@opengamehq/skills 1.3.0 → 1.5.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +44 -0
- package/README.md +12 -4
- package/dist/.claude-plugin/marketplace.json +26 -0
- package/dist/skills/manifest.json +1409 -145
- package/dist/skills/opengame-accessibility/CHANGELOG.md +4 -0
- package/dist/skills/opengame-accessibility/SKILL.md +48 -32
- package/dist/skills/opengame-all-in-one/CHANGELOG.md +21 -0
- package/dist/skills/opengame-all-in-one/SKILL.md +195 -0
- package/dist/skills/opengame-all-in-one/references/opengame-accessibility/README.md +102 -0
- package/dist/skills/opengame-all-in-one/references/opengame-api/README.md +107 -0
- package/dist/skills/opengame-all-in-one/references/opengame-assets/README.md +168 -0
- package/dist/skills/opengame-all-in-one/references/opengame-assets/references/pipeline.md +164 -0
- package/dist/skills/opengame-all-in-one/references/opengame-audio/README.md +238 -0
- package/dist/skills/opengame-all-in-one/references/opengame-audio/helpers/audio-chain.js +325 -0
- package/dist/skills/opengame-all-in-one/references/opengame-audio/references/mix-notes.md +125 -0
- package/dist/skills/opengame-all-in-one/references/opengame-audio/references/qa-without-ears.md +177 -0
- package/dist/skills/opengame-all-in-one/references/opengame-capture/README.md +128 -0
- package/dist/skills/opengame-all-in-one/references/opengame-capture/references/pipeline.md +114 -0
- package/dist/skills/opengame-all-in-one/references/opengame-capture/references/threejs.md +161 -0
- package/dist/skills/opengame-all-in-one/references/opengame-characters/README.md +180 -0
- package/dist/skills/opengame-all-in-one/references/opengame-characters/references/first-person-body.md +108 -0
- package/dist/skills/opengame-all-in-one/references/opengame-characters/references/integration.md +144 -0
- package/dist/skills/opengame-all-in-one/references/opengame-characters/references/player-og.md +160 -0
- package/dist/skills/opengame-all-in-one/references/opengame-cli/README.md +189 -0
- package/dist/skills/opengame-all-in-one/references/opengame-deploy/README.md +223 -0
- package/dist/skills/opengame-all-in-one/references/opengame-deploy/references/json-events.md +63 -0
- package/dist/skills/opengame-all-in-one/references/opengame-deploy/references/manifest.md +89 -0
- package/dist/skills/opengame-all-in-one/references/opengame-deploy/references/packaging.md +119 -0
- package/dist/skills/opengame-all-in-one/references/opengame-depth/README.md +154 -0
- package/dist/skills/opengame-all-in-one/references/opengame-enemies/README.md +155 -0
- package/dist/skills/opengame-all-in-one/references/opengame-enemies/references/pacing-and-bots.md +108 -0
- package/dist/skills/opengame-all-in-one/references/opengame-enemies/references/threejs.md +193 -0
- package/dist/skills/opengame-all-in-one/references/opengame-front-end/README.md +185 -0
- package/dist/skills/opengame-all-in-one/references/opengame-front-end/references/launch.md +50 -0
- package/dist/skills/opengame-all-in-one/references/opengame-front-end/references/menus.md +81 -0
- package/dist/skills/opengame-all-in-one/references/opengame-game-feel/README.md +181 -0
- package/dist/skills/opengame-all-in-one/references/opengame-game-feel/helpers/spring.js +146 -0
- package/dist/skills/opengame-all-in-one/references/opengame-game-feel/references/feel-numbers.md +142 -0
- package/dist/skills/opengame-all-in-one/references/opengame-game-feel/references/killcam.md +88 -0
- package/dist/skills/opengame-all-in-one/references/opengame-jam/README.md +114 -0
- package/dist/skills/opengame-all-in-one/references/opengame-lighting/README.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-lighting/helpers/light-pool.js +185 -0
- package/dist/skills/opengame-all-in-one/references/opengame-loading/README.md +135 -0
- package/dist/skills/opengame-all-in-one/references/opengame-loading/helpers/loading-count.mjs +115 -0
- package/dist/skills/opengame-all-in-one/references/opengame-loading/references/threejs.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-look/README.md +181 -0
- package/dist/skills/opengame-all-in-one/references/opengame-look/references/matching-a-reference.md +73 -0
- package/dist/skills/opengame-all-in-one/references/opengame-look/references/painterly.md +69 -0
- package/dist/skills/opengame-all-in-one/references/opengame-look/references/post-and-lens.md +73 -0
- package/dist/skills/opengame-all-in-one/references/opengame-maps/README.md +74 -0
