minutes or something to come up with a solution for a scramble in CsTimer, then with that solution, you time yourself solving it with the solution you made as fast as you can. PS: Love your vids!
I tried it. I found out, that if you make 1 side (doesn’t have to be a layer), the belt, then you do pll, if you don’t get lucky I think there’s only edge parity to fix, which can be done with slice and down moves. Then you just solve the other side, most likely without problems. That’s max 6 steps (I think) instead of the 9 steps you show us in the video. Also, you don’t have to do a layer, you can just try to set up an oll case, but that may still lead to edge parity (but idk if it does), and if you just seperate and get in the pieces on their side, but I don’t know why anyone would just randomly i they’re trying on time, it could add 1 step (corner parity fixing) I think it’s fun. You don’t have to like it, but you can’t stop me from liking it.
I Never actually thought of doing this. I thought about doing the First 2 layers differently. CFOP: Make a cross and then put in ALL of the pairs Roux: Make 2 cross edges and the pairs and do the last 2 cross-edges at the end Petrus: Make 2 adjacent cross edges and put in the pair between them and repeat that 3 more times. All of these work. Just they are very strict. So I combined Petrus F2L with Cfop F2l. So I can put in any of the pairs and any of the cross edges in any order. I went from 25avg to a sub21 average. People also say that Petrus is a bad method. I think that it is very good. Unlike Cfop, you don't have to do the crossedges first. But it is also easier to do than Roux Blocks. This is just my opinion and Petrus Helped me To Sub21
Are these worse? Ga: M2 U2 M2 U' (Jb-perm) U' M2 U2 M2 *Block on the right Gb: M2 U2 M2 (Jb-perm) M2 U2 M2 *Block on the left Gc: M2 U2 M2 U (Ja-perm only left hand) U M2 U2 M2 *Block on the left Gd: M2 U2 M2 (Ja-perm only left hand) M2 U2 M2 *Block on the right
It attual work only slightly chane 1 Slove one side any colour 2 face the oppoise colour 3 Make a belt as you will slove 2 nd layer by beginer method 4 do oll and pll 5 face white and do pll cause oll is already sloved
As a great man once said “Solutions that are the first thing you think of, and look sensible and are easy to implement. Are often terrible, ineffective solutions that once implemented, will drag on civilization forever”
I thought of this message where you first solve the corners like 2x2x2 and the solve the edges and then put the centres in place...takes a hell lot of time... especially when compared to CFOP or Belt. But it was worth it
Hey! New cuber (and subscriber) here (5 days in, making good progress). I was researching advanced methods and came across a newer method that looks to be an improved version of this belt method. It's called the "Mehta" method. It looks like it was first proposed in 2020, so it's relatively new. But seems to combine some Roux, belt, and CFOP methods. Supposedly it has low move count and mostly relies on algorithms. Would you consider doing a video over this method? I think it would be pretty cool and it seems many think it could be a new promising speedcubing method. I'd love to know your thoughts on it vs CFOP and ROUX.
PBL Method overview: Face-1: Make a face -1 corner on any color, not necessarily making a layer (if you make a layer just use 8355💀💀💀) Belt: Use the unsolved corner as a keyhole to insert all belt edges. for the last edge, act as if it was an f2l case and solve it with the last corner OLL: identify oll on top and do the alg, you won't get any parities until pbl 2-Look PBL: (side note: 2 look pbl is better than one look pbl because its more intuitive and easier to learn) solve the corners of both layers using an algorithm. There are 3 cases for cpbl, adj/adj, opp/opp, and adj/opp. i only know a good alg for adj/adj, which is just the bar/bar 2x2 pbl (R2 U' B2 U2 R2 U' R2) 3 times, but the other cpbls must be generated with a software like cube explorer to get an efficient alg. Now, solve the edges by combining 2 triggers: swap 2 edges oppositely on both layers (M Parity): M2 U2 M2 U2 (extra U2 to bring you back to original pos) swap 2 adjacent edges on both layers (S Parity) (top will swap back left, bottom will swap front right): S' M' U M U2 S U for the u perms, you'd want to set these up to M Parity, but sometimes you will get an adjacent front right on top and an opposite front back on bottom, in that case, do R U R' U' (M parity) U R U' R' now packed with this information, you should be able to solve EPBL intuitively with a bit of practice, pro tip: DONT USE THIS
Instead of doing M’ U2 M you could also use U M’ U’ M for last edge that you have to separate. The edge will go in at different orientations. For M’ U2 M the D sticker goes to Q, for U M’ U’ M the D sticker goes to K.
