slime_seed_finder

This program finds minecraft world generation seeds given enough slime chunks. Since slime chunks are 10% of all the chunks and there are 2^64 minecraft seeds, one would think that with 20 chunks you can find the seed, however thinks get interesting...

Web demo

There is a work in progress WebAssembly demo available at:

https://badel2.github.io/slime_seed_finder/

Also, a biome viewer (like AMIDST but without structures): https://badel2.github.io/slime_seed_finder/biomes.html

And a slime map which can be used to compare multiple seeds: https://badel2.github.io/slime_seed_finder/slime_map.html

Local instalation

To build this project you need to install the Rust programming language. Follow the instructions on https://rustup.rs Then, run the following commands

git clone https://github.com/badel2/slime_seed_finder
cd slime_seed_finder
cargo install --features="clap"

Now you should have the slime_seed_finder executable in your $PATH.

Usage

Put the slime chunks in a text file, one chunk per line, with the x and z chunk coordinates separated by a comma, and then run the following command to save the found seeds in the file seeds.txt

slime_seed_finder -c chunks_file.txt -o seeds.txt

If you don't have a list of slime chunks and want to try this program, use the generate option to generate slime chunks, just choose a numerical seed:

slime_seed_finder -s 1234 -o 1234.txt
slime_seed_finder -c 1234.txt -o 1234_and_some_more.txt

Ideally the program should only output one seed, but we can see this is not the case. To improve it, we can also specify non slime chunks: chunks that can't spawn slimes, with the -n flag. Since it is easy to miss one chunk, there are also options to leave an error margin: -f for slime chunks and -m for non slime chunks.

If you already have a list of possible 48-bit seeds, put them in a file one seed per line:

slime_seed_finder --candidate-seeds candidates.txt -c chunks.txt

To convert 48-bit seeds into 64-bit seeds (-j flag), put the 48-bit seeds in candidates.txt and run the program with an empty chunks.txt file:

slime_seed_finder --candidate-seeds candidates.txt -c empty_chunks.txt -j

Run slime_seed_finder --help for full details about the usage.

Building WebAssemblyy demo

cargo install cargo-web
cargo +nightly web build --target=wasm32-unknown-unknown --bin wasm_gui --features="stdweb serde1"
cp target/wasm32-unknown-unknown/release/wasm_gui.* static/
# and just open static/index.html with a web browser

Theory

Each slime chunk reduces the expected runtime by 50%, until colisions start to dominate. Non-slime chunks don't reduce the expected runtime, but are needed to remove all the false positives.

Java Random uses 48 bits, so we can only find the lower 48 bits of the seed using slime chunks. But instead of bruteforcing 2^48, we bruteforce 2^18 + 2^30. That's because those two lines are equivalent:

i % 10 == 0
(i % 2 == 0) && (i % 5 == 0)

And the parity of the output of Random.nextInt(10) depends only on the lower 18 bits, so if the parity is odd, we discard this 18-bit combination.

Cuhnt/slime_seed_finder