Entity Culling 1.10.5 for Minecraft 26.2

Minecraft Java occlusion performance

Entity Culling skips rendering entities and block entities that its visibility test considers hidden. The useful outcome is not only lower rendering work; visible objects must also reappear correctly. For Java 26.2, the current creator-listed release is Entity Culling 1.10.5 for Fabric, with Fabric API required.

Current compatibility snapshot

The creator’s 26.2 version feed does not list a NeoForge build. Earlier release notes specifically delayed NeoForge 26.2 because that platform was not working. Do not install a build for another Minecraft version as a substitute, even though the broader project supports NeoForge elsewhere.

Define success with correctness and timing

Occlusion culling is useful when a client spends time drawing objects completely hidden by solid geometry. It may show little improvement in an open landscape or a GPU-limited shader scene. A correct test therefore uses a repeatable place with many entities or block entities both behind and near the edge of an occluder.

A performance win is invalid if a visible chest, mob, item frame, sign, or modded entity vanishes. Whitelists and compatibility rules exist because renderers do not all behave like vanilla objects. Test the exact object types your pack or server uses rather than assuming a single zombie proves coverage.

Install and create a false-cull fixture

1. Duplicate a Java 26.2 Fabric profile and install matching Fabric API plus Entity Culling 1.10.5.
2. In a disposable world, place one ordinary entity and one block entity behind a solid wall. Keep a third object partially visible at the wall edge.
3. Fix render distance, graphics settings, packs, shaders, camera position, and time. Record the vanilla or disabled-culling baseline.
4. Enable culling, approach the edge slowly, turn away and back, open any relevant GUI, and repeat after a client restart.
5. Only after every visible object behaves correctly, compare frame-time behavior in the same route.

Use a copy of the profile so you can compare the exact scene without editing unrelated optimization settings. Peak FPS is not enough; note stutter and object pop-in at the visibility boundary.

Troubleshoot disappearing objects

If an object remains invisible when it should be visible, capture the object type, mod owner, camera position, and smallest repeatable movement. Disable Entity Culling in the same profile to confirm the baseline. Check the current whitelist and issue tracker before adding an object to local exclusions. A broad exclusion may hide the bug while removing the intended optimization.

If the client crashes, verify Fabric 26.2, Fabric API, and the exact JAR. Preserve the log. If there is no measurable improvement, test a workload where hidden entities are actually present; do not manufacture a favorable result by lowering render distance or changing shaders between runs.

Rollback

Exit Minecraft, retain the Entity Culling config and log, remove only its JAR, and relaunch the same test world. Confirm the hidden and edge-visible objects render normally. Restore a fresh default config before copying individual whitelist changes. The mod is client-side, so rollback should not require deleting the world or server data.

Compare more than one workload

After the correctness fixture passes, use one enclosed entity-heavy route and one open outdoor route. The enclosed route tests the situation where occlusion can skip meaningful work; the open route is a control where fewer objects are fully hidden. Keep entity counts, view direction, shaders, and warm-up time fixed. If only the enclosed route improves, that is a plausible scoped result. Do not convert it into a promise for every world or modpack.

Sources and unknowns

• Entity Culling creator versions
• Creator-owned source, license, and issues

MineBrush rechecked metadata August 28, 2026 but did not render a scene or benchmark hardware. False-cull frequency, modded-renderer coverage, shader interaction, whitelist completeness, and performance remain unknown. For a different optimization that targets broader rendering paths, compare the ImmediatelyFast guide without assuming the two effects add linearly.