-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
The Source Code (MUST READ)
- Thread starter silver
- Start date
- Status
Compression only comes at the cost of variation. Thats just how compression works, the patterns are what gets compressed, all that is outside the compression patterns has to be sored without compression.
In theory you could even make a doom clome with bump mapping in 64k
Here is one at 96k
http://www.theprodukkt.com/kkrieger.html
In theory you could even make a doom clome with bump mapping in 64k
Here is one at 96k
http://www.theprodukkt.com/kkrieger.html
OpinionatedCyborg
Member
The question why industry heads would get behind this was raised, and the answer's simple: when supposedly new technology comes around, you get behind it as quickly as possible. These industry heads had a tech demo, a crackpot explanation to work off of, and nothing more. By investing a relatively small amount on an unproven, yet potentially revolution idea, they're safeguarding their own interests. If the idea fails, whoops. If it does, you're in the money. If a revolutionary product presents itself, it's a good idea to be in on it--even if it fails, you aren't much worse off.
Have you ever heard of the "dot com" era? Supposedly smart people fell for all kinds of shit, and they even threw money away on stupid ideas!silver said:
Compression is the result of mapping the contents of one set of data to the contents of another, smaller set of data. As others have eloquently pointed out, the level of compression being claimed is utterly absurd.
(By the way, your little "GAYMING AGE" crack was noted and not appreciated.)
As Wolfram pointed out in his unnecessarily large and mostly boring book "A New Kind of Science", things of very high complexity can be described by small rulesets (like 4 logical operations) given small inputs (as small as 2 bits)... and enough free memory to output to of course.
Perhaps this guy figured out some way to find the right rulesets and input bits to break down large blocks of bits into self contained procedural cellular automata generators that could be described in a few bits.
Of course, we all know how incredibly hard it would be to do that. That's like solving NP in P time. Chances are it was all bunk. But theoretically, if someone could "reverse procedurally generate" things, you could get crazy crazy compression ratios.
Perhaps this guy figured out some way to find the right rulesets and input bits to break down large blocks of bits into self contained procedural cellular automata generators that could be described in a few bits.
Of course, we all know how incredibly hard it would be to do that. That's like solving NP in P time. Chances are it was all bunk. But theoretically, if someone could "reverse procedurally generate" things, you could get crazy crazy compression ratios.
You know, you may all think I'm retarded, but seriously: if somebody discovers a totally new way of compressing data that no one knows about, of course you think "impossible, retarded". Because if you didn't, you'd be the one working on that code.
I could tell you how it works from what I've read, but I doubt I could explain it well.
Right. One thing I didn't tell you is that Pieper quit (his multi-million dollar a year position) at Philips to head up Fifth Force.
Point taken. Now read up on how Sloot's technology worked, then come back to criticize.
For the record: Pieper gained complete control of Fifth Force, Inc. and the company still exists today, because the real code still hasn't been found. Sloot claimed that it was somewhere in a vault, but no one knows. Sloot was also a heart patient. He was not murdered. He was under pressure, had just travelled through the US with Pieper and died at a bad time. Still, if his story was bullshit... talk about going out with a bang.
I could tell you how it works from what I've read, but I doubt I could explain it well.
Right. One thing I didn't tell you is that Pieper quit (his multi-million dollar a year position) at Philips to head up Fifth Force.
Point taken. Now read up on how Sloot's technology worked, then come back to criticize.
For the record: Pieper gained complete control of Fifth Force, Inc. and the company still exists today, because the real code still hasn't been found. Sloot claimed that it was somewhere in a vault, but no one knows. Sloot was also a heart patient. He was not murdered. He was under pressure, had just travelled through the US with Pieper and died at a bad time. Still, if his story was bullshit... talk about going out with a bang.
RonaldoSan
Member
"No one will need more than 637 kb of memory for a personal computer." - Bill Gates
Technology evolves...
Technology evolves...
The Faceless Master
Member
i remember that other time when alot of the technology guys proclaimed something as the next coming of jesus christ, it was heralded by many as a revolution...
what did we get?
what did we get?
88,473,600,000/38.125=135,000 frames in a movie/program
thats 1/2 a byte per frame.