- package/dist/skills/opengame-all-in-one/references/opengame-maps/references/map-spec.md +77 -0
- package/dist/skills/opengame-all-in-one/references/opengame-mobile/README.md +86 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/README.md +183 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/briefs.md +293 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/day-one-questions.md +46 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/edge-cases.md +27 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/filmable.md +244 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/finishing.md +126 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/measure.md +282 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/montage-capture.md +151 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/montage-gates.md +84 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/montage-genres.md +198 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/montage-grammar.md +320 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/montage-ops.md +127 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/og-snipers-record.md +110 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/owner-eye.md +214 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/reference-study.md +293 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-make/references/song.md +64 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/README.md +193 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/briefs.md +293 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/day-one-questions.md +46 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/filmable.md +244 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/measure.md +282 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/montage-capture.md +151 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/montage-gates.md +84 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/montage-genres.md +198 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/montage-grammar.md +320 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/montage-ops.md +127 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/og-snipers-record.md +110 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/owner-eye.md +214 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/reference-study.md +293 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-plan/references/song.md +64 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/README.md +174 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/briefs.md +293 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/day-one-questions.md +46 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/filmable.md +244 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/finishing.md +126 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/measure.md +282 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/montage-capture.md +151 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/montage-gates.md +84 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/montage-genres.md +198 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/montage-grammar.md +320 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/montage-ops.md +127 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/og-snipers-record.md +110 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/owner-eye.md +214 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/reference-study.md +293 -0
- package/dist/skills/opengame-all-in-one/references/opengame-montage-review-and-improve/references/song.md +64 -0
- package/dist/skills/opengame-all-in-one/references/opengame-multiplayer/README.md +139 -0
- package/dist/skills/opengame-all-in-one/references/opengame-multiplayer/references/bot-lobbies.md +71 -0
- package/dist/skills/opengame-all-in-one/references/opengame-multiplayer/references/netcode.md +165 -0
- package/dist/skills/opengame-all-in-one/references/opengame-performance/README.md +177 -0
- package/dist/skills/opengame-all-in-one/references/opengame-performance/helpers/landing-pacer.mjs +86 -0
- package/dist/skills/opengame-all-in-one/references/opengame-performance/references/measuring.md +123 -0
- package/dist/skills/opengame-all-in-one/references/opengame-performance/references/webgpu.md +86 -0
- package/dist/skills/opengame-all-in-one/references/opengame-production/README.md +192 -0
- package/dist/skills/opengame-all-in-one/references/opengame-production/references/team.md +107 -0
