Tired of typing seeds over and over hoping they'll be good? Want a method that will let you find better seeds in less time? That is what Seed Theory is for.
Seed Theory is not guessing random numbers, is not hoping for good seeds, and is a just a theory that is based on an actual theory of finding similar seeds in Noise algorithms (it's also not actually even called seed theory ). Because it is a theory, I don't guarantee good results every time, just most the time. It is also important to keep in mind that biomes, trees, lavafalls and waterfalls will not be the same as the original seed. All of those are handled after noise generation, while this affects noise generation itself.
One important rule to seed theory is all seeds must be numerical values. If you type in a seed using letters and symbols, they get processed through what's called a "hash encoder" to convert it to a number and then used as a seed. The numerical value for the seed nyan will be much different than the numerical value for the seed myan, even though there was only one letter changed. So only use numerical values for this theory to work, otherwise results will vary.
Anyway, let me get to showing you how it works. The first step is to have a good seed you want to mimic, such as nyan or theedlerscrolls and maybe jocopa3. Pick one and make a new world using it.
I chose my own seed jocopa3
So, now that I have that, I'm going to find out the numerical by using iMCPEdit or any other program that can tell me the seed.
As it turns out, the numerical value is -1436175791, so what I'm going to do is split it into groups of 3 or 4 numbers each. It doesn't matter to much on how you split it up as long as you do not change the order the numbers are in. So I'm going to split it up into -1436, 175, and 791. This splitting of the seed is called the first degree of similarity, it promotes some similarity to the original seed, but not entirely.
Now, in the actual seed theory, a computerized program would split the seed in every possible way and try each one of those seeds and check to which one is the most similar to the original. For us, we will only try those three seeds and see which one we like most.
First I try -1436 and find some interesting overhangs and a cave.
Second I try 175 and find large mountains and cliffs in one side of the map, oceans and flat lands on the other.
Third I try 791 and found some weird strange shapes, flat areas and mountains.
At this point, you can either stick with the seeds you found or continue on to a second degree. The second degree of similarity involves combining the seed groups from the first degree that best fit the original seed, so for us that would give a total of 6 combinations. -1436175, 1751436, -1436791, 7911436, 175791, and 791175. You can either try all of those, or pick a few you want to see.
Keep this in mind when combining seed groups, -1436+175 = -1436175, but 175+-1436 = 1751436, not-1751436 (basically, if the negative number comes second, remove the minus sign).
Here is a good seed: 175791
Another cool seed I found from this: 1751436
I tried this theory on the seed nyan (numerical: 3396408) and got a cool seed 39 which has plenty of lavafalls, waterfalls and mountains for you to enjoy, just like the actual nyan seed.
A Partial Explanation:
It is a near fact that the shorter the seed the better chance of it being radical. Since this splits seeds up into smaller groups, there is a chance it will be good since small seeds are more radical. I don't really know the mathematics behind this though, since I don't know the exact noise generation algorithm used, but if I did, I could tell you the best seeds possible. Knowings its a 3D perlin noise algorithm, I can hopefully find something similar.
I really hope this theory works out for the best, I'll see if I can try some.
But, if I type a seed that is some random numbers, does that go through the has encoder thingy?
I have noticed that seeds end up being more mountainous and interesting without including spaces and/or the words: And, Or, But, The, So, For, Also, etc.
Rollback Post to RevisionRollBack
Follow me on Twitter! @Onlyadirector
Download my PE Adventure Map, The Legend of Koahtakula! Link Removed
I have noticed that seeds end up being more mountainous and interesting without including spaces and/or the words: And, Or, But, The, So, For, Also, etc.
Explain "Quantum Foam" it has a space and created some of the most interesting mountain formations I have yet seen. It would be cool though if we could find out how this random generation works
I have noticed that seeds end up being more mountainous and interesting without including spaces and/or the words: And, Or, But, The, So, For, Also, etc.
Coincidence. All letters and symbols are translated to numbers. The seed PB And J will not be the same as PB and J, in fact will be no where similar.
Rollback Post to RevisionRollBack
Did you know I make music? Just click my logo to listen to my awesome Electronic beats!
260 with any other combination is good for having "mini caves" I've tested it out and with 260 with another number generates maps with cave like things.
Ok guys, just type random crap for about 10 seconds in the seed section and make a new world, now do it again, and make sure not to start with the same letter. Look at that, same world, why though?
Ok guys, just type random crap for about 10 seconds in the seed section and make a new world, now do it again, and make sure not to start with the same letter. Look at that, same world, why though?
The seed 0.4.0 = the seed 0. No clue why, probably because of the . though.
Rollback Post to RevisionRollBack
Did you know I make music? Just click my logo to listen to my awesome Electronic beats!
