"Every Image Ever Theory" (I need alot of volunteers)

Author

Message

BMacZero

18

Years of Service

User Offline

Joined: 30th Dec 2005

Location: E:/ NA / USA

Posted: 23rd Nov 2008 20:52 Edited at: 24th Nov 2008 03:56

Link

Quote: "I didn't read the whole thread, so I don't know if this has been suggested, but if your program were designed, for example, to generate new smiles, like someone jokingly suggested, you could set up a simple neural net, and award points to the generations that had a round shape, or a mostly yellow colour, for example. While that would be extremely over complex, I bet over time you'd come up wiht some pretty sweet smilies."

Actually, you could probably help it out a lot by making it just draw round, yellow shapes to start with.

Back to top

Profile PM Email Website

code master

20

Years of Service

User Offline

Joined: 4th Dec 2003

Location: Illinois

Posted: 24th Nov 2008 02:34

Link

But that just takes the fun out of it...

Weapon Creator V1 Is for sale!

Back to top

Profile PM Email

BMacZero

18

Years of Service

User Offline

Joined: 30th Dec 2005

Location: E:/ NA / USA

Posted: 24th Nov 2008 04:05

Link

Hehe, fun. I've started working on a random-smiley program, anyway. It needs to be refined a lot though, the current version would require about 178 days and 3.1 terabytes of disk space to generate and store every possible smiley, and that's assuming 100% efficiency, which isn't going to happen yet because it's completely random. Plus, then at the end someone has to go through more than a million sheets with 2700 smileys on each. This random-image business definitely deals in very large numbers...

Back to top

Profile PM Email Website

Valle

18

Years of Service

User Offline

Joined: 10th Mar 2006

Location: in your girlfriends bed

Posted: 24th Nov 2008 16:57 Edited at: 24th Nov 2008 16:59

Link

+ Code Snippet

#CONSTANT SCREENW 1280
#CONSTANT SCREENH 1024
#CONSTANT PICW 20
#CONSTANT PICH 20
picsx = SCREENW/(PICW+1)-1
picsy = SCREENH/(PICH+1)-1

set display mode SCREENW, SCREENH, 32

input "Gimme a numba! -> ", seed
randomize seed

sync on : sync rate 0 : sync

cls
repeat
   for px = 0 to picsx
      for py = 0 to picsy
         lock pixels
            for x = px*(PICW+1) to (px+1)*(PICW+1)-2
               for y = py*(PICH+1) to (py+1)*(PICH+1)-2
                  dot x, y, rgb(rnd(255), rnd(255), rnd(255))
               next y
            next x
         unlock pixels
         sync
      next py
   next px
   wait key
until escapekey()

much faster

Back to top

Profile PM

Valle

18

Years of Service

User Offline

Joined: 10th Mar 2006

Location: in your girlfriends bed

Posted: 24th Nov 2008 17:44 Edited at: 24th Nov 2008 17:46

Link

Here we have some very nice pictures, generated, 20*20, black&white.

upper left: a nice medieval castle. Some houses, trees and a marketplace. very idyllic.

upper middle: a fierce dragon in the sky, in the bottom right lies the city he destroyed! You could also discover a heart in that one. how boring...

bottom middle: This one really shocked me! A dragon raping a duck while eating a bonefish

. Seriously. Don't show this to your children.

bottom right: an underwater scene. A sea horse and some underwater-plants.

Please help me to interpret the other two. I think theres a crowd of naked people running away from the sun (vampires?) in the upper right and a sunny pirate bay with stairs on the right in the lower left. But I may be wrong

.

Back to top

Profile PM

BMacZero

18

Years of Service

User Offline

Joined: 30th Dec 2005

Location: E:/ NA / USA

Posted: 25th Nov 2008 01:09

Link

I think I actually see the dragon...

Back to top

Profile PM Email Website

jasonhtml

20

Years of Service

User Offline

Joined: 20th Mar 2004

Location: OC, California, USA

Posted: 26th Nov 2008 07:28

Link

top right: a very happy man watching master chief shoot himself in the face!