- package/dist/skills/opengame-all-in-one/references/opengame-production/references/templates.md +153 -0
- package/dist/skills/opengame-all-in-one/references/opengame-quality-bar/README.md +170 -0
- package/dist/skills/opengame-all-in-one/references/opengame-quality-bar/references/cheap-tells.md +43 -0
- package/dist/skills/opengame-all-in-one/references/opengame-quality-bar/references/judges.md +95 -0
- package/dist/skills/opengame-all-in-one/references/opengame-quality-bar/references/review-prompt.md +32 -0
- package/dist/skills/opengame-all-in-one/references/opengame-rhythm/README.md +134 -0
- package/dist/skills/opengame-all-in-one/references/opengame-rhythm/references/charting.md +109 -0
- package/dist/skills/opengame-all-in-one/references/opengame-rhythm/references/timing.md +92 -0
- package/dist/skills/opengame-all-in-one/references/opengame-router/README.md +54 -0
- package/dist/skills/opengame-all-in-one/references/opengame-router/references/process.md +38 -0
- package/dist/skills/opengame-all-in-one/references/opengame-scene-audit/README.md +136 -0
- package/dist/skills/opengame-all-in-one/references/opengame-scene-audit/helpers/scene-audit.js +677 -0
- package/dist/skills/opengame-all-in-one/references/opengame-sdk/README.md +283 -0
- package/dist/skills/opengame-all-in-one/references/opengame-sdk/references/engines.md +93 -0
- package/dist/skills/opengame-all-in-one/references/opengame-sdk/references/frame.md +107 -0
- package/dist/skills/opengame-all-in-one/references/opengame-sdk/references/identity.md +111 -0
- package/dist/skills/opengame-all-in-one/references/opengame-sdk/references/leaderboards.md +115 -0
- package/dist/skills/opengame-all-in-one/references/opengame-sdk/references/radio.md +71 -0
- package/dist/skills/opengame-all-in-one/references/opengame-stability/README.md +112 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/README.md +160 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/archetypes.md +113 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/briefs.md +273 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/core.md +129 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/craft.md +299 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/edge-cases.md +27 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/filmable.md +244 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/finishing.md +126 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/gates.md +184 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/measure.md +282 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/music.md +148 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/og-clash-record.md +133 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/ops.md +154 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/owner-eye.md +214 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/review.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-make/references/title.md +92 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/README.md +172 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/archetypes.md +113 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/briefs.md +273 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/core.md +129 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/craft.md +299 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/edge-cases.md +27 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/filmable.md +244 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/finishing.md +126 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/gates.md +184 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/measure.md +282 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/music.md +148 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/og-clash-record.md +133 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/ops.md +154 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/owner-eye.md +214 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/reference-set.md +92 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/review.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-plan/references/title.md +92 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/README.