Mega Man, Made Out Of 97 Rubik’s Cubes

Anyone can grab some coloured sticky notes and bust out some impromptu pixel art. It takes real skill to transform 97 mini Rubik’s Cubes into a pixel-perfect Mega Man. Impressed? Wait until you see his Pokemon.

YouTube user and Rubik’s Cube expert Kyudan2 has been quietly busting out these incredibly impressive works of puzzle art for months now. Up until now, his focus has mainly been in recreating Pokemon sprites, like the simple Ash and Pikachu from Pokemon Yellow

… to more complex Pocket Monsters such as Togepi, rendered here in 99 cubes.

Amazing work. Kyudan2 has more in his YouTube channel, and plans new vids on a regular basis, so go subscribe the living heck out of him.


  • Apparently, (according to Laurence Leung), the secret to Rubik’s cubes is a mathematical algorithm. Complex in nature, yet inherently simple once you understand the mathematical process and are able to reliably apply it.
    I didn’t realize “being on acid” was an acceptable substitute.

    • Haha, it was because of him that I decided solving a Rubik’s Cube seemed like the kind of skill I should already have, so went off and learnt how to do it.

      I don’t know that I find it particularly mathematical though. I mean I can see how the creation of it would be based in mathematics, but for me it’s just a series of movements which have now more or less become muscle memory.

        • Hmm… apparently a skydive freefall from 10k takes 30 seconds? Oh, that’s the drop from 10 down to 4k. Then “up to six minutes to landing”.

          My best is 1:13 and average is about 1:30-2:00, so I reckon I could manage that 😛

          • Ha, good luck with it! You’ll be buying that video, regardless of the cost, I suspect …

      • While it eventually becomes muscle memory, you can break down most of those learned moves into simple algorithms, mostly A B A’ or A B A’ B’, which become very useful once you start solving higher-order cubes – 4×4 through 7×7, for example, and can be applied to any puzzle of this nature.
        Hmm… this makes me want to start doing pixel-art on my 7×7…

        • Yeah, I found it interesting that the dodecahedron my sister got me was nearly solvable with exactly the same moves as used on the cube. Would love one of the 4×4 cubes to try.

          • Yea, the megaminx is pretty much a 3×3 with more faces (sounds like an obvious thing, but it actually really helps to see it that way). I’ve even got a gigaminx (5×5 dodecahedron), but that took ages (so… many… edge… pieces…) – I had to do it over the course of three sittings, cos I literally spent an hour and a half just matching up edge pieces. So that one stays on the shelf, now. 😛
            If you start on the bigger cubes, I’d suggest doing an odd one first, as they only really require one algorithm beyond a 3×3, whereas the even cubes (except the 2×2) can use that same extra algorithm, but require another two really awkward ones to solve the parity cases. I can solve my 4×4 and 6×6 only about two times in three, because I’ve not figured out one of the parity cases.

          • Out of interest, what method do you use to solve the 3×3? I started out with the beginner one, doing cross -> corners -> edges -> all that stuff with the top, then the intermediate where you do the corners and edges together (though took the recommendation of figuring out how to do that yourself rather than just learning the moves, to help you learn how the pieces move around. Worked a treat), but then kind of went off the track a bit when learning the top layer stuff for that. I forget what the methods online told you to do, but I ended up kind of figuring out how the top layer ended up showing certain patterns on top, and worked out that they were all at most two moves away from a pattern that could be transformed straight into having the top all one colour. I tried looking into the “advanced” method, but just couldn’t get my head around anything I read in relation to it.

            Have you got the cube that has no colours, but instead has the centre offset so that the pieces are all physically different shapes instead? That one’s my favourite, especially trying to solve it by feel alone 😛

          • Same as you, I started with cross/corners/edges/stuff, then followed advice to learn my own F2L method for the same reasons, following the Fridrich method, but stopped when I got to the point that I could combine any two algorithms to solve OLL, ditto for PLL, and concentrated on speeding up recognition time. I might someday learn all the algs, but for now I can get away with knowing only half the Fridrich algs, and considering I spend 20-25 seconds on F2L, and about 5-10 on last layer, they’re not a top priority. Haven’t spent much time looking at the advanced methods, but they look like they either require a huge amount of memorisation, or a really good understanding of commutators/conjugates. Heise looks interesting, but I’m not sure my understanding is quite there yet…

            Are you talking about the mirror cube? (, third from the top, on the left)

            That one takes a bit of getting used to. I always get frustrated with mine cos it’s a cheapo, and doesn’t turn well. But it’s just a matter of figuring out what goes where, and then for me it’s just a frustratingly slow 3×3. I’ve still got a bunch of puzzles that bend my brain – I’m currently working on figuring out a 3x3x2, to work my way up to my 3x3x7 (and eventually get a 3x3x9), but I’ve got a 3x4x5 that is doomed to remain unsolved… 🙁 But of all of them, I still find I like my 3×3 the best.

Show more comments

Log in to comment on this story!