Using squares, circles and triangles with colours to represent bitcoin addresses

[remotemass]

Maybe it would get very visual to have a code to represent bitcoin addresses in a more visual way.
Uppercase letters would be squares.
Lowercase letters would be circles.
Numbers an equilateral triangle with horizontal base.
Colours would be used to distinguish them.

Who comes up with the better code?
Please post your artwork showing how would your code result.
We could also use 99designs.com to achieve great proposals!

[franky1]

Maybe it would get very visual to have a code to represent bitcoin addresses in a more visual way.
Uppercase letters would be squares.
Lowercase letters would be circles.
Numbers an equilateral triangle with horizontal base.
Colours would be used to distinguish them.

Who comes up with the better code?
Please post your artwork showing how would your code result.
We could also use 99designs.com to achieve great proposals!

not an actual submission. just food for thought
edit: google image search doctor who gallifrey

use of colours can be tricky due to light exposure making cameras not pick up colours that easily EG yellow street lghts can make yellow look white (missing transparent/white like the paper its printed on)

with 10 numerics 26 lowercase and 26 uppercase letters = 62 variations of shapes.

so heres a idea

Hexagon grey = 1
hexagon white=2
Square grey=3
Square white=4
Circle grey=5
Circle white=6
Triangle grey=7
Triangle white=8
basically multiply the outer shape by the inner shape and then add on the number of the centre shape
EG top shape is 1 4 7 so 1x4+7=11
EG bottom shape 6 7 5 so 6x7+5=47

now using the 0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ
top shape=a (lowercase a)
bottom shape=K (uppercase k)

[remotemass]

very rough draft:

[greyhawk]

I'm sure that's awesome for synesthetes, but how many of them are there even?

[remotemass]

Also for showing on television seems much better and less prone to human error.

[remotemass]

Very rough, but just to give an idea. What about having both, with the alphanumerics overlayed on top?

[franky1]

for geniune legit businesses a service that has registered bitcoin addresses linked to a user. kind of like tinyurl would work better

eg tinybitcoinaddress yet to be made.com/franky1
would take you to 1FrankZ7t5Wbf5uTMxMtCiQy9eKDsj1fUn

heck, even blockchain.info is already tagging known addresses so using their service to find the lengthy address is better. such as satoshi dice
http://blockchain.info/fb/1dice8

[BlackLilac Jordan]

for geniune legit businesses a service that has registered bitcoin addresses linked to a user. kind of like tinyurl would work better

eg tinybitcoinaddress yet to be made.com/franky1
would take you to 1FrankZ7t5Wbf5uTMxMtCiQy9eKDsj1fUn

heck, even blockchain.info is already tagging known addresses so using their service to find the lengthy address is better. such as satoshi dice
http://blockchain.info/fb/1dice8

I think what you're asking for already exists: https://btc.to/

[franky1]

BLFC Jordan, yes  thats what im talking about
great idea

[runeks]

I've thought about this too.

An often cited problem with verifying a bitcoin address is that the long string of letters and numbers is very hard to compare for a human being. This is relevant in the case of "hacker proof" hardware wallets where the user has to confirm that the address he copy/pasted from a site really is the address he wants to send the funds to.

I've considered simple color-coding, where the 25 byte address is converted into an image, which is experienced more as a whole instead of the letters, deciphered one-by-one when reading.

I just made a little script that encodes an address into a 12 by 12 pixel image. Each pixel is either red, green or blue. It basically just converts the address into a number, then converts this number into a base-3 number, and each pixel gets its color from the digit in the base three number (0 for red, 1 for green, 2 for blue). Here are some examples of random addresses converted into this image format:

1FbyYynZwHukrQZeAzPYb91KvbFkgp1kgN

	1LQxi4tDHca5ZAZW3HuBxof7Uxh54rTd9t
	1AdN2my8NxvGcisPGYeQTAKdWJuUzNkQxG
	124GJ3HCtYvaD2zDwJPKJLMTurWU1PdLic
	14jMgd1vQ8Pi9H3eKjsxwnb2xo8DZLiezf
	13xQfLkQ2q7xLDsEZfgv6weoZP98rmZnLx
	14cZMQk89mRYQkDEj8Rn25AnGoBi5H6uer

When uploading these pictures to imgur.com, their file names were lost (the filename had the address in it), but it was pretty quick just looking at the picture to see which were which. They are all distinct in various ways.

I'm sure there are ways to improve this algorithm though, so different patterns become more distinct.

Here's the script used to generate these images. By default the script will show the image encoding of the address contained in the address variable.