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/archetypes.md +113 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/briefs.md +273 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/core.md +129 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/craft.md +299 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/edge-cases.md +27 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/filmable.md +244 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/finishing.md +126 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/gates.md +184 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/measure.md +282 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/music.md +148 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/og-clash-record.md +133 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/ops.md +154 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/owner-eye.md +214 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/review.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-trailer-review-and-improve/references/title.md +92 -0
- package/dist/skills/opengame-all-in-one/references/opengame-verification/README.md +197 -0
- package/dist/skills/opengame-all-in-one/references/opengame-verification/helpers/browser-evidence.mjs +74 -0
- package/dist/skills/opengame-all-in-one/references/opengame-weapons/README.md +179 -0
- package/dist/skills/opengame-all-in-one/references/opengame-weapons/references/handling-numbers.md +114 -0
- package/dist/skills/opengame-all-in-one/references/opengame-weapons/references/reloads-and-hands.md +96 -0
- package/dist/skills/opengame-all-in-one/references/opengame-weapons/references/sights-and-scopes.md +98 -0
- package/dist/skills/opengame-all-in-one/references/opengame-weapons/references/threejs.md +240 -0
- package/dist/skills/opengame-all-in-one/references/opengame-world-physics/README.md +141 -0
- package/dist/skills/opengame-all-in-one/references/opengame-world-physics/references/measured-failures.md +79 -0
- package/dist/skills/opengame-api/CHANGELOG.md +12 -0
- package/dist/skills/opengame-api/SKILL.md +39 -13
- package/dist/skills/opengame-assets/CHANGELOG.md +11 -0
- package/dist/skills/opengame-assets/SKILL.md +79 -31
- package/dist/skills/opengame-assets/references/pipeline.md +67 -0
- package/dist/skills/opengame-audio/CHANGELOG.md +11 -0
- package/dist/skills/opengame-audio/SKILL.md +129 -41
- package/dist/skills/opengame-audio/references/mix-notes.md +44 -1
- package/dist/skills/opengame-audio/references/qa-without-ears.md +177 -0
- package/dist/skills/opengame-capture/CHANGELOG.md +10 -0
- package/dist/skills/opengame-capture/SKILL.md +44 -20
- package/dist/skills/opengame-capture/references/pipeline.md +33 -1
- package/dist/skills/opengame-capture/references/threejs.md +17 -1
- package/dist/skills/opengame-characters/CHANGELOG.md +15 -0
- package/dist/skills/opengame-characters/SKILL.md +94 -39
- package/dist/skills/opengame-characters/references/first-person-body.md +108 -0
- package/dist/skills/opengame-characters/references/player-og.md +1 -1
- package/dist/skills/opengame-cli/CHANGELOG.md +20 -0
- package/dist/skills/opengame-cli/SKILL.md +12 -11
- package/dist/skills/opengame-deploy/CHANGELOG.md +30 -0
- package/dist/skills/opengame-deploy/SKILL.md +43 -23
- package/dist/skills/opengame-deploy/references/json-events.md +6 -5
- package/dist/skills/opengame-deploy/references/manifest.md +9 -3
- package/dist/skills/opengame-deploy/references/packaging.md +32 -30
- package/dist/skills/opengame-depth/CHANGELOG.md +8 -0
- package/dist/skills/opengame-depth/SKILL.md +61 -33
- package/dist/skills/opengame-enemies/CHANGELOG.md +10 -0
- package/dist/skills/opengame-enemies/SKILL.md +96 -40
- package/dist/skills/opengame-enemies/references/pacing-and-bots.md +108 -0
- package/dist/skills/opengame-enemies/references/threejs.md +59 -0
- package/dist/skills/opengame-front-end/CHANGELOG.md +4 -0
- package/dist/skills/opengame-front-end/SKILL.md +85 -39
- package/dist/skills/opengame-front-end/references/menus.md +42 -8
- package/dist/skills/opengame-game-feel/CHANGELOG.md +11 -0
- package/dist/skills/opengame-game-feel/SKILL.md +95 -47
- package/dist/skills/opengame-game-feel/references/feel-numbers.md +59 -0
- package/dist/skills/opengame-game-feel/references/killcam.md +88 -0
- package/dist/skills/opengame-jam/CHANGELOG.md +12 -0
- package/dist/skills/opengame-jam/SKILL.md +48 -33
- package/dist/skills/opengame-lighting/CHANGELOG.md +9 -0
- package/dist/skills/opengame-lighting/SKILL.md +85 -34
- package/dist/skills/opengame-loading/CHANGELOG.md +4 -0
- package/dist/skills/opengame-loading/SKILL.md +78 -38
- package/dist/skills/opengame-loading/helpers/loading-count.mjs +115 -0
- package/dist/skills/opengame-loading/references/threejs.md +32 -7