Back to top

Profile PM Email Website

hmm

20

Years of Service

User Offline

Joined: 11th Oct 2003

Location:

Posted: 28th Nov 2008 20:37

Link

Always kind of understood the idea that with enough time a random sequence of whatever could form a meaningful something, but never really thought about it in detail before. I like the idea that you could generate every image ever. It'd be nice if you developed some automated process that could distinguish meaningless noise from an image of interest so that you could leave it the program going, saving only the 'good' images. Though whats to say meaningless noise is meaningless? Maybe thats what the beginning of the universe looked like.

I'd recommend perhaps limiting the experiment to greyscale images first. Significantly less possible images. I've been studying computational neuroscience and all the experiments involving visual stimulii are limited to greyscale to simplify things. But then, I'm an engineer, not a mathematician or a physicist, so I like simplifying things.

Back to top

Profile PM Website

IanM

Retired Moderator

21

Years of Service

User Offline

Joined: 11th Sep 2002

Location: In my moon base

Posted: 28th Nov 2008 22:35

Link

Quote: "I'd recommend perhaps limiting the experiment to greyscale images first"

I'd recommend not bothering at all. The numbers involved, even on a 20x20 black/white image are so immense that we're talking about billions of years to cover all the possible combinations, even at 1000 images per second.

Oh, and adding 'random' into the mix will actually make it take even longer - with random, you have duplication.

20x20 black/white has 2^400 possible images.
20x20 8 bit grey scale has 2^3600 possible images.

Do the math - windows calculator is capable of giving you an approximate answer.

Utility plug-ins collection (updated 27/11/08) and
http://www.matrix1.demon.co.uk

Back to top

Profile PM Email Website

Carlos508

15

Years of Service

User Offline

Joined: 28th Nov 2008

Location:

Posted: 28th Nov 2008 22:39

Link

Nice idea, won't work. In fact, I suggest you stop now before you waste your time. Here's why:

Quote: "implify there literally can only be (10000^16777216) different images on any 100x100 24-bit image. Now I know that thats ALOT"

No, you don't know. Dude, there are only 10^79 electrons in the entire universe(!!!) and your saying that there can only be 10000^16777216...think about it man, for all intents and purposes, that is infinite.

If you where to store the random seed value that could produce the image, it would take up more memory than every computer that has and ever will exist, multiplied by at least 10^79 (that's one with 79 zeros remember). Think about it, 10000^2 = 100000000, and 10000^4 = 10,000,000,000,000,000 and we still have 16777212 to go!

I doubt you could replicate even a 2x2 image to the exact RGB value with a single random seed, even multiple, in a reasonable amount of time to make it worth it.

"I don't have anything to say, I just want my cake!"
- Louie Ferrigno, Pumping Iron

Back to top

Profile PM

IanM

Retired Moderator

21

Years of Service

User Offline

Joined: 11th Sep 2002

Location: In my moon base

Posted: 29th Nov 2008 02:53

Link

2x2x24 is 2^96, at 1000 images per second that would take over 2.5 billion billion years ... approximately. Hmm, and the universe is only 14 billion years old.

Now, as for and image 100x100 at 24 bit colour, which is 2^(100*100*24) => 2^240000 ... well, I'm not even going to bother working that out.

Utility plug-ins collection (updated 27/11/08) and
http://www.matrix1.demon.co.uk

Back to top

Profile PM Email Website

Butter fingers

18

Years of Service

User Offline

Joined: 20th Mar 2006

Location: Mecca

Posted: 29th Nov 2008 03:41

Link

Quote: "I would describe it as being exactly like the infinite monkey theory, rather than the pigeon-hole theory."

"The internet has proved that a billion monkeys bashing a billion keyboards will not produce Shakespeare"

Back to top

Profile PM Email Website

BMacZero

18

Years of Service

User Offline

Joined: 30th Dec 2005

Location: E:/ NA / USA

Posted: 29th Nov 2008 16:01

Link

Quote: ""The internet has proved that a billion monkeys bashing a billion keyboards will not produce Shakespeare""

Well, those monkeys aren't bashing randomly. Oh, wait a second...

.

Can someone tell me if my math is off here? I calculated the number of possible smileys in RGB color and got semi-reasonable results (the data are actual data from my program).

