id: design: visual-feature cost ranking — measured ranks, per-world gates, per-viewer minimums, fallback substitution #6012

Part of the worlds design program: #5955. Wave-2 follow-up from the prior-art research (Lane B item 40); parent topic issue: #5970. Research source: workspace/scratch/worlds-capabilities-brief-2026-11-04/prior-art-research.md.

Precedent: VRChat, the strong one

VRChat computes an Avatar Performance Rank (Excellent → Very Poor) from measured metrics — triangles, materials, texture memory, PhysBones, particles, lights, shader complexity. Each user sets a minimum displayed rank; below-threshold avatars are replaced by fallbacks/impostors; Very Poor is hidden by default; Quest/mobile has a separate, stricter rank table and only SDK mobile shaders are permitted on Android avatars at all. World/instance settings add another gate layer, and per-user "Show Avatar" overrides exist for trusted individuals — the same override shape as our permission and license systems (https://wiki.vrchat.com/index.php?title=Avatar_Performance_Ranking ; https://vrcreators.net/docs/vrchat/quest-compatibility-checklist).

Scope

#5970 owns per-world enablement of assets/filters; this follow-up designs the ranking mechanism that makes gating automatic and legible:

• Measured cost metrics for sprites/filters/shaders (the shader asset format is a sibling follow-up under id: design program: worlds — advanced Roc libraries, capabilities, permissions, dynamic user functions, licensing (2026-11-04 brief) #5955) and the rank table.
• Three gate layers: per-world (author policy — e.g. filters off because invisibility cheats), per-viewer (minimum rank), per-person override.
• Fallback substitution: below-threshold content is replaced by its authored fallback (per id: design: user presentation assets — shaders, sprites by ID, filters, per-world gating, graceful degradation #5970's fallback chain), never just vanished — except where the world says hidden.

Users accept automatic visual downgrade when the rule is legible (a rank) and overridable per-person. That is the usability bar.

Done looks like

A ranking design: metrics, rank table, gate layers, substitution rules.