I've have never thought of this.... maybe I'm stupid. Also is 28 seconds and an average of 38 after 3 and a half weeks since first learning to cube a good time. If so, then J Perm is even more awesome at cubing and teaching than I thought he was, because he is the only I have watched.
![The BEST Speedcubing Method? [CFOP/ROUX/ZZ] Comparison](/img/n.gif)
When you realise everyone has but you have never thought of the belt method
Here is my bro
Totally relatable
Cubefinity same. Although I have seen it on the speedsolving wiki
@@goldenwarrior1186 r u srs
I actually thought of this method and already knew the problems I was going to have...the parities...hence I never tried it.
RedRobbie7 Yes
Tymon Kolasinski tried the belt method during Cyoubx's stream yesterday ("Sub-6's on Stream EZ. Learning Pseudoslotting and CN"). He modified the method slightly - he forced a parity skip during piece separation and rearranged the steps. He also got a 14 and then called a 15 "bad".
Tymon is the type of guy that would get 15 seconds on belt method
Advanced method:
Step 1: white/ yellow manipulation ( on correct sides, no parites )
Step 2: belt
Step 3: OBL (Orient both layers)
Step 4: PBL ( Permute Both Layers)
this is not a square 1
WBOP
nice name
ortega moment
WYBOP
this is square 1 but for 3x3 lol
So, I have boarded the dummy train and I have tried to optimize this method the best I can with my limited knowledge.
First start with ZZ's EO, this way you don't have to deal with the pain of edge parity later.
Then solve the belt with moves only. Which is very easy actually.
Then separate edges and corners using bars. Make bars by placing an edge between two adjacent corners on the opposite side using M' U2 M. Place the top bar on the left and bottom bar on the right and do R2 U2 R2. If you have an odd amount of edges or corners, solve using bars until your at one corner or one edge on each side that needs to swap. To swap one corner on each side, place one corner on top of the other and do (U R U' R')3. to swap one edge on each side, place one edge on top of the other and do M' U2 M.
From there do OLL and PLL, doing corner twist parity as needed, flip over, OLL, PLL, finish!
I feel like this would be the method you’d teach someone who for some reason knows how to solve a square one but not a normal cube
Yeah, though only squan corner algs would work, and you’d have to orient the corners first. Edges work a bit different because of the way it slices on squan.
Try Shadowslice Snow Columns. It solves the E slice and corner orientation at the same time, then solves corners with Square 1 algs.
Keyhole method is fire for noobs🔥
I learned how to solve the ghost cube before the 3x3 lol
yes
5:41 "+2s don't count at home" - Kevin Hays
Kevin Hays ❌
Everyone Says ✅
When I “invented” this method, I actually realized this potential flaw, so I actually would solve the white side, similar to F2L, but only the edges for the belt were in the right spot. This way, I did OLL once on the yellow side (solving mismatch, edge and corner parities) and the would only do PLL on both sides. I still think it’s decently viable but CFOP is just better in every way
Very interesting. Before I ever watched a single tutorial on solving, I intuitively created a belt around the E axis. I still do a solve like this occasionally.
Oh god yes this is just nostalgic and hours of trying to perfect this
Yeah my parents would always use this method on me
Its so nostalgic
@@xcalivrr3982 exactly
@@xcalivrr3982 u won
You may be able to make separating pieces easier if you don’t preserve the belt, but only require those 4 edges stay in the middle. One algorithm at the end could fix it easily or be combined with PLL to make it one step (I think this would triple the number of PLL’s)
Stupid cubing method: *Exists*
JPerm: Its free real estate
*its free real estate*
.
such an original comment....