- package/dist/skills/opengame-look/CHANGELOG.md +9 -0
- package/dist/skills/opengame-look/SKILL.md +122 -51
- package/dist/skills/opengame-look/references/matching-a-reference.md +73 -0
- package/dist/skills/opengame-look/references/painterly.md +69 -0
- package/dist/skills/opengame-look/references/post-and-lens.md +10 -3
- package/dist/skills/opengame-maps/CHANGELOG.md +9 -0
- package/dist/skills/opengame-maps/SKILL.md +78 -0
- package/dist/skills/opengame-maps/references/map-spec.md +77 -0
- package/dist/skills/opengame-mobile/CHANGELOG.md +4 -0
- package/dist/skills/opengame-mobile/SKILL.md +18 -18
- package/dist/skills/opengame-montage-make/CHANGELOG.md +5 -0
- package/dist/skills/opengame-montage-make/SKILL.md +187 -0
- package/dist/skills/opengame-montage-make/references/briefs.md +293 -0
- package/dist/skills/opengame-montage-make/references/day-one-questions.md +46 -0
- package/dist/skills/opengame-montage-make/references/edge-cases.md +27 -0
- package/dist/skills/opengame-montage-make/references/filmable.md +244 -0
- package/dist/skills/opengame-montage-make/references/finishing.md +126 -0
- package/dist/skills/opengame-montage-make/references/measure.md +282 -0
- package/dist/skills/opengame-montage-make/references/montage-capture.md +151 -0
- package/dist/skills/opengame-montage-make/references/montage-gates.md +84 -0
- package/dist/skills/opengame-montage-make/references/montage-genres.md +198 -0
- package/dist/skills/opengame-montage-make/references/montage-grammar.md +320 -0
- package/dist/skills/opengame-montage-make/references/montage-ops.md +127 -0
- package/dist/skills/opengame-montage-make/references/og-snipers-record.md +110 -0
- package/dist/skills/opengame-montage-make/references/owner-eye.md +214 -0
- package/dist/skills/opengame-montage-make/references/reference-study.md +293 -0
- package/dist/skills/opengame-montage-make/references/song.md +64 -0
- package/dist/skills/opengame-montage-plan/CHANGELOG.md +5 -0
- package/dist/skills/opengame-montage-plan/SKILL.md +197 -0
- package/dist/skills/opengame-montage-plan/references/briefs.md +293 -0
- package/dist/skills/opengame-montage-plan/references/day-one-questions.md +46 -0
- package/dist/skills/opengame-montage-plan/references/filmable.md +244 -0
- package/dist/skills/opengame-montage-plan/references/measure.md +282 -0
- package/dist/skills/opengame-montage-plan/references/montage-capture.md +151 -0
- package/dist/skills/opengame-montage-plan/references/montage-gates.md +84 -0
- package/dist/skills/opengame-montage-plan/references/montage-genres.md +198 -0
- package/dist/skills/opengame-montage-plan/references/montage-grammar.md +320 -0
- package/dist/skills/opengame-montage-plan/references/montage-ops.md +127 -0
- package/dist/skills/opengame-montage-plan/references/og-snipers-record.md +110 -0
- package/dist/skills/opengame-montage-plan/references/owner-eye.md +214 -0
- package/dist/skills/opengame-montage-plan/references/reference-study.md +293 -0
- package/dist/skills/opengame-montage-plan/references/song.md +64 -0
- package/dist/skills/opengame-montage-review-and-improve/CHANGELOG.md +5 -0
- package/dist/skills/opengame-montage-review-and-improve/SKILL.md +178 -0
- package/dist/skills/opengame-montage-review-and-improve/references/briefs.md +293 -0
- package/dist/skills/opengame-montage-review-and-improve/references/day-one-questions.md +46 -0
- package/dist/skills/opengame-montage-review-and-improve/references/filmable.md +244 -0
- package/dist/skills/opengame-montage-review-and-improve/references/finishing.md +126 -0
- package/dist/skills/opengame-montage-review-and-improve/references/measure.md +282 -0
- package/dist/skills/opengame-montage-review-and-improve/references/montage-capture.md +151 -0
- package/dist/skills/opengame-montage-review-and-improve/references/montage-gates.md +84 -0
- package/dist/skills/opengame-montage-review-and-improve/references/montage-genres.md +198 -0
- package/dist/skills/opengame-montage-review-and-improve/references/montage-grammar.md +320 -0
- package/dist/skills/opengame-montage-review-and-improve/references/montage-ops.md +127 -0
- package/dist/skills/opengame-montage-review-and-improve/references/og-snipers-record.md +110 -0
- package/dist/skills/opengame-montage-review-and-improve/references/owner-eye.md +214 -0
- package/dist/skills/opengame-montage-review-and-improve/references/reference-study.md +293 -0
- package/dist/skills/opengame-montage-review-and-improve/references/song.md +64 -0
