[Help Thread] Kilominx Discussion and Help

[biscuit]

At the Run Module part, I get this:
"
============ RESTART: C:\Users\Christo\Desktop\Kilo\kilosolver.py ============
Traceback (most recent call last):
File "C:\Users\Christo\Desktop\Kilo\kilosolver.py", line 6, in <module>
from PIL import ImageDraw, Image, ImageColor, ImageFont
ImportError: No module named 'PIL'

After that, I can't insert print_move_sequence(solve(random_state()))

Wasn't there a library that had to be downloaded to be able to draw scrambles? It sounds like that's missing.

 

[muchacho]

I ran this command to get rid of that error:
pip install image

 

[DGCubes]

At the Run Module part, I get this:
"
============ RESTART: C:\Users\Christo\Desktop\Kilo\kilosolver.py ============
Traceback (most recent call last):
File "C:\Users\Christo\Desktop\Kilo\kilosolver.py", line 6, in <module>
from PIL import ImageDraw, Image, ImageColor, ImageFont
ImportError: No module named 'PIL'

After that, I can't insert print_move_sequence(solve(random_state()))

You downloaded the version that also gives images of scrambles; not sure if you wanted that one. I'm not particularly well-versed in Python, so I don't know exactly how to fix that problem, but if you download the original kilosolver file and run it as I described, you'll get scrambles without pictures. Your problem has something to do with not having the PIL module, but I'm not sure how you get it or how exactly to get that working.

 

[Solar]

So I got the normal one to work (^-^) but now I need some help with the Image one.

 

[Kit Clement]

So I got the normal one to work (^-^) but now I need some help with the Image one.

Note that to run this, you will need to have the PIL or pillow library installed along with your Python, as this is necessary for image generation.

Use the googles to find how to install Python packages/modules to your interpreter.

EDIT: apparently once you have pip setup, the command "pip install image" works according to @muchacho, but given that my code calls the PIL module, "pip install PIL" would probably be safest. (I'm assuming you're working from a command line, which appears to be the case by your error)

 

[Solar]

And there's the problem, I tired installing PIL (and failed miserably).
Here's the tutorial I used.

I get stuck at 7:32
I insert this:
C:\Users\lolno\Downloads>C:/K/Scripts/pip install "Pillow-3.1.1-cp26-none-win_amd64.whl"
and I get this:
The system cannot find the path specified.
help

EDIT: "K" is my Python folder

 

[Kit Clement]

And there's the problem, I tired installing PIL (and failed miserably).
Here's the tutorial I used.

I get stuck at 7:32
I insert this:
C:\Users\lolno\Downloads>C:/K/Scripts/pip install "Pillow-3.1.1-cp26-none-win_amd64.whl"
and I get this:
The system cannot find the path specified.
help

EDIT: "K" is my Python folder

You need to do this from the folder containing pip in your python folder, which is usually Scripts. Also, I just verified myself that PIL doesn't work from the command line, but "pillow" does. Which is weird to me, since it gets referred to as PIL in the py file.

anyway, this is what I did from command line: (mine is already installed)

You should really find a nice interpreter to get off the command line, it'll make installing packages a lot nicer

 

[Solar]

I did it!

 

[Solar]

There's a chance Kilominx could be an event at Calgary Fall/Autumn 2016

 

[Solar]

I'll be making a video tutorial on this (To show the organizers). You'll all be credited.

 

[WACWCA]

I know there's already a thread debating whether this should be an official event, but we should also have a thread to discuss methods, algs, etc. Cause this is a really fun puzzle and there seems to be plenty of interest right now.

It seems like everyone is using generally the same method, this is what I'm doing:
1. Solve one side/layer intuitively (white)
2. Solve an adjacent layer, three pieces (green)
3. Solve a layer adjacent to both the previous two layers, two pieces (purple). Usually done by creating a two-piece bar and inserting it
4. Continue step three in the same direction until one piece remains (yellow, blue layers)
5. Insert the final non-LL corner (red/pink/cream)
6. OLL (16 cases)
7. PLL (3 cases)

I've been using the same colors, but I can see some variation of color neutrality being viable, at least middle layer color neutrality. Also it seems really structured and I'm sure there's a better way to go about this, so maybe you guys solve differently or have some ideas

And here are my LL algs. I use the first alg in each list currently.

