Methodology · Updated May 2026
How ChunkScan finds Minecraft seeds worth playing
Most Minecraft seed lists are written by one person testing a handful of worlds and picking favorites. ChunkScan takes a different approach: we scan seeds programmatically, index what is near spawn, and let you search the results. No opinions, no curated top-ten lists — just data about what each seed actually contains.
As of May 2026, the database holds 257,700 scanned seeds across Java and Bedrock editions, with over 14 million structure records and 59 million biome region records. Here is how that data gets generated.
The scanning engine: cubiomes
At the core of ChunkScan is cubiomes, an open-source C library created by Cubitect that reimplements Minecraft's world generation algorithm. Cubiomes can compute biome maps, locate structures, and determine terrain features for any seed without launching Minecraft itself. This is what makes scanning hundreds of thousands of seeds practical — each seed takes seconds to analyze rather than minutes of loading a world in-game.
We run cubiomes-based scanner workers that process seeds in batches. Each worker takes a seed number, edition (Java or Bedrock), and Minecraft version as input, then generates a complete analysis of the spawn area. The scanner is multithreaded and runs continuously to expand the database over time.
What gets indexed for each seed
For every seed, the scanner records several categories of data about the area around spawn:
Structures
Every structure within the scan radius is recorded with its type and exact distance from spawn. We track 16 structure types including villages, ruined portals, strongholds, ancient cities, ocean monuments, mansions, desert pyramids, jungle temples, igloos, witch huts, outposts, shipwrecks, buried treasure, trail ruins, and trial chambers.
Biome regions
The scanner maps biome regions around spawn, recording each biome's area and distance. This powers the biome diversity score and lets you search for specific biomes near spawn — like cherry groves, mushroom fields, or deep dark.
Terrain features
Island detection, ocean coverage, elevation analysis, and cave system presence are all computed from the terrain data. The island spawn flag, for example, checks whether the land area around spawn is surrounded by water — which is why only 0.2% of seeds qualify.
Spawn point
The exact spawn coordinates and spawn biome are recorded. The most common spawn biomes are forest (19%) and plains (17.6%), while rare spawn biomes like mushroom fields are almost nonexistent.
Structure distance: why it matters
Most seed tools tell you whether a structure exists. ChunkScan tells you how far away it is. A village 50 blocks from spawn is fundamentally different from one 800 blocks away, even though both technically count as "village seeds." Distance data is what makes the difference between a useful search result and a misleading one.
Some real numbers from our database: the average stronghold distance is 1,686 blocks from spawn, with the closest we have found at 167 blocks. Mansions average about 754 blocks and appear in only 4.8% of seeds — they are the rarest structure type we track. Villages are the most useful proximity metric: about 10% of seeds have a village within 200 blocks, which is close enough to reach before the first night.
Biome diversity scoring
Each seed gets a biome diversity score based on how many different biomes are present in the spawn area. Scores in our database range from 29 to 51. A score above 45 means the area around spawn has a wide variety of biomes — forests, plains, mountains, oceans, and potentially rare biomes like cherry groves or mushroom fields all within reach. A lower score means the area is more uniform, which might be ideal for builders who want consistent terrain.
Edition and version handling
Java and Bedrock use different structure placement algorithms, so the same seed number produces different results in each edition. We scan them completely separately. The database currently holds 218,500 Java seeds and 39,200 Bedrock seeds. Within each edition, we also scan multiple Minecraft versions (1.18, 1.19, 1.20, 1.21) because structure positions can change between updates. Java 1.21 has the most coverage with 114,500 seeds.
When you select an edition and version in ChunkScan, you are searching only the seeds that were scanned with those exact parameters. The results match what you will see in-game if you use the same settings.
Map preview generation
The map preview shown for each seed is generated from the same cubiomes data used for scanning. The biome map is rendered at a resolution that shows the major terrain features — oceans, mountains, forests, deserts, and rivers — along with markers for structure positions. This gives you a visual overview of the spawn area before you copy the seed into Minecraft, so you can evaluate the layout without loading the world.
Limitations and accuracy
ChunkScan's results are as accurate as the cubiomes library, which closely matches Minecraft's own generation code. However, there are some things to keep in mind:
- Structure interiors (chest loot, villager types, portal completeness) are not scanned — only positions and types
- Very rare edge cases in world generation may not be captured by cubiomes
- Minecraft updates can change generation rules, making older scan data less reliable for newer game versions
- The scan radius is centered on spawn, so structures beyond the scan area are not indexed
We recommend always verifying important seeds in your own game before relying on them for server setups or long-term survival worlds. ChunkScan is a search and preview tool, not a guarantee.
Data scale
257,700
Total seeds scanned across both editions
14 million
Structure data points indexed
59 million
Biome region records mapped
4 versions
Minecraft 1.18, 1.19, 1.20, and 1.21 covered