compression

Krypton Hybrid Overview Krypton Hybrid is a fork based on Krypton fnp legacy（1. 19. 2),Completed the missing optimizations in the Krypton fnp lite version, added support for the zstd compression algorithm, and made improvements to the original encoding and Netty program processing.while retaining the features of Krypton fnp and RecastLib. Differentiation Compatibility support for hybrid servers is offered. Commands are provided to monitor traffic usage. This project prioritizes using zstd as the primary compression algorithm, while also providing optimizations for entity tracking and communication encoding from the original version. Technical Details 1. Network Compression Pipeline Krypton Hybrid replaces Minecraft's default zlib codec by injecting into Connection.setupCompression . Feature Detail Primary algorithm Zstd via zstd-jni (native) Fallback algorithm zlib via velocity-native Level range 1 – 22 (default 3) Multi-threaded workers ≥ 1 activates native thread pool Dictionary Optional .zdict pre-trained file, fail-open or fail-fast Advanced tuning overlap_log , job_size , strategy , LDM + window_log Both server and client must use the same algorithm. A mismatch corrupts the session immediately. 2. Light Payload Optimization — Uniform-RLE Sky-light sections above the surface are almost always uniform ( 0xFF = all-15). Instead of writing 2 048 bytes per uniform layer, Krypton encodes them as 2 bytes. Wire format ( ClientboundLightUpdatePacketData ): – Krypton marker (1 byte) – BitSet – BitSet – BitSet – BitSet – VarInt For each sky DataLayer: + 1 byte – uniform (all nibbles == byte) + 2048 bytes – raw (fixed size, no VarInt prefix) – VarInt For each block DataLayer: same as above Savings guard: only activates when 2048 × uniform_count + raw_count − 1 0 . Max saving: ~40 KB per chunk in open-sky environments (view-distance 12 → up to 20 uniform sky layers). 3. Chunk Data Optimization — XOR-Delta Heightmaps + Biome Stream Wire format ( ClientboundLevelChunkPacketData ): – Krypton chunk-data marker (1 byte) ──── Heightmaps (binary + XOR-delta) ──── VarInt entry_count Per entry: UTF-8 key VarInt long_count long[] XOR-delta encoded (delta = raw ^ raw) ──── Block data (sections without biomes) ──── VarInt blocks_length byte[] blocks (short nonEmptyCount + PalettedContainer BlockState per section) ──── Biome data (compact stream) ──── VarInt section_count Per section: 0x01 + VarInt(biomeId) – single-value section (2 bytes vs. 3+) 0x00 + VarInt(len) + raw bytes – multi-value section ──── Block entities ──── LIST_STREAM_CODEC (unchanged vanilla encoding) Optimization Mechanism Benefit XOR-delta heightmaps Adjacent columns correlate → many near-zero longs Better Zstd/zlib ratio + ~40 B NBT overhead eliminated Biome stream split Biome data grouped; single-value sections compacted Single-value: 2 B vs. 3+ B; cross-section redundancy explo