So i thought of this LL method which seems to be between beginners LL and full CLL.
The idea is to orient all of your pieces and also permute two of them so that you would either get the easy PLL or a PLL skip
recognition seems to be easy as it is only recognising where two stickers are in relation to each other. I've found the algs of a few sets and i feel like they're only slightly slower/ the same speed than the regular algs. There would only be 64 algs which is quite good compared to the amount you would need for full CLL.
Here are the T set algs: (the orange represents which two colours have to be the same)

R' U' F U R U' R' F' R

(U') F R U R' U' R' F' R U R U' R'

(U) R U R' F' L' U' L F

(U2) R' F' L F R F' L' F

What did you guys use to make those images for the algorithms

 

[fabdabs1234]

What did you guys use to make those images for the algorithms

Paint [emoji23][emoji23]

 

[Skullush]

What did you guys use to make those images for the algorithms

Yeah I just made them in paint

 

[xyzzy]

I've been working on improving the solver/scrambler and I'm wondering if people are okay with the notation used in it or have some other preference. What we have now is U, R, F, L, BL, BR and flip, and I'm pretty sure the first four are very uncontroversial and can be kept as they are, but do people have objections to the others? I know some prefer "bL"/"bR" over "BL"/"BR" and some prefer "x2" over "flip".

Kit's suggested to get rid of one of BL/BR after the first phase, which might also help readability. Or we could even get rid of one of them entirely at the expense of slightly longer scrambles.

Ideas? Feedback from people who've used it for kilominx events at comps would be especially welcome.

Edit: Looks like the v2 will be about four times as fast at generating all the lookup tables while using considerably less RAM. Haven't touched the graph searching code yet, but that'll probably have a similar speedup. Edit #2: see sig. GUI to come later.

 

[Jaysammey777]

I've been working on improving the solver/scrambler and I'm wondering if people are okay with the notation used in it or have some other preference. What we have now is U, R, F, L, BL, BR and flip, and I'm pretty sure the first four are very uncontroversial and can be kept as they are, but do people have objections to the others? I know some prefer "bL"/"bR" over "BL"/"BR" and some prefer "x2" over "flip".

Kit's suggested to get rid of one of BL/BR after the first phase, which might also help readability. Or we could even get rid of one of them entirely at the expense of slightly longer scrambles.

Ideas? Feedback from people who've used it for kilominx events at comps would be especially welcome.

The scrambles work great! The only issue is that they are not necessarily "fast" to preform.as far as readability having all upper case I think is key. And the flip is intuitive for a lot of people, but some people don't get it immediately.

 

[xyzzy]

The scrambles work great! The only issue is that they are not necessarily "fast" to preform.as far as readability having all upper case I think is key. And the flip is intuitive for a lot of people, but some people don't get it immediately.

Cool. I take it that nobody else has any feedback?

I might be able to get the scramble sequences a bit shorter in the next version by doing partially-CN solves (it's currently not even <y>-neutral). As for current progress on the GUI, here's a screenshot (with random-move scrambles; this is configurable). Multithreading and such are in the works, and I'll release the code when that's done. This should overall be a huge improvement over the Python versions.

Edit (2016-12-20): link in sig updated. Now comes with a decent partially-random-state scrambler (with a rough analysis on how far from uniform the state distribution is) that doesn't take forever to initialise. Actual release on GitHub within a few days, maybe.

 

[xyzzy]

Is there a consensus on whether colour neutrality is worth it?

I used to use the same orders as I do for megaminx S2L, but now I just start wherever and extend the solved block however. Statistically speaking, the probability of having at least one free pair is around 40% with full CN versus only 8% for fixed colour, so an easy first layer should be a lot more common with CN.

The same downsides to CN for a megaminx also apply, of course. Using a fixed order means you don't have to think about what colours to track at every step, whereas with CN you'd have to be very familiar with the colour scheme, especially since there're no centres. (Or if you're me and have crappy tps, you can still "look ahead" to determine the colour scheme on the fly, but this takes away from actually looking ahead for pieces.)

 

[applezfall]

new kibiminx http://www.lightake.com/p/Fangge-2x2-Megaminx-Speed-Cube-Colorful_2285715.html

 

[Tabe]

Is Fangge the same as Cubestyle?

https://thecubicle.us/cubestyle-kilominx-p-9918.html

There's also a Fanxin kilominx on Ebay that looks exactly like the Fangge and Cubestyle. Different names for exact same product?

Note that these are much closer to an actual megaminx in that they have colored centers unlike the Shengshou kilominx, which has black/uncolored centers.

 

[xyzzy]

Is Fangge the same as Cubestyle?

The Chinese name on the box reads "Fangge Mofang", so yes, they're the same.