- package/dist/skills/opengame-multiplayer/CHANGELOG.md +15 -0
- package/dist/skills/opengame-multiplayer/SKILL.md +143 -0
- package/dist/skills/opengame-multiplayer/references/bot-lobbies.md +71 -0
- package/dist/skills/opengame-multiplayer/references/netcode.md +165 -0
- package/dist/skills/opengame-performance/CHANGELOG.md +8 -0
- package/dist/skills/opengame-performance/SKILL.md +106 -65
- package/dist/skills/opengame-performance/helpers/landing-pacer.mjs +86 -0
- package/dist/skills/opengame-performance/references/measuring.md +51 -0
- package/dist/skills/opengame-performance/references/webgpu.md +86 -0
- package/dist/skills/opengame-production/CHANGELOG.md +13 -0
- package/dist/skills/opengame-production/SKILL.md +119 -54
- package/dist/skills/opengame-production/references/team.md +107 -0
- package/dist/skills/opengame-production/references/templates.md +106 -16
- package/dist/skills/opengame-quality-bar/CHANGELOG.md +10 -0
- package/dist/skills/opengame-quality-bar/SKILL.md +94 -42
- package/dist/skills/opengame-quality-bar/references/cheap-tells.md +14 -7
- package/dist/skills/opengame-quality-bar/references/judges.md +95 -0
- package/dist/skills/opengame-quality-bar/references/review-prompt.md +10 -5
- package/dist/skills/opengame-rhythm/CHANGELOG.md +5 -0
- package/dist/skills/opengame-rhythm/SKILL.md +138 -0
- package/dist/skills/opengame-rhythm/references/charting.md +109 -0
- package/dist/skills/opengame-rhythm/references/timing.md +92 -0
- package/dist/skills/opengame-router/CHANGELOG.md +12 -0
- package/dist/skills/opengame-router/SKILL.md +14 -7
- package/dist/skills/opengame-scene-audit/CHANGELOG.md +15 -0
- package/dist/skills/opengame-scene-audit/SKILL.md +64 -47
- package/dist/skills/opengame-scene-audit/helpers/scene-audit.js +97 -19
- package/dist/skills/opengame-sdk/CHANGELOG.md +16 -0
- package/dist/skills/opengame-sdk/SKILL.md +27 -10
- package/dist/skills/opengame-sdk/references/engines.md +2 -2
- package/dist/skills/opengame-sdk/references/identity.md +10 -11
- package/dist/skills/opengame-sdk/references/leaderboards.md +11 -1
- package/dist/skills/opengame-sdk/references/radio.md +1 -1
- package/dist/skills/opengame-stability/CHANGELOG.md +4 -0
- package/dist/skills/opengame-stability/SKILL.md +59 -27
- package/dist/skills/opengame-trailer-make/CHANGELOG.md +5 -0
- package/dist/skills/opengame-trailer-make/SKILL.md +164 -0
- package/dist/skills/opengame-trailer-make/references/archetypes.md +113 -0
- package/dist/skills/opengame-trailer-make/references/briefs.md +273 -0
- package/dist/skills/opengame-trailer-make/references/core.md +129 -0
- package/dist/skills/opengame-trailer-make/references/craft.md +299 -0
- package/dist/skills/opengame-trailer-make/references/edge-cases.md +27 -0
- package/dist/skills/opengame-trailer-make/references/filmable.md +244 -0
- package/dist/skills/opengame-trailer-make/references/finishing.md +126 -0
- package/dist/skills/opengame-trailer-make/references/gates.md +184 -0
- package/dist/skills/opengame-trailer-make/references/measure.md +282 -0
- package/dist/skills/opengame-trailer-make/references/music.md +148 -0
- package/dist/skills/opengame-trailer-make/references/og-clash-record.md +133 -0
- package/dist/skills/opengame-trailer-make/references/ops.md +154 -0
- package/dist/skills/opengame-trailer-make/references/owner-eye.md +214 -0
- package/dist/skills/opengame-trailer-make/references/review.md +141 -0
- package/dist/skills/opengame-trailer-make/references/title.md +92 -0
- package/dist/skills/opengame-trailer-plan/CHANGELOG.md +5 -0
- package/dist/skills/opengame-trailer-plan/SKILL.md +176 -0
- package/dist/skills/opengame-trailer-plan/references/archetypes.md +113 -0
- package/dist/skills/opengame-trailer-plan/references/briefs.md +273 -0
- package/dist/skills/opengame-trailer-plan/references/core.md +129 -0
- package/dist/skills/opengame-trailer-plan/references/craft.md +299 -0
- package/dist/skills/opengame-trailer-plan/references/edge-cases.md +27 -0
- package/dist/skills/opengame-trailer-plan/references/filmable.md +244 -0
- package/dist/skills/opengame-trailer-plan/references/finishing.md +126 -0
- package/dist/skills/opengame-trailer-plan/references/gates.md +184 -0
- package/dist/skills/opengame-trailer-plan/references/measure.md +282 -0
- package/dist/skills/opengame-trailer-plan/references/music.md +148 -0
- package/dist/skills/opengame-trailer-plan/references/og-clash-record.md +133 -0
- package/dist/skills/opengame-trailer-plan/references/ops.md +154 -0
