SSC (Shadowslice Snow Columns) 3x3x3 Method

[sqAree]

Example: L2 D B2 F2 U R2 U' L2 U2 F2 U R F U L' B2 L F' D2 B' R'

B2 U' L2 B L' //EO + line
R' // separate edges + RG pair
U2 R2 D// RB triple
R U R2 // OL5C
U D2 R' // CO + belt

Step 2 is 10 moves HTM and 9 moves ATM.

Compare with this:

B2 U' L2 B L' //EO + line
U R2 //RB triple
U D //place DL
R //RG pair
U' R2 //OL5C
U D2 R' //CO + belt

10 moves HTM and 8 moves ATM.

In the first example solve we even get a 1x1x2 skip for U. The second example doesn't really use your approach as the RG edge is hidden in the belt slots until the end.
But I admit there is only a slight difference. I think one can learn your 5 steps and once mastered try to find other ways to deal with certain cases. There will probably be lots of skips and cancellations also. Overall, well done.

Now, my question is how do we generate the algs for CPETL and EPBL? I especially don't want to see S moves in those algs.
But I think even with the original idea of just CPBL --> LEE it's very efficient (and as I pointed out one can always learn a few more CPBL algs to influence LEE). With that I finally decided to try to get fast with SSC now.

I can gen the OL5C algs if you want to.

 

[shadowslice e]

Example: L2 D B2 F2 U R2 U' L2 U2 F2 U R F U L' B2 L F' D2 B' R'

B2 U' L2 B L' //EO + line
R' // separate edges + RG pair
U2 R2 D// RB triple
R U R2 // OL5C
U D2 R' // CO + belt

Step 2 is 10 moves HTM and 9 moves ATM.

Compare with this:

B2 U' L2 B L' //EO + line
U R2 //RB triple
U D //place DL
R //RG pair
U' R2 //OL5C
U D2 R' //CO + belt

10 moves HTM and 8 moves ATM.

In the first example solve we even get a 1x1x2 skip for U. The second example doesn't really use your approach as the RG edge is hidden in the belt slots until the end.
But I admit there is only a slight difference. I think one can learn your 5 steps and once mastered try to find other ways to deal with certain cases. There will probably be lots of skips and cancellations also. Overall, well done.

Yeah that's exactly what I meant and thanks! regarding skips, you actually have a high skip chance (5/9 for 1x1x2, same that one corner will be oriented for the 1x1x3 and a 1/9 chance it will already be done and almost guarenteed that it will be less than 2 moves away.)

Now, my question is how do we generate the algs for CPETL and EPBL? I especially don't want to see S moves in those algs.
But I think even with the original idea of just CPBL --> LEE it's very efficient (and as I pointed out one can always learn a few more CPBL algs to influence LEE). With that I finally decided to try to get fast with SSC now.

Well, I can gen the algs if you want using cube explorer this weekend (fingers crossed). They'll probably be quite intuitive half the time anyway though.

I can gen the OL5C algs if you want to.

that would be great especially if you could match up the cases to algs. Most of them will also likely be guimond algs with a rotation.

 

[sqAree]

Well, I can gen the algs if you want using cube explorer this weekend (fingers crossed). They'll probably be quite intuitive half the time anyway though.


that would be great especially if you could match up the cases to algs. Most of them will also likely be guimond algs with a rotation.

How to tell Cube Explorer to not use S/E moves?

I think it's 81 cases with ~1/3 being mirrors?

 

[shadowslice e]

How to tell Cube Explorer to not use S/E moves?

I think it's 81 cases with ~1/3 being mirrors?

You don't

Just ignore any solutions like that if they come up.

For the step 2 algs, I think there are 16 individual cases that need something like 10 algs

 

[sqAree]

You don't

Just ignore any solutions like that if they come up.

For the step 2 algs, I think there are 16 individual cases that need something like 10 algs

Good luck then, when I tried there was always some S.

For step 2, we assume one edge is already in DL with the two adjacent corners "correctly" oriented and a pair in U. As we can freely AUF there are 3 corners in U with arbitrary orientation and the FRD one (for recognition we don't have to look at BDR), makes 3^4=81 imo.
I know there are two possible way to build a pair in U, so in 1/3 of the cases we have the second type of pair and can use a mirror.
I think I just overlook something but if it's indeed that many cases I'm not sure it's worth.
Recognition is nice though, and the algs I found so far tend to have less moves than 5 if we don't count pre-AUFs.