The program creates 15x15 smileys, but only 137 pixels actually drawn on (because the smileys are drawn as a circle).

137 * 256^3 = 2,298,478,592 possibilities. (I'm probably not multiplying it by enough 256s, though.)

2700 smiles drawn per screen = 851,289 screens

Average of 500 ms per screen (drawing and storing time) = only 118 hrs (about 5 days) to draw every possible 15x15 smiley in RGB color and store it.

Each sheet is 3,145,784 bytes, so about 2.44 TB of hard drive is required.

Back to top

Profile PM Email Website

IanM

Retired Moderator

21

Years of Service

User Offline

Joined: 11th Sep 2002

Location: In my moon base

Posted: 29th Nov 2008 18:45

Link

Take the number of pixels (x times y), multiply by the number of bits (1 for black/white or on/off, 8 for greyscale, 24 for true colour).

Then raise 2 by the power of that number, eg 2^(15*15*1) => 2^225 => 5.3912 * 10^67 - that's a 5 with 67 digits following it.

2700 smilies every 500ms, it would take 3.166 * 10^56 years to cover every combination - longer if you use random numbers.

If you have 137 pixels of 24 bits each, then the numbers are far larger.
2^(137*24) => 2^3288 => 6.118 * 10^989, or 3.593 * 10^978 years.

Every extra bit you add, you double the number of possible combinations - it makes no difference whether the bits are for existing pixels (more colours) or for new pixels (extra rows or columns).

Utility plug-ins collection (updated 27/11/08) and
http://www.matrix1.demon.co.uk

Back to top

Profile PM Email Website

BMacZero

18

Years of Service

User Offline

Joined: 30th Dec 2005

Location: E:/ NA / USA

Posted: 30th Nov 2008 15:52

Link

Oh, dang. Thanks, Ian!

Back to top

Profile PM Email Website

Kevin Picone

21

Years of Service

User Offline

Joined: 27th Aug 2002

Location: Australia

Posted: 30th Nov 2008 17:46

Link

I'm probably missing something here, but what exactly is the point of storing/generating every permutation ?

Learn to code for FREE with PlayBasic Learning Edition

Back to top

Profile PM Website

C0wbox

17

Years of Service

User Offline

Joined: 6th Jun 2006

Location: 0,50,-150

Posted: 30th Nov 2008 22:36

Link

Quote: "I'm probably missing something here, but what exactly is the point of storing/generating every permutation ?"

So people in years to come can look back at human society of the 21st century and go:
"What a bunch of nutters they were, what use is all this data?"

xD

Back to top

Profile PM Email Website

Neuro Fuzzy

16

Years of Service

User Offline

Joined: 11th Jun 2007

Location:

Posted: 23rd Dec 2008 21:28

Link

Ha!

one side is arguing "It could happen, but it's improbable beyond belief"
the other side's saying "it's improbable beyond belief, but it could happen"

Sounds cool, i'd definitely run a program like this for hours if you could get one that would methodically go through, and wouldn't come up with the same image twice. I don't think you would need to save any of the images though, i could imagine more of a video with images popping up a little different from the last. Maybe start out with completely white image, and work your way there, or a button that would set every pixel to random.

Back to top

Profile PM Email

Diggsey

17

Years of Service

User Offline

Joined: 24th Apr 2006

Location: On this web page.

Posted: 29th Dec 2008 15:52 Edited at: 29th Dec 2008 15:56

Link

This code is pretty good at making random smileys:

data 0,0,0,0,0,1,1,1,1,1,0,0,0,0,0
data 0,0,0,1,1,1,1,1,1,1,1,1,0,0,0
data 0,0,1,1,1,1,1,1,1,1,1,1,1,0,0
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 0,0,1,1,1,1,1,1,1,1,1,1,1,0,0
data 0,0,0,1,1,1,1,1,1,1,1,1,0,0,0
data 0,0,0,0,0,1,1,1,1,1,0,0,0,0,0

cls 0
lock pixels
for y = 0 to 14
    for x = 0 to 14
        read a
        if a
            dot x,y,0xFFFFFF00
        endif
    next x
next y
unlock pixels
get image 1,0,0,15,15,1
cls 0