The only way I can see it possible is. Assuming that you have a single algorithmic shape that stays there or moves(for pans etc), you know minor variations and is there for say 256 frames or 9 seconds, you can get can get a for example a 256 pixel single colour square object that would normally take atleast 256 kilobytes into ~10 bytes(4 bytes for colour ~6for square algorithm) reduction space used by a factor of 25 600 But there are elements that always that exist outside formulas. And because of that they can take tons of space. But doing so in 64k is an extreme event where all the algorithm fall into place and there are no outside elements and no change in light which changes the pixels colour slightly.
thats 1/2 a byte per frame.
The only way I can see it possible is. Assuming that you have a single algorithmic shape that stays there or moves(for pans etc), you know minor variations and is there for say 256 frames or 9 seconds, you can get can get a for example a 256 pixel single colour square object that would normally take atleast 256 kilobytes into ~10 bytes(4 bytes for colour ~6for square algorithm) reduction space used by a factor of 25 600 But there are elements that always that exist outside formulas. And because of that they can take tons of space. But doing so in 64k is an extreme event where all the algorithm fall into place and there are no outside elements and no change in light which changes the pixels colour slightly.
The Faceless Master said:
:lol
this is EXACTLY what I was thinking. I remember reading a quote from Steve Jobbs when the Segway was still heavily under wraps.
It was something along the lines of "Cities will be built around this technology". Something to that effect anyway. I think the only result *I* noticed from this revolution was the chorused laughter of everyone when the first sneak peak video of it was released.
"That's it? A walker with wheels?! HAHAHAHAHA!"
God's Hand
Banned
I'm confused. If they know how to do it, why haven't they done it?
God's Hand said:
They don't. Here's a summary of the thread:
This guy claims to have created something impossible and didn't tell anyone how to do. A few people poured millions into it because they believed it was possible without knowing how it was possible. The guy died before he told anyone how he did this impossible thing. People like silver come on GAF and try to convinced everyone it's actually possible but can't prove it either way because the guy died and since the guy is dead no one will ever know if he truly made the impossible possible. Yeah.
Naked Snake
Member
EviLore said:
No kidding :lol :lol :lol
I think silver deserves a tag.
Hitokage said:
Yep. But I suppose it's possible, although unlikely, that for any given block of data, there is a procedural algorithm that could precisely generate it. Maybe if someone was able to construct a huge database of mappings of all possible block values (heh...) to a generating procedural algorithm it could work. Then they'd just look up the block in the database and only output the procedural algorithm parameters to the file. You could then take the output and repeat the process until there was no further compression realized.
this thread is funny as shit :lol
Not to contribute to the lunacy, but what about quantum data sets. If a piece of information can have multiple simultaneous states wouldn't that reduce the amount of sets required to describe something? (sorry about the complete laymen speak). but thats not compression.
Not to contribute to the lunacy, but what about quantum data sets. If a piece of information can have multiple simultaneous states wouldn't that reduce the amount of sets required to describe something? (sorry about the complete laymen speak). but thats not compression.
border said:
Short answer: No.
Slightly longer answer: No, it would never work, in any form.
Long answer: to understand why it won't work, read this FAQ: http://www.faqs.org/faqs/compression-faq/part1/section-8.html
Dsal said:
It won't work. The pigeonhole principle will guarantee that I can generate a sequence that won't be covered by any of your procedural algorithms, as long as the number of bits of input to your system is less in length than the number of bits I get back out.
http://www.dogma.net/markn/FAQ.html#Q19
LinesInTheSand
Banned
lol @ this thread
that is seriously one of the stupidest stories ive ever read
anyone who buys into that is retarded
that is seriously one of the stupidest stories ive ever read
anyone who buys into that is retarded
BugCatcher
Member
.kkrieger is great. But before that, there was this:
Back in '97, Sanction made this demo, showing a recreation of the first level of Descent in only 4K! It's even got music!
http://www.sanction.de/
Back in '97, Sanction made this demo, showing a recreation of the first level of Descent in only 4K! It's even got music!
http://www.sanction.de/
About this whole compression issue... there's an idea I had a while ago, that could shrink every movie down to less than a kb, the only problem is, you'd need really fast computers (or a lot of time) to "decompress" them.
In the file sharing program eMule, each file (<4 GB) gets its unique 128-bit MD4 hash. Wouldn't this mean that when you know the MD4 hash of a certain version of a movie, your PC could create a file starting with 00000...01, hash it, check the newly generated hash with the wanted hash and if the hashes didn't match, create the next file (00000...11). After a long time, your computer would have created the right file, the hashes would match and you would have the movie without having to download it, in a way all the information was contained in the MD4 hash...
http://www.amule.org/wiki/index.php/MD4_hash
Would this be possible with very fast computers and a lot of patience or where is my mistake?
