Help - Other QR code - Can it be reverse engineered?

@QuantumForce - Did you really extract the data from Python's zlib.decompress function? I tried and even build my own function to extract QR Code data straight from zbar module (which is used by qrtools) but the result still the same: "unknown compression method". I have compressed and uncompressed custom data, no problem. Something strange...

Here is the complete hex result from a fresh new card (just a new void card, with no modification). Can you tell me if you've got the same?

Code:

01 13 0d fc 78 01 ed c0 41 0a 00 10 10 40 51 5b 1b 67 70 15 7b d9 39 c2 94 a2 51 c2 5c df 3d f4 9f 2f 62 31 cb 95 11 aa 76 9d a6 31 9d dd e6 72 00 00 00 00 00 e0 7b 0f 43 81 08 dd 

Just in case the zbar implementation isn't good and I have to find another way...

Note: I pass to the zlib function the string from the 5th Byte, starting with 7801.

 

Got it!!!
Stupid types convertions issues with Python 2. :mad: If only zbar was ported to Python 3...
Now free games begin (but again, not as official level)

For those who ever wonder what a blank level looks like:

Spoiler: Blank new level

Well... That's av void one

 

Wow... That's soooo cool. Everything is so clear. The 3D organisation (from bottom to up, line per line), the blocks (01=stone, 05xx = stair #, 0c = grass, 02 = brick,...). If I didn't work tomorrow I could have stayed all night.

But now it's time to close my eyes... to dream of hex number and enter into the matrix
Cheers

 

@QuantumForce - Yep, my Python2 code works. For the moment it is only able to read and extract the QR Code data (the content of the level) from the card image file to a raw file. Eventually zbar works fine, but I achieved that using Linux as it is a mess to install using Windows (but possible, I did that once). Working with Python seems interesting because it would be great to spread the data in a numpy matrix and rotate the data with that (despite I never did that with 3D matrix...). Of course oriented blocks (2 bytes) will need to be rotate too.

Your list of blocks is wonderfull! It will save a lot of time. I have guessed the 3D organisation of the data level in the hex result (after the title and author), but maybe you did it too. Eventually, it is not as difficult as I thought.

The next step I want to test is :

1. Trying to regenerate the QR Code to determine the appropriate parameters (the level param is not the only one, there is a Wbits one for zlib)
2. Changing a block type and regenerate the QR Code image with the custom signature.
3. Puting all data in an array of matrix and try to play with rotations functions
4. Adapting the oriented blocks to the final orientation.
5. Then... Well, we'll see, but for sure: reorganising my code!

 

Well... For the moment, since I regenerate the zlib data after extraction I can't get the same than the original compressed data. I didn't try to modify anything yet as getting back to a readable QR Code is a key feature (otherwise anything is useless). When I try to pack the zlib data with the custom signature to a PNG QR Code, this is what I get:

Spoiler: Scan Error - This QR code is not a Mekorama game.

So I still investigate on that part better than the game block part (and @QuantumForce did/is doing a great job on that).

Indeed it's really interesting to know how all of this is packaged inside. But I don't wanna do too much work without knowing that I can use and test it.

 

Phew... I have succeeded to reproduce a valid QR Code for a level despite the fact that the re-encoded data are not done on the same format than the original one. I have even been able to compress more the zlib data! So it is not necessary to encode the level param in the same way than the original one, with the zlib no/low compression (0x78 01) but a more strong one (0x78 9c) works.

Of course, when the data are decoded and re-interpreted by the software then everything still the same on the rendered card.

Now I am confident in working with the level data as I know that they will be accepted by the software

So fun...

 

@QuantumForce - I was curious and checked the desert effect on the grass. The "green" grass is 0x0c and the "desert" grass is 0x12. The value for the grass in your code is the desert one

EDIT: Sorry you did right. You're expressing the value for grass (and other) in decimal, not hex
GRASS = 0x0c = 12
DESERT = 0x12 = 18

Really nice list! Thanks again

 

@sawdust for sure this is something that we're gonna check I don't know yet what will be find and if something will be find. But it will be easy to say that in a short amount of time (depending on my free time). Practically this can be achieved by changing a byte value with another one and check the results in the application.

@cpw I am not sure that speaking of changing bytes could be called a direct way At least it is not the easiest. As we don't have a 3D editor of the objects and plan things have to be done as we can but are possible. I have looked at a 3D svg representation of the blocks in a html5 container but we are far, far away from that . So everything seems possible now, in the scope of the game and according to the "tools" we have.

 

@cpw
So... What do you think of this?

Spoiler: From that

Spoiler: To that

Without any zapper trick!
It really starts to get interesting.

 

@cpw Yes, I admit that's a bit theoretical (or kind of nerds langages)... I need to release some code soon but for the moment I am just playing and don't really care of a code organisation. So that's a dirty brush for now

Just imagine that it is possible to open the box of the QR-Code, view all the objects/blocks of the game in the 3D location and give them another identity, in order to produce a new QR-Code. So we can add or modify them freely. I could even change the R-Bots to a grass block, or whatever.

If @QuantumForce didn't challenge us I am not sure I would have spent my time and dig on this...

 

Still playing... A bot orientation is N, S, E, W. But what happen when you try to set a bot orientation U (Up) Or D (Down) - which is usually impossible?

Spoiler: Orientation Down - 0x04

Spoiler: Orientation Up - 0x05

This is how everything can be messed up...

@QuantumForce - I have watched your Java code and it is beautifully organised. I think I will take example on your for my Python code. I have opened a github account for my mekorama pseudo (not my personnal - usually I work with Bitbucket) and will share with you my code as soon as I could be proud of something