Spine Optimizer

Atlas Splitter

Create child .atlas / .atlas.txt files from one shared Spine atlas without touching the texture pages.
Back to Optimizer

1. Total Data

Drop the exported Spine package. Exactly one .atlas or .atlas.txt is required. Texture pages are optional for validation and are never modified.
Drop total Spine data hereAtlas + texture pages + optional skeleton/json

2. Child Data

Drop one or more .json or .skel.bytes files. Each child creates its own atlas with the same base name.
Drop child data here.json / .skel.bytes

3. Analyze & Generate

Waiting for Total Data and Child Data.
Strict output: page declarations and exact original region blocks are preserved. Only unused regions are removed. Texture PNG/JPG/WebP/KTX2 files are never rewritten or cropped. If any child references a region missing from the Total Atlas, that child is FAILED and has no download.

Processing rules

JSON and Binary: both use the official Spine 4.2.43 SkeletonJson/SkeletonBinary loaders to deserialize the child data. The official Spine 4.2.43 TextureAtlas + AtlasAttachmentLoader resolves attachment paths against the Total Atlas. A validation proxy supplies placeholder regions only for missing names so every missing region can be reported. All loaded attachment regions and every Sequence frame region are collected and matched exactly; no basename fallback is used. Missing regions fail that child and block its download. Spine binary/runtime versions must match; Spine documents binary compatibility as version-dependent. Spine 4.2 API.