- package/dist/skills/opengame-trailer-plan/references/owner-eye.md +214 -0
- package/dist/skills/opengame-trailer-plan/references/reference-set.md +92 -0
- package/dist/skills/opengame-trailer-plan/references/review.md +141 -0
- package/dist/skills/opengame-trailer-plan/references/title.md +92 -0
- package/dist/skills/opengame-trailer-review-and-improve/CHANGELOG.md +5 -0
- package/dist/skills/opengame-trailer-review-and-improve/SKILL.md +145 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/archetypes.md +113 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/briefs.md +273 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/core.md +129 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/craft.md +299 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/edge-cases.md +27 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/filmable.md +244 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/finishing.md +126 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/gates.md +184 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/measure.md +282 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/music.md +148 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/og-clash-record.md +133 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/ops.md +154 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/owner-eye.md +214 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/review.md +141 -0
- package/dist/skills/opengame-trailer-review-and-improve/references/title.md +92 -0
- package/dist/skills/opengame-verification/CHANGELOG.md +11 -0
- package/dist/skills/opengame-verification/SKILL.md +113 -48
- package/dist/skills/opengame-weapons/CHANGELOG.md +11 -0
- package/dist/skills/opengame-weapons/SKILL.md +129 -55
- package/dist/skills/opengame-weapons/references/handling-numbers.md +42 -0
- package/dist/skills/opengame-weapons/references/reloads-and-hands.md +96 -0
- package/dist/skills/opengame-weapons/references/sights-and-scopes.md +98 -0
- package/dist/skills/opengame-weapons/references/threejs.md +71 -0
- package/dist/skills/opengame-world-physics/CHANGELOG.md +9 -0
- package/dist/skills/opengame-world-physics/SKILL.md +55 -24
- package/dist/skills/opengame-world-physics/references/measured-failures.md +28 -1
- package/package.json +1 -1
|
@@ -0,0 +1,197 @@
|
|
|
1
|
+
> Carried from `opengame-verification` 1.1.0. Read it when: Prove a browser game works by using it as players and the owner do. Covers the owner's exact path, defaults only from a booted tree, items closed by someone else on the live build, fixes that name their neighbour, hiding one layer at a time, counters proven on a known fault, gates that fail closed, tools that must not change the game, injected faults and senses no agent has. Use when asked does it work, test my game, verify the fix, before deploying, or when tests pass and the game is broken.
|
|
2
|
+
|
|
3
|
+
|
|
4
|
+
# Verification
|
|
5
|
+
|
|
6
|
+
Every defect worth knowing about in our own games was found by playing, not by a build,
|
|
7
|
+
a type check or a test suite, and several were hidden by the very check that should
|
|
8
|
+
have caught them: a test script that entered a sign-in code by focusing each cell by its
|
|
9
|
+
label, which no person can do; an audit that measured zero because it measured nothing.
|
|
10
|
+
So the question to ask is not only "does the check pass" but "what is making this look
|
|
11
|
+
fine?"
|
|
12
|
+
|
|
13
|
+
Three newer games sharpened that. On OS the owner found about 93% of new defects. On LC
|
|
14
|
+
about 45 of 190 lessons were the measuring tool changing what it measured, and every
|
|
15
|
+
complaint the owner had to repeat lived on a path nobody tested. On OH and LC no agent
|
|
16
|
+
heard a sound or watched a film at speed, so the owner's ears and hands were the only
|
|
17
|
+
test of feel. The rules below are how to get closer to what the owner sees, and how to
|
|
18
|
+
say so where you cannot.
|
|
19
|
+
|
|
20
|
+
Rules carry five tags; `opengame-quality-bar` explains them and the game codes.
|
|
21
|
+
|
|
22
|
+
## Run a player check
|
|
23
|
+
|
|
24
|
+
Use [helpers/browser-evidence.mjs](helpers/browser-evidence.mjs) from a project with
|
|
25
|
+
Playwright installed. The helper starts a muted headless browser, clears network
|
|
26
|
+
cache, clicks Play, holds a key, reads position changes and captures the page.
|
|
27
|
+
Run it through your machine's job queue when one is required.