In the file sharing program eMule, each file (<4 GB) gets its unique 128-bit MD4 hash. Wouldn't this mean that when you know the MD4 hash of a certain version of a movie, your PC could create a file starting with 00000...01, hash it, check the newly generated hash with the wanted hash and if the hashes didn't match, create the next file (00000...11). After a long time, your computer would have created the right file, the hashes would match and you would have the movie without having to download it, in a way all the information was contained in the MD4 hash...
http://www.amule.org/wiki/index.php/MD4_hash
Would this be possible with very fast computers and a lot of patience or where is my mistake?
silver said:
No.
It's a comforting idea for the tin-foil hat conspiracy theorists, but this isn't a question of different types of compression. It's a question of whether what's being discussed is possible at all, regardless of the approach.
You're like someone posting about a technique someone discovered for turning lead into gold through purely chemical reactions. "Ah," you might cry, "but of course you don't believe it's possible, or you'd be out there turning lead into gold yourselves." But it's still fundamentally impossible, to the point that we don't care what you think the technique is, or how many incomprehensible pseudo-science websites you point us at.
Same holds true with this compression issue.
Acrylamid said:
You can create a hash of any size.
So let's take your idea further - from a movie, we create a single-bit hash. Then from that single bit of data we can reconstitute the original movie through the process you describe.
Here are some movies for you to download:
Spiderman 2: 0
Constantine: 1
Battlefield Earth: 1
Star Wars Episode 2: 1
Robots: 0
Enjoy.
Mister Zimbu
Member
Acrylamid said:
Let's do a little test run on a small file. 2kb sound reasonable to you?
2kb = 2048 bytes = 16384 bits
To brute force this, you would have to try only 2^16384 combinations.
Let's say that it took one microsecond to generate a file and test a combination.
Generating the file would take 3.7E4912 years.
I'm not *that* patient.
Granted, I'm not good at math, but I think I did this right.
Warm Machine
Member
gofreak said:
This is the best part of this thread. Most impressive. 96KB!!
(I kinda got "stuck" midway through, controls are clunky etc. but..wow)
I wonder if Will Wright hired these guys? I wonder how readable their code is? :lol
The produckt relies heavily on model instancing and positioning and deforming those instances with basic values. I believe the Jak and Daxter engine does something quite similar for the creation of the environments. The level editor would be 3d-space where you bring in the environment model and duplicate it over and over again then only export the coordinates as opposed to the total mesh you've created. Very clever concept though limited to the number of base objects you bring in.
border said:
I study CS at the University of Twente. Yeah, Pieper is a bit of joke here, I've heard people refer to Sloot (Dutch for ditch) & Pieper (Dutch for potato) as "the people who could store a byte in a bit".
Pieper isn't a proper professor, he's just a PR front. He doesn't actually do anything but give the occasional nonsensical visionary lecture. The only reason why he's associated with this university is because he used to run Philips which is a huge company, research wise.
This whole Sloot deal is embarassing but no big deal.
Yup. Though even if you did that, you'd need to store the offset into pi for someone else to get the data back.Hitokage said:It'd be easier in a programming sense to just itterate digits of pi until you came across the desired values.
You can show that the offset would have to be, on average, just as long as the data you wanted to generate (and that's only if the digits of pi are completely random, which I don't think has been proven yet).
aaaaa0 said:
Yeah about 5 minutes after I posted that I figured that out
, so yeah, the precisely thing ain't gonna happen. The "offsets into procedural sequence" thing could still work too though for some of the blocks.If someone generated a big database of block values that could be procedurally generated, you could recursively try to find those patterns in the data and shrink things down. You could compress stubborn non-matching blocks with more traditional compression as well.
I might try a quick experiment with this stuff. It won't work but it'll be fun to try
CrunchyB said:
I do feel Pieper is a visionaire. I know his ideas are somewhat out of the ordinary, but in essence, his thoughts hold a lot of truth. The story is true, I don't know about those new compression tools. He had some pretty having investing behind him, especially for a one-man-plan. It has been suggested that he was killed because of his ideas.
- Status