@@pk251 lol i know
@@pk251 well I mean at least it isn't so overused, not like those annoying "nobody:" or "20 likes 0 views" comments
My variation on this was to build the top, and bottom faces, then the belt. That way, you don't have to worry about piece separation. You can also build the first face and the belt at the same time, since it's sorta like F2L.
The major flaw in this method came in the PLLS. Originally, I thought I'd just fix the parity with M2 U2 M2, then (I assumed) I could memorise all the algs to solve both faces at once. That was a no, so I figured I'd ignore parity and just memorise all PLLs, including parity. Still a no go (547 algs. Possible, but probably not worth it). I think the best way would just to do the two faces separately, like you did (I have no idea why I didn't think of that at the time lol).
Still pretty garbage due to the move count and the rotation at the end, but better than the version you chose ( I think? I'm working on sub 20, so I'm not exactly experienced in this kind of stuff yet).
This is the Exact Method my Mom uses to Beat me.
Lmfaoo
😂
I had made a method by myself I do f2l first and then cross and oll pll
@@ninjasparkzz5097 i Dont know this method, i only know cfop (jk)
Rip
1:eo line 2: belt 3 OYE orient yellow/white edge 4: EFP edge flip parity 5: OWE orient white/ yellow corner 6 OLL 7 PLL
This is legitimately the first time I've heard about this for 3x3! I like using this for weird non shapeshifting 3x3xn cuboids and I'd also imagine this could work well for other 3x3xn cuboids where n is an odd integer greater than 3.
If I'm not mistaken, this is like solving a 3x3 as a square-1. Square-1 uses this method.
Nobody:
J perm: solves cube faster then me with the belt method
@Your Mom same
Than* and same
Not me i average 30 seconds feels good to know im not the slowest cuber ever
@@jaydonturner2367 and he averaged around 20.
LOL
Here's a way I found to do it
Build belt
(Fix oll parity, F perm or E perm if only corners)
Solve both oll
Move edges to correct side
Commutate the corners loke for 5x5 centers
(Fix pll parity, M2 U2 M2)
Solve both pll
I never once thought of trying to solve a cube this way.
Same
me2
Me 🌳
same
Yeah who would?
this is less efficient but it works i guess . 1. make a cross dosent have to be correct but make a cross. 2. orientate corners
3. make belt boom! no oll parity 4. oll 5. pll ,sometimes pll parity just use 4x4 algorithm for pll parity 6. flip 7. oll 8.pll 9.finished
1 month later: how do get sub 10 with the belt method
Underrated
No, more like how to get sub 0.000001 with belt method
Sub 10 half a MINUTE
Was bout to write this
@@hvrst6241 How to get WR with belt
I've been speedcubing 2 years and have actually never thought of the Belt method. I was expecting this to be about the CFOP-Roux Hybrid method that lots of people come up with.
Meanwhile , in an alternative universe: uses belt method to set wr
I mean, if SSC counts as belt method, it might be viable.
i thought advanced cross was tough, but after trying to solve the belt a few times, i realized how much easier the cross was
When Dylan’s belt method pb is better than my regular pb😂
More, or better?
Yup
same
My pb is 17
My pb is 19
My pb is 24 he beat me by 8 seconds
4×4×4 cube
1. 6 centers
2. belt
3. separate pieces
4. edge flip parity
5. corner rotation
6. OLL(T)
7. PLL parity
8. PLL(T)
9. OLL(B)
10. PLL(B)
"back in the day of the speedsolving forums"
The forums still exist and people like me are very active on it.
oh my god are you itsowen? what's up dude i'm pussyslayer68 i just got 1k posts
I am also a very active member of the forums, but I don’t think that Owen will figure it out before he overtakes Cuberstache in mega 🤪
@@a1b3a3c14nbcv no, I am Owen Morrison on the forums.
@@fatherfilth3784 haha😆. Who are you on the forums?
That’s for you to figure out
BLBL (Belt Layer by layer)
1. Make the belt (obviously)
2.fix parity (by repurposing algs)
3.swap corner
4.layer by layer oll and pll
I remember thinking about a similar way to solve it, and thought it would be good, but for a different reason:
1. Belt
2. OBL (Orientation of Both Layers)
3. Top and bottom face separation (if not color neutral, yellow and white)
4. PBL (Permutation of Both Layer)
The goal was to save time in bottom layer and permute everything at the end with only one algorithm, like ortega in 2x2 (which might give better algs than normal PLL for some cases (ex opposite J-Perms (R2 U R2 U' D' R2 D R2)). The problem is that it would need about 400 algs for PBL (no idea for OBL) and I don't think it would even be faster.
3249 for OBL, 441 for PBL
1st OLL algs would be a lot easier because the bottom pieces dont matter.
I feel like doing EO before the belt could make things more efficient. They maybe you could combine some steps. Something like:
1. EO
2. Belt
3. Separate pieces (edges are easy to recognize when oriented so it might be easier to combine edge and corner separation)
4. Corner parity
5. COLL #1
6. COLL #2
7. Permute edges
This actually sounds really obscure? I'm surprised that this many cubers thought of doing it.
Optimized Belt Method:
1) White Cross (doesn't have to be aligned)
2) Belt (leave 1 piece out)
3) White Corners (4th corner paired w/last belt piece)
4) OLL Yellow
5) PLL Yellow (if not real PLL, use M2 U2 M2 to fix it)
6) PLL White (no OLL because step 1 and 3)
Imma try this
TITLE: The Speedcubing Method We All Invented...
Me seeing the video: I have never seen or thought of that method in my life
Yes, i think we've all tried it. Good practice for OLL/PLL. Cause you have to do it twice.
Bad idea when you realize you messed up and have to do much more than you originally planned when even though the pieces are separated, they are still in a configuration that can't be solved without fixing the cube. Really good if you want to do a bunch a u perms.
I have a feeling this involves the Middle Layers a lot.
pp
@@rexyaxy4314 stop spamming
Equator layers
@@cubebutpro298Slice moves (M, E, and S)
Not a bad idea, I'd do it like this:
1) Put all the white pieces (randomly - non-permuted and non-oriented) on the top layer
2) Build the belt
3) Now we should understand if we can fix the eventual parity
4) oll and pll white pieces (top layer)
5) oll and pll yellow pieces (lower layer)
There's a WAY better method than CFOP called ZB. You first do F2L up to your last pair but then use an algorithm
to orient all of the last layer edges. After that, you use a ZBLL algorithm to solve the rest. If there wasn't as many
algorithms, would you recommend this method or is it still not worth learning? Also, why do the few cubers who use
this method get slower times than any top speedcubers (like Feliks)?
btw amazing content
Same
k
@@abstractcuber236 I wasn't thinking of actually learning it but I was just curious about why it was slightly slower. But anyway thanks for the explanation : )
When He's FASTER Than You With BELT METHOD
finally a method where j perm's times are slower than my normal solves lmao
While you make the belt you can try to do piece/edge separation at the same time to save time, like I think it could be possible to get fast at this method if you blend some of the steps together like the first 2 Edit: If you use Roux like me, you can use on of the triggers M' U2 M to swap edges, I think the Belt method could work with roux solvers
4:31 Finally faster than J Perm
Method idea:
1: First 4 Pairs
2: CMLL
3: Edge seperation
4: Solve the rest of edges
4a. Fix Edge Parity
4b. ELL (Top)
4c. ELL (Bottom)
I literally invented this a week ago. J perm is spying on me
Step 1: solve the belt(7 or less moves max) step 2. Orient all the edges(max of like 10 moves) step 3: solve the f2l pairs(rotationless) step 4: solve the 4 cross edges(rotationless) step 5:ZBLL(or last layer)
When he solves faster with belt than you do normally
I also have created a method: the Minus2Y method
Step 1: Make a line just as you make the cross on CPOF/Beggibers method, but instead of a cross it is a line or a minus
Step 2: Make the 2 blocks next to the line's centers like in the Roux method matching with the center's colour
Step 3: OLL and PLL
Conclusion: this mixes roux and CPOF (And maybe ZZ?)
I just realized he solves faster than me in the Belt method than I can in Cfop LOL.
(This is what I get for taking a 2 year break from cubing 🤦♂️)
same
same I took a break for like a year
Jokes on you I got a 7 year break
@@wernerziegler2873 I haven't even cubed even before this comment but sheesh
I actually solved my old 5x5 for years using a sort of layer by layer method. Do the first 3 layers and then the bottom ring of edges and then the final 4th layer edges. Finally, do the remaining center pieces. It was slow, but can be adjusted to work with any size cube
Lol, I really thought about this method as well and initially thought the only extra step would be the white/yellow sorting. Luckily while solving 2x2 with Ortegra I noticed that I mostly get invalid OLL cases for the top face when the bottom face is not oriented. So I didn't bother trying "Belt" ;)
What about the instead of the cross, you do an X? Corners first, if advanced, do 2nd layer edges, then do 1st layer edges, or do 1st layer edges, then 2nd layer edges, then do last layer however you want?
Jperm: gets low 20s using the belt method
Me: gets low 30s using intermediate cfop
Me again: quits cubing
dont give up really
Forgive me for asking but what's intermediate CFOP?? I know what beginning and advanced are but what do you mean by intermediate?
Siri M he means he’s not advanced but he’s not beginner
Siri M difficulty easy=beginners medium=intermediate hard=advanced
Mekenzie Brown that is so me
Just suggestion, how about doing "oll" on both top and bottom and use 5x5 center building to complete the faces. Then do E perm if both top and bottom have adj. corner swaps by moving the corners to be swapped from the bottom to the top. Then do EPLL and parity (if any).
I never had such a problem with CFOP that I needed to "invent" a method it was difficult but I learned
I have a possibly sightly better way of doing the method:
1: EO-Belt. (Do edge orientation and the belt)
2: Piece seperation. (to keep EO use only R2, U & D moves. My method is to first have a 1x1x2 block (1 edge and center) of the same colour on top, then insert one 'pair' to make it a 1x2x2 then another 'pair' to make a 1x2x3. Finally you pair up the last two corners and edge then insert to finish separation. (if needed use R2 U2 R2 U2 R2 to flip R2 layer).)
3: OLL on top with one corner ignored.
4: PLL parity. (M2 U2 M2)
5: PLL on top
6: Flip cube, Then use EJLS to twist the last corner and finish OLL at the same time. (If the corner isn't twisted you could use ZBLL to skip step 7)
7: PLL to finish the cube.
How many of you keep touching your cubes in during online classes?
Me
He's right, at first I was like "yknow, that actually sounds like a good, fast method"
then i was like "oh"
Some guy: Who exactly invented the Belt method?
Me: *It's hard to say who exactly...*
Hi JPerm i have just started watching your videos about 2 weeks ago and your videos have really insiperd me to get a runik's cube myself! I just wanna say thank you for the amazing content you give us
Update?
Actually we can make this method more efficient by reducing OLL paritys. Bassicly, when we put wrong for example white edge on the bottom layer, we can determine if it should be white in front or bottom. Same with corners, we can use "edge on place" F2L (only 3 algs, not that much) to insert pieces in correct rotation. Then the only parity left is PLL parity which can be done in 3 moves.
i personally like to make a pll case on the bottom by solving just the side, then i do the belt with no problem, i do the oll and pll at the top, and if there is a parity then i keep moving anyways, i do the pll case at the bottom, and then i solve the parity there (if there is), then i do the top, and it is solved
@@neezduts6933 how do u not break the side when doing belt?
@@floseatyard8063 i do f2l pretty much
@@neezduts6933 oh yeah that's actually kinda efficient now that I think about it, you just do a pll on top of belt and flip the cube 180 to do pll on the layer on bottom since the bottom isn't aligned into a cross?
I came up with a method that is the same as belt. 1. white face (no need for correct orientation 2. Belt 3. permute white 5. parity (If needed and there is only on case and can be solved by M2 D2 M2) 6. just do the yellow normally.
Can you solve a 3x3... Like a square-1?
I wonder if you could do it this way:
1. EO + Belt + Corner separation, and the corners at the bottom are all white and are a valid CLL case. To simplify it, make adjacent corner pairs that are either "lizard eyes" (as in the T case), solved or headlights.
2. Edge separation. If the corners aren't solved then make an invalid edge permutation at the bottom and if the corners are solved then make a valid edge permutation on the bottom (U, Z, H perm etc)
3. ZBLL both sides. There should be no parities of any kind
This should work but at this point just solve with CFOP or even with ZZ
Lmao I also thought I invented something similar🤣. Though my pattern is a bit different since im not a speed cuber and not for the purpose of speedcubing.
My pattern is to build a white cross then just put white pieces randomly in that cross. Build the second layer, then third layer, adjust things that needed to adjust then go back to the first layer and do the third layer thing🤣.
Its fun making your own formula🤣.
very fun method
There's an alternate method you could use. The pseudocuboid method
1. Solve belt.
2. Orient edges on both sides. The edge parity is solved by turning one of the belt centers with an unsolved edge 180 degrees.
3. Orient corners on both layers. Same as edge parity.
4. Solve corners like you would on a 2x2x3. Hope you know your cuboids.
5. Put white edges on white side and yellow edges on yellow side.
6. Permute edges.
Am I the only one who never thought of this method?
I actually got my PB with the belt method by trying it right now for the first time after watching your video haha. I use layer by layer with algs invented by me and my bestie and average around 40 seconds, and the belt gave me a 25 seconds time smashing my old PB by 10 seconds
J-perm: here's a method we all invented
Me who used roux and has no idea how to do oll and pll: no
I actually tried to improve on this method, and here is my current revision.
Step 1: Belt
You know how this goes.
Step 2: Parity Swap
First, you need to learn the swapping algorithms.
Edge Swap - M U2 M’
This swaps the UF and DF edges.
Corner Swap - Sledgehammer x3
This swaps the URF and DRF corners.
Back to step 2. Ignore the fact that yellow and white are different colors. Are there an odd number of bad edges on the top layer? Just swap one of the good edges with a bad one from the bottom. Similar process for corners.
Step 3: OFLL
Do OLL on the top and bottom, still ignoring the fact that yellow and white are different.
Step 4: Color Swap
Unfortunately we can’t ignore the colors forever, so this is the step we solve the colors. It should be obvious how to do this step.
Step 5: PFLL
Do PLL on the top and bottom. If you have parity, do M2 U2 M2.
51 secs ago!
Last time I was this early, I was allowed to leave the house !
I made a belt method for beginner method.
1. Midldle layer.
2. white cross.
3. yellow cross.
4. yellow edges
5. white and yellow corner seperation.
6. permutate first layer
7. permutate last layer
8. orient first layer
9. orient last layer
10. orientation parity
lmao I remember "inventing" this when I was first getting into speedcubing
"Turns out, during inspection, I have a tough time planning the belt"
that hit so hard
4:30 that is literally my Pb on 3x3
OMG J PERM!! Its been sooooooooooooooo----- long since u uploaded your video, I was waiting for this. The only entertainment on
youtube for me is your channel. I have a GAN 356 RS and I want to magnetize it, but its too hard.
Huh not me though i have invented 2 methods of my own.
Method-1: solve all the edges and then all the corners.
Method-2: solve all the corners and then all the edges
isnt thats just a method for bllindfolded?
It is used a few times for blindfold and also blindfold 1hand
@@YesHelloHi nah its just a method of my own and it is not a blindfold method
@@voidman6419 congrats youve literally copied ZLL OH Blind. lol/
Jperm, I have a really weird way of solving the cube, first make the first two blocks in roux, then do cmll, then position the edge pieces that matches the colors of the first two blocks, then solve it with cfop
Couldn't you just do this in Reverse basically were instead you solve the top and bottom layer simultaneously and then put in the belt afterwards?
Roux moment
Here is a similar method that might work better:
1) belt
2) oll on both sides. treat yellow and white like the same color, and just do single side roatations for the parities
3) now you effectively have a square-1. Solve one of the layers (yellow or white) intuitively, but only rotate one side
4) if you performed an odd number of rotations in the last step (two belt pieces swapped), fix it
5) oll and pll on the top layer
1) So far fine
2) I guess we can ignore parity by treating edge parity as solved and corner parity as a solvable orientation. Then we can have 2 flipped edges and 2 twisted corners which can be solved
3) This is called domino reduction and kociemba's algorithm and is used for computers to solve cubes. We should solve the bottom side
4) R2 U2 R2 U2 R2 U2
5) Since yellow and white are all facing up, no need for OLL
Working but quite impractical
I actually did not invent this. F2L is really easy and intuitive.
Lucky
J Perm, I know that you said that in this method you can get 9 things that make this method bad, but there is actually a way of doing this method and only getting 1 PARITY, this is how it works:
1st: You do a face on the cube, not a layer, just a face.
2nd: You do the belt, not necessarily with F2l, just with the algorithm of putting edges on the beginners method.
3rd: You do Oll, and there is no parity here because the opposite face is solved, and all of the edges and corners are orientated correctly so that you can do Oll with no worries.
4th: Do Pll on the two opposite faces.
5th: If there is parity on Pll, just do the algorithm: M2, U2, M2, that will fix the two pll's left. ( Do this step only if you get an Pll case that seems impossible. )
I hope that this was helpful for anyone using this method. Bye!
3:19 it’s just like communism
I was literally about to type this!!! 😮
@@WTF-Cubing lol
I invented the belt method too, although slightly different. I first made the cross without worrying about orientation, then I paired up the belt edge with any bottom layer corner piece in any orientation and inserted it. Then I dealt with parity and then oll/pll. I then improved it by making the cross with 0, 2 or 4 edges oriented correctly which saved the parity later. It was still bad but the steps of putting in cross pieces first in a good orientation and putting corners on the correct layer when doing the belt pieces makes in immensly better. You don't need to worry about edge flip or corner twist parity with my improved version.
I watch your videos u are so cool make more
@@door3545 thanks. My life circumstances are not conducive to making videos now. I don't have the time.
3:04 What loser made a post called "Cat Solves Rubik's Cube In Three Seconds!"
...
...
...
...
...
...
...
wait a minute
I wanna see it now
Jo_ovin I’m not so sure that you do.
Lmaoo
ohh its u
Yes
For a challenge vid, can you speedsolve with an FMC solution. So basically, you get like 45 minutes or something to come up with a solution for a scramble in CsTimer, then with that solution, you time yourself solving it with the solution you made as fast as you can.
PS: Love your vids!
List of step: 1.Belt 1.5.Parity 2.OLL 2.5.Parity 3.PLL 5.OFL 6.PFL (LL is last layer so FL is first layer)
jperm: everybody came up with this idea
Me:never heard of it, never came up with it and heard of it when watching jperm
I tried it. I found out, that if you make 1 side (doesn’t have to be a layer), the belt, then you do pll, if you don’t get lucky I think there’s only edge parity to fix, which can be done with slice and down moves. Then you just solve the other side, most likely without problems. That’s max 6 steps (I think) instead of the 9 steps you show us in the video. Also, you don’t have to do a layer, you can just try to set up an oll case, but that may still lead to edge parity (but idk if it does), and if you just seperate and get in the pieces on their side, but I don’t know why anyone would just randomly i they’re trying on time, it could add 1 step (corner parity fixing)
I think it’s fun. You don’t have to like it, but you can’t stop me from liking it.
Jperm: every speedcuber invented the belt method
Me: *Laughs in beginner*
I Never actually thought of doing this. I thought about doing the First 2 layers differently.
CFOP: Make a cross and then put in ALL of the pairs
Roux: Make 2 cross edges and the pairs and do the last 2 cross-edges at the end
Petrus: Make 2 adjacent cross edges and put in the pair between them and repeat that 3 more times.