nx = screen width()/20
ny = screen height()/20

set image colorkey 1,1,1

do
    for y = 0 to ny-1
        for x = 0 to nx-1
            px = x*20
            py = y*20
            paste image 1,px,py
            lock pixels
            n = rnd(30)+5
            for i = 0 to n
                dot px+rnd(12)+1,py+rnd(12)+1,0
            next i
            unlock pixels
        next x
    next y
    do
        omx = mx
        omy = my
        mx = mousex()/20
        my = mousey()/20
        if mx <> omx or my <> omy
            ink 0xFF000000,0
            line omx*20-1,omy*20-1,omx*20+16,omy*20-1
            line omx*20-1,omy*20-1,omx*20-1,omy*20+16
            line omx*20-1,omy*20+16,omx*20+16,omy*20+16
            line omx*20+16,omy*20-1,omx*20+16,omy*20+16
            ink 0xFFFFFFFF,0
            line mx*20-1,my*20-1,mx*20+16,my*20-1
            line mx*20-1,my*20-1,mx*20-1,my*20+16
            line mx*20-1,my*20+16,mx*20+16,my*20+16
            line mx*20+16,my*20-1,mx*20+16,my*20+16
        endif
        omc = msc
        msc = mouseclick()
        msd = msc-omc
        if msd = 1
            get image 2,mx*20,my*20,mx*20+15,my*20+15,1
            save image "RNDIMAGE_"+hex$(rnd(0xEFFFFF)+0x100000)+".png",2
            delete image 2
        endif
        sync
        if scancode() then exit
    loop
loop

+ Code Snippet

data 0,0,0,0,0,1,1,1,1,1,0,0,0,0,0
data 0,0,0,1,1,1,1,1,1,1,1,1,0,0,0
data 0,0,1,1,1,1,1,1,1,1,1,1,1,0,0
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 0,1,1,1,1,1,1,1,1,1,1,1,1,1,0
data 0,0,1,1,1,1,1,1,1,1,1,1,1,0,0
data 0,0,0,1,1,1,1,1,1,1,1,1,0,0,0
data 0,0,0,0,0,1,1,1,1,1,0,0,0,0,0

cls 0
lock pixels
for y = 0 to 14
    for x = 0 to 14
        read a
        if a
            dot x,y,0xFFFFFF00
        endif
    next x
next y
unlock pixels
get image 1,0,0,15,15,1
cls 0

nx = screen width()/20
ny = screen height()/20

set image colorkey 1,1,1

do
    for y = 0 to ny-1
        for x = 0 to nx-1
            px = x*20
            py = y*20
            paste image 1,px,py
            lock pixels
            n = rnd(30)+5
            for i = 0 to n
                dot px+rnd(12)+1,py+rnd(12)+1,0
            next i
            unlock pixels
        next x
    next y
    do
        omx = mx
        omy = my
        mx = mousex()/20
        my = mousey()/20
        if mx <> omx or my <> omy
            ink 0xFF000000,0
            line omx*20-1,omy*20-1,omx*20+16,omy*20-1
            line omx*20-1,omy*20-1,omx*20-1,omy*20+16
            line omx*20-1,omy*20+16,omx*20+16,omy*20+16
            line omx*20+16,omy*20-1,omx*20+16,omy*20+16
            ink 0xFFFFFFFF,0
            line mx*20-1,my*20-1,mx*20+16,my*20-1
            line mx*20-1,my*20-1,mx*20-1,my*20+16
            line mx*20-1,my*20+16,mx*20+16,my*20+16
            line mx*20+16,my*20-1,mx*20+16,my*20+16
        endif
        omc = msc
        msc = mouseclick()
        msd = msc-omc
        if msd = 1
            get image 2,mx*20,my*20,mx*20+15,my*20+15,1
            save image "RNDIMAGE_"+hex$(rnd(0xEFFFFF)+0x100000)+".png",2
            delete image 2
        endif
        sync
        if scancode() then exit
    loop
loop

Just click on any smileys you like and it will save them out to the current directory.
Press a key to generate a new page.

Some nice ones: