What If Every Picture You've Ever Seen Already Exists? | Hacker News Reader

What If Every Picture You've Ever Seen Already Exists?

20 pointscin4ed28 comments

I was thinking recently about how images work at the data level, and it kind of broke my brain.

Take a simple case: a 3×3 pixel image with only black and white pixels. There are only 9 pixels, and each has 2 options (black or white), so the total number of possible unique images is:

2^9 = 512

That’s tiny, you could generate and look at every one of those images in a few seconds. But already, you’re looking at the complete universe of 3×3 B/W images. Every possible shape, face, glitch, symbol, if it can exist in that resolution and color range, it’s already in there.

Now scale up.

A 1920×1080 image (full HD), with each pixel using 24-bit RGB (i.e., 16.7 million colors), has:

(2^24)^(1920×1080) = 2^49,766,400 ≈ 10^14,983,365

That number is incomprehensibly massive. It’s orders of magnitude larger than the number of atoms in the observable universe (≈10⁸⁰). And yet, it’s finite.

Which means:

- Every possible frame of every possible movie is mathematically there.

- Every photo you never took exists in this space.

- Every piece of digital art, every childhood memory, every face, every impossible scene, all of it is representable by just one of those possible combinations.

Of course, almost all of those images are noise. Pure entropy. But buried in that space is literally everything.

Makes you wonder, are we creating images? Or are we just exploring a tiny, meaningful subset of a space that already contains them all?

Reminds me of: https://en.wikipedia.org/wiki/Sloot_Digital_Coding_System

The Sloot Digital Coding System is an alleged data sharing technique that its inventor claimed could store a complete digital movie file in 8 kilobytes of data — which, if true, would dramatically disprove Shannon's source coding theorem, a widely accepted principle of information theory that predicts how much data compression of a digital file is mathematically possible. The alleged technique was developed in 1995 by Romke Jan Bernhard Sloot (27 August 1945, Groningen – 11 July 1999,[1] Nieuwegein), an electronics engineer from the Netherlands.[2] Several demonstrations of his coding system convinced high-profile investors to join his company, but a few days before the conclusion of a contract to sell his invention, Sloot died suddenly of a heart attack. The source code was never recovered, the technique and claim have never been reproduced or verified, and the playback device he used for demonstrations was found to have contained a hard disk drive, contrary to what he told investors.

This is really interesting thought. Could someone help me understand where I'm going wrong? Because I assume that Sloot made the same mistake that I'm making here, and I'm struggling to understand the Shannon's source code theorem article, so I'm thinking in practical terms:

If you wanted any movie in 1080x768, and wanted to have it in full colour, that would be 16,777,216 possible colours per pixel, so the number of IDs in your sequentially generated noise database would be 16777216 * 1024 * 768 = 1.319414e+13. This is where I think I might be misunderstanding something, because that number in decimal form is apparently 13194140000000, which means every frame will have an ID that is only 14 digits long.

Where have I made the mistake in concluding that I could store any 1024x768 image in 14 characters so long as my computer could generate an image of that ID?

Edit: I've just realised that 1024x768 is not a standard resolution, but I suppose the point remains with 1920x1080

There is a small mistake in the formula: the actual cardinality of the noise database would be 16777216 ^ (1024 * 768), which is a much bigger number.

Ah yes, the way I was thinking of it, I was only considering the variations of one pixel in each position of the image, not combined. Thanks!

I wonder if this could be done with some kind of seed generation (as in minecraft), but it sounds like magical thinking to have a seed and generator anywhere near small enough to not just contain all the data anyway.

is this yet another tech hoax?

Dying is a pretty extreme length to go to for a hoax

You mean the Library of Babel's universal slideshow? https://babelia.libraryofbabel.info

You can go through picture by picture or search for a picture you already have.

I think there's a problem with tense in the title, obviously all the pictures I have seen so far in my life existed (or I wouldn't have seen them...) The question posed is more about whether all the pictures I will see in the future are already in some sense present, which might be true, but I still haven't seen them yet...

Yeah agree, exists can be debatable.

Similar to the plot of the library of babel (https://en.wikipedia.org/wiki/The_Library_of_Babel)

And the actual website: https://libraryofbabel.info/

As well as their slideshow: https://babelia.libraryofbabel.info/

We live in a many dimensional phase space. Our life is a trajectory line in it, moving along the line is percepted as time.

God sees all the space at once (this is an “all knowing” part).

But we are left with the choice of which turn to take at every moment (this is the “free will” part), which picture to draw, which movie to create (this is the “in god’s image” part) etc.

From human perspective there is no difference if you create or explore the phase space. Just a wordplay. Because by definition you can’t distinguish those.

One of these images is the screenshot of this conversation, including your reply to this comment that you haven't posted yet.

One of these images is a picture of your great-great-great-great-great-great-grandchild, living in the year 3460, whom you never met.

The core assumption here that the color depth is fixed or that the frame size cannot grow is flawed. We can always increase both, creating infinite possibilities.

Also you'd have to generate these images for them to exist and not merely be probability.

The logic applies for any given resolution or color set, I'm just using full hd here because is the most common (same for color) and enough to display a pretty detailed picture, the core idea here is that no matter what kind of image you are trying to make, it is already constrained into a finite set of combinations.

In my opinion, this set of images contains no information at all. A very short program can generate it. I could offer the whole repository for an affordable cost on a cloud image hosting service because the filename for each image would have to be at least as long as the image itself. When users request an image I need only echo the request back to them.

Here is one - there are finitely many mathematical symbols (or at least, all mathematical symbols can be defined in terms of a finite core of symbols).

That means the set of all mathematical definitions is countable (i.e. you could assign a whole number to each one, putting them into an infinitely long ordered list).

However, the set of real numbers is uncountable (by Cantor's argument).

Therefore the vast majority of numbers ("almost all" numbers, in a mathematical sense) cannot be defined, even in principle.

The big question is, can we ever know if the laws of the universe are governed by those undefinable ("uncomputable") numbers?

Can I move an object X meters away from me, where X is an uncomputable number?

Whether the answer is yes or no, the consequences are very interesting to me.

The vast majority of numbers also aren't useful or interesting.

Damn, I think I need chatgpt to explain me this one

This seems like an isomorphism of "is math invented or discovered"?

https://philosophy.stackexchange.com/questions/1/is-mathemat...

that really makes sense while it is very hard to imagine about. i have thought about this several times before too, but rather than that its for files, not just images, with the similar idea: how many distinct files can be created from a fixed file size, say, 4KB?

in fact, you "create" something when its not existed yet, or its never been perceived to exist by anyone. despite in theory everything are already in track of an algorithm, we cant really sure that its already there, or its creating enough effect to be known to exist, i.e. hardly anyone will think about it.

well, after all, whether it is "created" or "discovered" depends on the context and people's general perception about it.

You should look at the jpeg/mpeg compression algorithm, definitely contains bits of this idea but uses the actual images as a starting point

Here is a number that has all those pictures encoded: π

It is unknown whether the digits of pi contains every single finite sequence of numbers. But the Champernowne Constant does[1].

[1] https://en.m.wikipedia.org/wiki/Champernowne_constant

See also: What Colour are your bits? - https://ansuz.sooke.bc.ca/entry/23

Reply on news.ycombinator.com