 

[shadowslice e]

Good luck then, when I tried there was always some S.

For step 2, we assume one edge is already in DL with the two adjacent corners "correctly" oriented and a pair in U. As we can freely AUF there are 3 corners in U with arbitrary orientation and the FRD one (for recognition we don't have to look at BDR), makes 3^4=81 imo.
I know there are two possible way to build a pair in U, so in 1/3 of the cases we have the second type of pair and can use a mirror.
I think I just overlook something but if it's indeed that many cases I'm not sure it's worth.
Recognition is nice though, and the algs I found so far tend to have less moves than 5 if we don't count pre-AUFs.

If you take the cube and rotate so R is on top, you get what is essentially a last slot case meaning that there can't be any more cases than, for example, Winter variation which has 27 iirc. However, it must be lower than this as WV has to play around a pair. I think this means in total there are 23 cases. However, we need fewer algs than this as you can freely ARF (adjustment of right face) to get cases which mean that you can orient all corners so they are facing F/B as opposed to U/D if the alg is nicer and just ARF back.

 

[shadowslice e]

CPETL algs. I'll be updating this as I gen the algs and will try to add a few options for each case. Later I'll porbably change them all to <R,U,r>

Spoiler: Corners solved

1/1, both on front.
M F2 M'
M' U2 M
2/2, adj/adj. F/R
R2 F2 R2 F2 R2 F2
2/2, Adj/Opp, F/R, F/B

2/2 Opp/Opp
M2 U M2
3/3 L/F/R, L/F/R
M' U2 M' U D' M2
4/4
M2 U D' M2

 

[sqAree]

OL5C algs:

http://imgur.com/FoUYLWg

The images show the R face with the top-left corner being RFU. Circles are oriented corners, otherwise CO is indicated by the bars. All of that in respect to the R face, i.e. the circle means the U/D color is pointing to the right.
We assume one of the belt edges in DL with two adjacent accordingly oriented corners and the other belt edge in UF forming a pair with UFL.
Not counting pre-ARFs (adjust R face) the OL5C algs have an average of ~3.9 moves ATM or 4.1 moves HTM.
For recognition it's not needed to look at ULB.
You can also build the pair in U mirrored, in that case just place the edge in UB. It's really easy to mirror those algs, just invert the single moves.

 

[crafto22]

Okay this seems super cool except that I have no idea what's going on. So you do the EOLine on left thing whilst separating the remaining E-slice edges, and then I get lost. How does this work?

 

[wir3sandfir3s]

Okay this seems super cool except that I have no idea what's going on. So you do the EOLine on left thing whilst separating the remaining E-slice edges, and then I get lost. How does this work?

I have a video tutorial on the method, can be found somewhere earlier in the thread (or just search the method in YouTube).

 

[shadowslice e]

Okay this seems super cool except that I have no idea what's going on. So you do the EOLine on left thing whilst separating the remaining E-slice edges, and then I get lost. How does this work?

1) EoEdge
2a) separate FR and BR
2b) Form a 1x1x3 "triplet" with one of the edges
2c) Form a 1x1x2 "pair" with the other one
2d)O5C
2e) finish.

It looks a lot more cumbersome than it actually is because you can combine steps 1&2a, 2b&2c and 2d and 2e quite easily and you get a lot of skips
Example:
Scramble: L2 U2 R2 U2 B2 L D2 L' D2 U2 F2 D' L2 F R' U2 L' U F' U' B'
R' B U' F' L D L'// EoEdge (7/7)
U2//pair+triplet (1/8)
R D R' D' R'// O5C (5/13)
(U' D2) (R/R')// Finish (2/15)

I'll admit I was slightly lucky with the pair/triplet but I did influence 1 corner of the triplet and the others were a 1 in 9 chance of skipping. It was also one of the ever so slightly longer O5C algs too.

A couple more example solves from sqAree just to show the movecount wasn't a total fluke:

Example: L2 D B2 F2 U R2 U' L2 U2 F2 U R F U L' B2 L F' D2 B' R'

B2 U' L2 B L' //EO + line
R' // separate edges + RG pair
U2 R2 D// RB triple
R U R2 // OL5C
U D2 R' // CO + belt

Step 2 is 10 moves HTM and 9 moves ATM.

Compare with this:

B2 U' L2 B L' //EO + line
U R2 //RB triple
U D //place DL
R //RG pair
U' R2 //OL5C
U D2 R' //CO + belt

10 moves HTM and 8 moves ATM.