Code:

#!/usr/bin/python

import Image

#base58 magic borrowed from Gavin Andresen's bitcointools (and modified slightly)

__b58chars = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
__b58base = len(__b58chars)

def b58decode(v):
   """ return the number representation of the base58-encoded string v
   """
   long_value = 0
   for (i, c) in enumerate(v[::-1]):
      long_value += __b58chars.find(c) * (__b58base**i)

   return long_value

def b3decode(num):
   """ return the base 3 encoding of num as an array of numbers from 0 to 2
   """
   array = []
   base = 3
   remainder = num
   while remainder >= 2:
      div, mod = divmod(remainder, base)
      array.insert(0, mod)
      remainder = div
   return array

address = "1HB5XMLmzFVj8ALj6mfBsbifRoD4miY36v"

addresses = ["14cZMQk89mRYQkDEj8Rn25AnGoBi5H6uer", "13xQfLkQ2q7xLDsEZfgv6weoZP98rmZnLx", "14jMgd1vQ8Pi9H3eKjsxwnb2xo8DZLiezf", "124GJ3HCtYvaD2zDwJPKJLMTurWU1PdLic", "1AdN2my8NxvGcisPGYeQTAKdWJuUzNkQxG", "1LQxi4tDHca5ZAZW3HuBxof7Uxh54rTd9t", "1FbyYynZwHukrQZeAzPYb91KvbFkgp1kgN"]

width = 12
height = 11
image = Image.new("RGBA", (width, height))
pixels = image.load()

for (index, num) in enumerate(b3decode(b58decode(address))):
   colors = [(255, 0, 0, 255), (0, 255, 0, 255), (0, 0, 255, 255)]
   pixels[index % width, index / width] = colors[num]

image.resize((width*5, height*5), Image.NEAREST).show()

if False: #if this is run, it will save the image representation of the addresses contained in the "addresses" list to the current folder, using the file name "<address>.png"
   for addr in addresses:
      for (index, num) in enumerate(b3decode(b58decode(addr))):
         colors = [(255, 0, 0, 255), (0, 255, 0, 255), (0, 0, 255, 255)]
         pixels[index % width, index / height] = colors[num]

      image.resize((width*5, height*5), Image.NEAREST).save(addr+".png")

The resulting images are 12x11 pixels. Though the above images are 12x12 (with an empty bottom row since I didn't realize this until I had made these images).

[phelix]

http://en.wikipedia.org/wiki/Identicon



generator: http://finalboss.org/g/

[runeks]

Oh, my three-color images feel so primitive now.

But seriously, that might be useful if they really can encode 160 bits of entropy, and the changes aren't too subtle.

[Lethn]

I like that idea a lot, in case you didn't gather from my rants about math I'm not a mathematical thinker, shapes etc. are generally how I see the world not only that, if these could be turned into QR codes or even our usernames etc, that would be fucking badass!

They remind me a lot of the Japanese clan mons ( emblems ) in Shogun 2, maybe we could have our own unique symbols for Bitcoin addresses I bet this would get quite a bit of attention from normal people too.

Edit: I like this a lot lol! I put my username in and it popped out the avatar I put up

[Ente]

I like where this is going!
Exactly, this will be very important for visually comparing adresses on my computer and a dedicated wallet hardware with display.
Identicon: So, how many bits would fit into one of those? Wikipedia didn't help with technical details..
Runeks, your system is nice too! 12x12? Champion!

Ente

[runeks]

Check out this paper: https://sparrow.ece.cmu.edu/group/pub/old-pubs/validation.pdf

Looks like they're taking the task seriously (defining the image-producing function to have the same properties as a hash function). The images look nice and distinct. But I'm not sure it can produce 2^160 sufficiently visually distinct images. Interesting none the less.

EDIT: This looks cool too, and it's open source: http://www.thevash.com/

[paybitcoin]

Don't forget about randomart used with ssh keys:

http://superuser.com/questions/22535/what-is-randomart-produced-by-ssh-keygen

sample:

Code:

Your identification has been saved in test.
Your public key has been saved in test.pub.
The key fingerprint is:
20:e9:b0:5b:5a:2b:ad:e8:4d:e4:b3:a0:32:49:2d:97 evan
The key's randomart image is:
+--[ RSA 2048]----+
|                 |
|     .           |
|  . o .          |
|   + . .         |
|  o.=   S        |
| ooE .           |
|.o*+o            |
|=.+oo            |
|=o.o             |
+-----------------+

[theymos]

very rough draft:

Straight substitution isn't very creative/interesting IMO. I do like the idea of representing addresses as cool-looking images, though.

[Matthew N. Wright]

http://en.wikipedia.org/wiki/Identicon



generator: http://finalboss.org/g/

Is there a way to read these in reverse and get the string from the image?

[Red Emerald]

http://en.wikipedia.org/wiki/Identicon



generator: http://finalboss.org/g/

Is there a way to read these in reverse and get the string from the image?

Nope. Thats by design since they were originally meant to identify IPs without revealing the IP.

If the input fields for addresses in the client had a small identicon next to them, and whatever people were manually copying from also had an identicon, it would be a quick way to check for typos in addition to bitcoin's built in checksum.  If you are copy/pasting from the same computer, or using a QR scanner, there isn't really a need.

[phelix]

one could use hash-colored qr codes as identicons...

   

if your client calculates the same code from the address you can be pretty sure you copy/pasted the right one.