This topic contains 3 replies, has 2 voices, and was last updated by Haplo 9 months, 1 week ago.
December 13, 2017 at 2:32 am #23623
What if the programmer created a junior version of the Babel Image Archive which consisted of pictures about 70% of the current size, say 300 x 462 pixels, and a smaller color palette. This would result in far less possible combinations. I understand part of the idea of the Library is to boggle our minds with it’s immensity, but a smaller image archive would for example be a more usable tool for artists looking for inspiration.
Actually I don’t know what would help artists, but what I really want is to actually find some neat randomly generated images.
It would still be incredibly huge. Would someone please calculate the total number of image combinations for a 300 x 462 pixel, 256-color palette archive?
256 colors are plenty right? Does it really need to be 4096 different colors? That seems like overkill.December 13, 2017 at 3:52 am #23624
An image library at that size would be 256^(300 x 462) images. Still pretty big:
Even an image archive of 10 x 10 pixels, or 256^100 is stupidly big.
Making a smaller library is just not feasible. I’ve thought about making a micro Library of Babel. Instead of books, it would be fortune cookies. But even that would be ridiculously big. I took a list from here and stripped away all the junk stuff like “(sic)” or “(2)”:
The shortest fortune is 15 characters, longest is 92. Average length is about 48, and the average of the shortest and longest is about 54. With the same 29 characters, and with a decent length of 60, it would be 29^60. Let’s take the shortest length of 15. The number is small enough to be written out without scientific notation, but it’s still massive (8,629,188,747,598,184,440,949). If you could read 60 fortunes a second, it would take 1 million people approximately 275 million years to go through them all.
EDIT: Oh, and it would look something like this
http://nickm.com/poems/babel.htmlDecember 13, 2017 at 11:31 am #23626
But still wouldn’t there be a great increase in the probability of finding something more meaningful than noise if you were using an archive say 100 x 154 pixels at 256 colors?
so 256^(15400) instead of 4096^(266240)
How many less combinations would this filter out?
Why wouldn’t this be feasible? Just change a few parameters in the algorithm, right?December 13, 2017 at 11:58 am #23627
the ratio of meaningful images to noise images would actually stay the same though, right? with the number of combinations to look through still being incomprehensible, your chances of finding something meaningful would be the ‘same’? This is beyond my math to understand.
I suppose ‘meaningful’ is relative term also – if you watch the slideshow enough you start to see things in the noise.