 

[TyeDye]

I have a video tutorial on the method, can be found somewhere earlier in the thread (or just search the method in YouTube).

Yeah, I understood your version as well, which I'm pretty sure was the absolute basic version of SSC but, I'm pretty sure that this method has changed like six times and I get more and more confused each time.

 

[TyeDye]

1) EoEdge
2a) separate FR and BR
2b) Form a 1x1x3 "triplet" with one of the edges
2c) Form a 1x1x2 "pair" with the other one
2d)O5C
2e) finish.

It looks a lot more cumbersome than it actually is because you can combine steps 1&2a, 2b&2c and 2d and 2e quite easily and you get a lot of skips
Example:
Scramble: L2 U2 R2 U2 B2 L D2 L' D2 U2 F2 D' L2 F R' U2 L' U F' U' B'
R' B U' F' L D L'// EoEdge (7/7)
U2//pair+triplet (1/8)
R D R' D' R'// O5C (5/13)
(U' D2) (R/R')// Finish (2/15)

I'll admit I was slightly lucky with the pair/triplet but I did influence 1 corner of the triplet and the others were a 1 in 9 chance of skipping. It was also one of the ever so slightly longer O5C algs too.

A couple more example solves from sqAree just to show the movecount wasn't a total fluke:

Also, a lot of the posts have a lot of confusing abbreviations and the original solves are still unsolvable and most of the examples aren't explained very well and combine steps so I can't get any idea of what the individual steps are

 

[shadowslice e]

Also, a lot of the posts have a lot of confusing abbreviations and the original solves are still unsolvable and most of the examples are explained and combine steps so I can't get any idea of what the individual steps are

In plain English, the current steps are (assuming green on front, white on top)
1) Orient edges and place the Front-left (orange/green) and Back left (orange/blue) edges

2a) separate the front right (green/red) and back right (blue/red) edges (ie place either the front right (green/red) edge in the U layer and the back right (blue/red) in the D layer or vice versa)
2b) create a "triplet" (a 1x1x3 or column) using the edge in the D-layer (technically a pseudo triplet) of pieces so that if the edge in in the Down Left potition, there would be a white or yellow sticker in the front down left position and a yellow or white sticker in the back down left position. Place this in the down left slot.
2c) create another similar pseudopair (not triplet: this one is a 1x1x2) using the remaining e-slice edge in the U-layer. Place this in the front up slot with the yellowor white sticker on the corner facing to the left (ie being in the left-up-front position).
2d) Look at the right side to determine the O5C (orient 5 corners) case and execute the alg
2e) bring the unsolved e-slice edges to RU and RD using only U and D moves then do an R or R' to solve them.

3) separate the corners: make sure white corners are facing up and yellow corners are facing down. This is similar to the first step of Vandenburgh on square one. You can learn algorithms for it if you want.
4) permute (solve in relation to each other) the corners while separating the edges (so yellow edges face down and white edges face up): this has 48 algorithms and can all be learnt.
5) EZD: permute all of the edges so the cube is completely solved.

 

[TyeDye]

In plain English, the current steps are (assuming green on front, white on top)
1) Orient edges and place the Front-left (orange/green) and Back left (orange/blue) edges

2a) separate the front right (green/red) and back right (blue/red) edges (ie place either the front right (green/red) edge in the U layer and the back right (blue/red) in the D layer or vice versa)
2b) create a "triplet" (a 1x1x3 or column) using the edge in the D-layer (technically a pseudo triplet) of pieces so that if the edge in in the Down Left potition, there would be a white or yellow sticker in the front down left position and a yellow or white sticker in the back down left position. Place this in the down left slot.
2c) create another similar pseudopair (not triplet: this one is a 1x1x2) using the remaining e-slice edge in the U-layer. Place this in the front up slot with the yellowor white sticker on the corner facing to the left (ie being in the left-up-front position).
2d) Look at the right side to determine the O5C (orient 5 corners) case and execute the alg
2e) bring the unsolved e-slice edges to RU and RD using only U and D moves then do an R or R' to solve them.

3) separate the corners: make sure white corners are facing up and yellow corners are facing down. This is similar to the first step of Vandenburgh on square one. You can learn algorithms for it if you want.
4) permute (solve in relation to each other) the corners while separating the edges (so yellow edges face down and white edges face up): this has 48 algorithms and can all be learnt.
5) EZD: permute all of the edges so the cube is completely solved.

Thank you, this makes a lot more sense

 

[crafto22]

In plain English, the current steps are (assuming green on front, white on top)
1) Orient edges and place the Front-left (orange/green) and Back left (orange/blue) edges

2a) separate the front right (green/red) and back right (blue/red) edges (ie place either the front right (green/red) edge in the U layer and the back right (blue/red) in the D layer or vice versa)
2b) create a "triplet" (a 1x1x3 or column) using the edge in the D-layer (technically a pseudo triplet) of pieces so that if the edge in in the Down Left potition, there would be a white or yellow sticker in the front down left position and a yellow or white sticker in the back down left position. Place this in the down left slot.
2c) create another similar pseudopair (not triplet: this one is a 1x1x2) using the remaining e-slice edge in the U-layer. Place this in the front up slot with the yellowor white sticker on the corner facing to the left (ie being in the left-up-front position).
2d) Look at the right side to determine the O5C (orient 5 corners) case and execute the alg
2e) bring the unsolved e-slice edges to RU and RD using only U and D moves then do an R or R' to solve them.

3) separate the corners: make sure white corners are facing up and yellow corners are facing down. This is similar to the first step of Vandenburgh on square one. You can learn algorithms for it if you want.
4) permute (solve in relation to each other) the corners while separating the edges (so yellow edges face down and white edges face up): this has 48 algorithms and can all be learnt.
5) EZD: permute all of the edges so the cube is completely solved.

Okay thanks this makes a lot more sense now.

 

[crafto22]

Hey guys, here are most of the EZD algs I use, I'm finishing up the list right now:
https://docs.google.com/spreadsheets/d/1EHGcr92ksjRZflXDwgBwhA8o04mrGoJShV3bYLzY-gI/edit?usp=sharing

 

[crafto22]

So am I doing this right?
Scramble: U F' B' D' F' R' U2 D F R2 D2 B2 D L2 U R2 F2 U R2 L2
z2 // inspection
R F x M D2 U F' L' F2 // EOLine (8/8)
R2 // Separation (1/9)
U2 R' D2 R2 D // Column things (5/14)
U' R2 // OL5C (2/16)
U' D2 R' // CO + Belt (3/19)
L2 D L2 D R2 // Corner separation (5/24)
some CPETL alg // (~13/37)
M2 U' D' M2 U M2 U D' M2 // A random EZD alg (9/46)

So yeah this method seems really nice but I feel like there's a better way to finish the solve. Notice how we solved up to corner separation with 24 moves, then it takes 22 moves just to finish. This doesn't seem as efficient as it could be.

EDIT: Wouldn't solving CP + DL and DR then go into L6EP be more efficient? That would be like roughly 18 moves, which is about 4 or 5 moves less than doing CPETL then EZD

 

[shadowslice e]

So am I doing this right?
Scramble: U F' B' D' F' R' U2 D F R2 D2 B2 D L2 U R2 F2 U R2 L2
z2 // inspection
R F x M D2 U F' L' F2 // EOLine (8/8)
R2 // Separation (1/9)
U2 R' D2 R2 D // Column things (5/14)
U' R2 // OL5C (2/16)
U' D2 R' // CO + Belt (3/19)
L2 D L2 D R2 // Corner separation (5/24)
some CPETL alg // (~13/37)
M2 U' D' M2 U M2 U D' M2 // A random EZD alg (9/46)

Yep

so yeah this method seems really nice but I feel like there's a better way to finish the solve. Notice how we solved up to corner separation with 24 moves, then it takes 22 moves just to finish. This doesn't seem as efficient as it could be.

EDIT: Wouldn't solving CP + DL and DR then go into L6EP be more efficient? That would be like roughly 18 moves, which is about 4 or 5 moves less than doing CPETL then EZD

That does seem nice but I just feel like it would be hard to recognise the cases. It's definately worth having a look though.

 

[crafto22]

Guys, I think I have an idea to make this more efficient and better for the last few steps. Instead of doing belt + CO then CTL, then CPETL, then EPBL, which ends up being about 21 + 5 + 12 + 11 = 49 or so, why don't we do this:

1. Belt + CO (20 moves)
2. Solve FB in the bottom-left (3 or 4 moves)
3. CP + FD + BD (5 or 6 moves)
4. L11P, which would have 45 cases (~11 moves)

Total: 39 STM