Woah... You just blew my mind... I had always thought that a random seed was generated for your word or collection of words and then it was saved into like some database or something... I seriously thought that that was how people were able to end up with the exact same seed every time... Urgh... All this information at 10:17 in the morning is hurting my brain... I need some coffee... Well thanks for the info, Jocopa3. It helps a lot.
Also... I've been screwing around with the generation a bit and a question popped up... How does the "-" symbol affect the seed? I typed in 1790 as the seed and it was pretty normal, but then I added a "-" symbol at the beginning to make it -1790 and it was a completely different seed... It would be nice to be able to understand this...
The seed 0.4.0 = the seed 0. No clue why, probably because of the . though.
The numbers in the generator are handled as integers, when you enter a decimal the generator most likely catches a "Value does not match assigned type" error, then uses the number anyway. When a float or double is assigned to an integer the run-time environment will always lop off the decimal values.
In short, the seed 2.32493659156438957 is the same as the seed 2 in the eyes of the generator.
Also... I've been screwing around with the generation a bit and a question popped up... How does the "-" symbol affect the seed? I typed in 1790 as the seed and it was pretty normal, but then I added a "-" symbol at the beginning to make it -1790 and it was a completely different seed... It would be nice to be able to understand this...
it is the negative sign, any symbol will drastically change the seed anyways
Seed Theory is not guessing random numbers, is not hoping for good seeds, and is a just a theory that is based on an actual theory of finding similar seeds in Noise algorithms (it's also not actually even called seed theory ). Because it is a theory, I don't guarantee good results every time, just most the time. It is also important to keep in mind that biomes, trees, lavafalls and waterfalls will not be the same as the original seed. All of those are handled after noise generation, while this affects noise generation itself.
One important rule to seed theory is all seeds must be numerical values. If you type in a seed using letters and symbols, they get processed through what's called a "hash encoder" to convert it to a number and then used as a seed. The numerical value for the seed nyan will be much different than the numerical value for the seed myan, even though there was only one letter changed. So only use numerical values for this theory to work, otherwise results will vary.
Anyway, let me get to showing you how it works. The first step is to have a good seed you want to mimic, such as nyan or theedlerscrolls and maybe jocopa3. Pick one and make a new world using it.
I chose my own seed jocopa3
So, now that I have that, I'm going to find out the numerical by using iMCPEdit or any other program that can tell me the seed.
As it turns out, the numerical value is -1436175791, so what I'm going to do is split it into groups of 3 or 4 numbers each. It doesn't matter to much on how you split it up as long as you do not change the order the numbers are in. So I'm going to split it up into -1436, 175, and 791. This splitting of the seed is called the first degree of similarity, it promotes some similarity to the original seed, but not entirely.
Now, in the actual seed theory, a computerized program would split the seed in every possible way and try each one of those seeds and check to which one is the most similar to the original. For us, we will only try those three seeds and see which one we like most.
First I try -1436 and find some interesting overhangs and a cave.
Second I try 175 and find large mountains and cliffs in one side of the map, oceans and flat lands on the other.
Third I try 791 and found some weird strange shapes, flat areas and mountains.
At this point, you can either stick with the seeds you found or continue on to a second degree. The second degree of similarity involves combining the seed groups from the first degree that best fit the original seed, so for us that would give a total of 6 combinations. -1436175, 1751436, -1436791, 7911436, 175791, and 791175. You can either try all of those, or pick a few you want to see.
Keep this in mind when combining seed groups, -1436+175 = -1436175, but 175+-1436 = 1751436, not -1751436 (basically, if the negative number comes second, remove the minus sign).
Here is a good seed: 175791
Another cool seed I found from this: 1751436
I tried this theory on the seed nyan (numerical: 3396408) and got a cool seed 39 which has plenty of lavafalls, waterfalls and mountains for you to enjoy, just like the actual nyan seed.
A Partial Explanation:
It is a near fact that the shorter the seed the better chance of it being radical. Since this splits seeds up into smaller groups, there is a chance it will be good since small seeds are more radical. I don't really know the mathematics behind this though, since I don't know the exact noise generation algorithm used, but if I did, I could tell you the best seeds possible. Knowings its a 3D perlin noise algorithm, I can hopefully find something similar.
But, if I type a seed that is some random numbers, does that go through the has encoder thingy?
Hmm, have you tried this? I'm sure it reveals the seed.
iFunBox, copy level.dat, open level.dat with MCPEedit.
Download my PE Adventure Map, The Legend of Koahtakula! Link Removed
Coincidence. All letters and symbols are translated to numbers. The seed PB And J will not be the same as PB and J, in fact will be no where similar.
The seed 0.4.0 = the seed 0. No clue why, probably because of the . though.
The numbers in the generator are handled as integers, when you enter a decimal the generator most likely catches a "Value does not match assigned type" error, then uses the number anyway. When a float or double is assigned to an integer the run-time environment will always lop off the decimal values.
In short, the seed 2.32493659156438957 is the same as the seed 2 in the eyes of the generator.