Vandor Labs

Full changelog: https://github.com/rheide/vandorlabs/blob/master/docs/gallery/version-1.4.md

• Replace propulsion particle On/Off with Off/Light/Medium/Heavy. Keep existing On settings and emission unchanged as Light; Medium and Heavy add density and wider plumes around the original stream. Preserve direction, particle types, connected assembly centers, saved settings and copying.

• Match diagonal screen picking and the selection outline to the solid wedge, and use cached collision slices so its empty space remains accessible.

• Make complete redstone-screen rows clickable, add a saved editable header, and constrain edited text to the visible screen width. Copy headers and complete row definitions with the Duplifier, enabling screen copying on older saved tools while preserving subsequent exclusions. Match configured diagonal screen inventory icons to their solid housing.

• Batch redstone-screen panel geometry, avoid unused animation work, and provide distinct row-control inventory icons. Keep text clear on all mounting directions, preserve power while editing row labels, and send row-status updates only when their visible highlight changes.

• Add Redstone... to the artwork pickers of existing programmable screens, inputs and consoles. Configure independent labelled channel controls on each surface, with full-row interaction, editable headers, saved definitions and Duplifier copying. Limit half-height surfaces to four rows at readable text proportions and hide their headers in the world and editor. Remove the separate alpha redstone blocks and recipes.

• Rename Programmable Viewscreen to Programmable Screen in the interface, preserving its registry ID and existing worlds.

• Configure comma-separated redstone channel lists on switches, levers, buttons and channel-aware programmable blocks. Controls operate every listed channel; consumers respond when any listed channel is active. Preserve complete lists through save/load, picked items, Duplifier settings and joined assemblies, with automatic migration of existing single-channel settings.

• Extend diagonal chunk rendering to section-boundary positions whose conservative geometry remains inside the section. Keep tile rendering for actual overhangs, with a shared routing rule for model and tile paths.

• Keep Programmable Walls and Porthole Walls visible throughout the loaded render distance, matching diagonal walls instead of disappearing beyond the tile renderer's 64-block cutoff.

• Merge compatible coplanar Diagonal Wall faces during chunk construction, preserving texture interpolation, normals, clipping and two-sided visibility. Simple walls use 20 quads instead of 44.

• Fix corrupted triangles on chunk-rendered Diagonal Walls when using the vanilla block renderer. Use the standard baked-quad vertex layout and retain the existing surfaces, normals and lighting.

• Validate the actual vanilla output buffer as well as the Forge rendering path.

• Render static Programmable Diagonal Walls through cached chunk geometry. Preserve existing surfaces, material coordinates, two-sided faces and uniform lighting. Keep the existing tile renderer for section overhangs and portholes; see the diagonal rendering notes.

• Refresh nearby diagonal-wall chunk meshes when geometry settings arrive in a tile update packet.

• Reuse bounded Static/Off screen texture identifiers and the existing loaded-neighbor lighting sampler, reducing render-loop allocation while preserving texture paths, resource-pack resolution and light values.

• Reject offset-trapdoor collision scans in loaded sections whose live block palette contains no trapdoors. Avoid per-cell world reads for ordinary entity/particle queries; retain the original scan for possible owners and unsupported storage.

• Release light and porthole group-cache world references and geometry on client-world unload, including when no later light or porthole is rendered.

• Store housing mesh settings in compact immutable snapshots and create full Forge property maps only on demand. Preserve listed/unlisted transitions, validation errors and concurrent reads.

• Reuse private per-thread corner and artwork-coordinate buffers during trapdoor rendering. Preserve independently owned geometry results for collision callers and isolate nested draws.

• Use scalar face intersection for diagonal trapdoor picking, preserving exact hit coordinates, edge tolerance and nearest-face precedence.

• Resolve saved ramp materials once per slice and reuse its face emitter, reducing repeated source-tag decoding and per-face temporary objects without changing UVs or lighting.

• Prepare immutable door quad groups once per resource bake, avoiding per-frame list assembly while preserving Forge lighting transitions, item transforms and custom-renderer fallbacks.

• Cache immutable enum-property hashes and directly read canonical trapdoor properties, avoiding repeated map hashing in rendering, collision and state queries. Preserve vanilla property equality, metadata and transitions.

• Avoid redundant chunk-dirty notifications for unchanged trapdoor group power. Preserve saved channel signals and group state, and stop physical input reads once the group is known powered.

• Reuse bounded ray-candidate coordinate lists for offset trapdoor picking, preserving hit order and current loaded-world reads. Offset collision scans reuse a mutable query position.

• Precompute fixed catalog texture names for square, unlit and storage surfaces, avoiding per-draw JSON traversal and string allocation while retaining live Custom material and atlas resolution.

• Reuse exact diagonal trapdoor collision meshes in a bounded cache, avoiding repeated subdivision and allocating only intersecting boxes. Preserve collision bounds, box order and edge contacts.

• Skip physical power scans when a screen's display mode is independent of power or its virtual channel is already powered. Joined lights stop reading inputs after the first powered source while still synchronizing the entire group.

• Reduce Block, Slab, Storage and Stairs chunk-mesh allocation with single-pass extended states, reused neighbor positions and bounded caches of common housing face lists. Preserve exact quads, face overrides, storage artwork and Forge property behavior. See the performance measurements.

• Reduce loaded redstone channel work by revalidating a known powered source before searching other members. Preserve immediate OR propagation and coalesced input handling.

• Share immutable trapdoor assembly membership after validation, reuse each draw's group snapshot, and avoid redundant corner/edge allocations and opposing-cover lookups.

• Reuse trapdoor texture-clipping layouts across frames and moving poses. Keep lighting and resource-pack texture coordinates live, with bounded cache storage and unchanged leaf geometry, artwork, and controls.