All of these work. Just they are very strict. So I combined Petrus F2L with Cfop F2l. So I can put in any of the pairs and any of the cross edges in any order. I went from 25avg to a sub21 average.
People also say that Petrus is a bad method. I think that it is very good. Unlike Cfop, you don't have to do the crossedges first. But it is also easier to do than Roux Blocks. This is just my opinion and Petrus Helped me To Sub21
Are these worse?
Ga:
M2 U2 M2 U' (Jb-perm) U' M2 U2 M2
*Block on the right
Gb:
M2 U2 M2 (Jb-perm) M2 U2 M2
*Block on the left
Gc:
M2 U2 M2 U (Ja-perm only left hand) U M2 U2 M2
*Block on the left
Gd:
M2 U2 M2 (Ja-perm only left hand) M2 U2 M2
*Block on the right
It attual work only slightly chane
1 Slove one side any colour
2 face the oppoise colour
3 Make a belt as you will slove 2 nd layer by beginer method
4 do oll and pll
5 face white and do pll cause oll is already sloved
As a great man once said “Solutions that are the first thing you think of, and look sensible and are easy to implement.
Are often terrible, ineffective solutions that once implemented, will drag on civilization forever”
What cube is the best?
Thanks @J Perm for this method! It's really fun to play with this method lol
I thought of this message where you first solve the corners like 2x2x2 and the solve the edges and then put the centres in place...takes a hell lot of time... especially when compared to CFOP or Belt. But it was worth it
Hey! New cuber (and subscriber) here (5 days in, making good progress). I was researching advanced methods and came across a newer method that looks to be an improved version of this belt method. It's called the "Mehta" method. It looks like it was first proposed in 2020, so it's relatively new. But seems to combine some Roux, belt, and CFOP methods. Supposedly it has low move count and mostly relies on algorithms.
Would you consider doing a video over this method? I think it would be pretty cool and it seems many think it could be a new promising speedcubing method. I'd love to know your thoughts on it vs CFOP and ROUX.
PBL Method overview:
Face-1: Make a face -1 corner on any color, not necessarily making a layer (if you make a layer just use 8355💀💀💀)
Belt: Use the unsolved corner as a keyhole to insert all belt edges. for the last edge, act as if it was an f2l case and solve it with the last corner
OLL: identify oll on top and do the alg, you won't get any parities until pbl
2-Look PBL: (side note: 2 look pbl is better than one look pbl because its more intuitive and easier to learn)
solve the corners of both layers using an algorithm. There are 3 cases for cpbl, adj/adj, opp/opp, and adj/opp. i only know a good alg for adj/adj, which is just the bar/bar 2x2 pbl (R2 U' B2 U2 R2 U' R2) 3 times, but the other cpbls must be generated with a software like cube explorer to get an efficient alg.
Now, solve the edges by combining 2 triggers:
swap 2 edges oppositely on both layers (M Parity): M2 U2 M2 U2 (extra U2 to bring you back to original pos)
swap 2 adjacent edges on both layers (S Parity) (top will swap back left, bottom will swap front right): S' M' U M U2 S U
for the u perms, you'd want to set these up to M Parity, but sometimes you will get an adjacent front right on top and an opposite front back on bottom, in that case, do R U R' U' (M parity) U R U' R'
now packed with this information, you should be able to solve EPBL intuitively with a bit of practice, pro tip: DONT USE THIS
Just solve 1 side (not a layer) then the belt and then oll and pll, turn the cube upside down then oll and pll again
Instead of doing M’ U2 M you could also use U M’ U’ M for last edge that you have to separate. The edge will go in at different orientations. For M’ U2 M the D sticker goes to Q, for U M’ U’ M the D sticker goes to K.
I’m talking about 6:31
I've have never thought of this.... maybe I'm stupid. Also is 28 seconds and an average of 38 after 3 and a half weeks since first learning to cube a good time. If so, then J Perm is even more awesome at cubing and teaching than I thought he was, because he is the only I have watched.