|
|
28
|
+
|
|
29
|
+
Create a synthetic scenario file with `url`, `playSelector`, `key`, `screenshot`,
|
|
30
|
+
`statePath` and `positionPath`. Example paths are `["__og", "state"]` and
|
|
31
|
+
`["player", "position"]`; use the game's actual read-only contract. Supply
|
|
32
|
+
`criteria.movementMin` in game units and `criteria.frameP95MsMax` from a measured
|
|
33
|
+
baseline and the target hardware. `holdMs`, `soakMs` and `cpuThrottle` set the budget.
|
|
34
|
+
|
|
35
|
+
Run `node <installed-skill>/helpers/browser-evidence.mjs scenario.json`. Exit zero
|
|
36
|
+
means movement, frame time and browser error criteria passed. Inspect the screenshot
|
|
37
|
+
and confirm the level and phase yourself. A script result alone cannot establish
|
|
38
|
+
that the correct game is visible. Re-run with a blocked player or delayed frames to
|
|
39
|
+
verify that the measurement fails. Keep credentials out of the scenario and capture.
|
|
40
|
+
The helper clicks through Playwright, which grants the page user activation, so it does
|
|
41
|
+
not exercise the strict autoplay path; the owner's-path rule below covers that.
|
|
42
|
+
|
|
43
|
+
## Rules
|
|
44
|
+
|
|
45
|
+
### Capture the frame one second after Play
|
|
46
|
+
- **Invariant:** The first moment of play is seen, through the real menu, on every change that could affect it.
|
|
47
|
+
- **Technique:** Drive a real browser from a cold load through the title and menu with real clicks and keys, press Play, wait one second, and capture the composited page (canvas and HUD together). Serve the build without caching so the browser runs the code on disk.
|
|
48
|
+
- **Reference configuration:** ST captures the frame after Play on every checkpoint; our originals' rule is to serve with no-store, because a caching static server ran stale modules for three separate agents. LC's director booted the owner's served build every 30 minutes, unflagged, at the owner's window and at 1280 x 720, with every cheat saved, a first and a second run, and logged console errors, failed requests, seconds to the menu with the machine load, and press-to-play time.
|
|
49
|
+
- **Applies when:** Every game.
|
|
50
|
+
- **Verify:** The capture shows the game world, the HUD and the player in control, not a loading screen, a black frame or the menu.
|
|
51
|
+
|
|
52
|
+
### Test on the owner's exact path
|
|
53
|
+
- **Invariant:** A fault the owner reports is reproduced, and its fix seen, on the path the owner actually takes: the build the owner is served, begun the way the owner begins it, with the owner's saved settings and cheats, a return visit, the owner's window size and pixel ratio, the owner's line speed, a real key press, the GPU shared the way the owner's is, the browser's strict autoplay policy, and a second run in the same page.
|
|
54
|
+
- **Technique:** The first time the owner reports a fault you cannot reproduce, write the owner's path down as a recipe and make it a standing capture: a cold browser with the owner's save seeded by an init script and checked on a still, the line throttled from navigation, the owner's window and pixel ratio, one real Enter, a wait before any second press, and a timeline of phase and picture from the press. Slow each async load in turn. Clicks sent through script evaluation do not count, because they grant user activation and the autoplay and pointer-lock refusal paths never run. Prove the fault reproduces on the recipe before fixing it, and is gone after.
|
|
55
|
+
- **Reference configuration:** LC's owner played at about 1790 x 860 with every cheat saved, over a laptop line, with a screen recorder holding the GPU, under strict autoplay. For three days the team tested two saved cheats, first visits, an autostart flag and 1280 x 720, and each difference hid a fault the owner then reported: six perk injections put the gun aside for 3.5 s at spawn (reported four times), a return visit sent New run back to the menu, 170 MB of late fetches queued the menu's 21 KB script, the wider window lost the lowered pistol's hands, and every streamed loop died at its first hand-over under strict autoplay, so there was no rain. "None of my 34 earlier spawns was begun by a real press." With the recipe standing (20 and 5 Mbit, 40 ms), spawns came right 20 of 20, 22 of 22 and 11 of 11. OH's blank board appeared only with gesture-gated autoplay, a pixel ratio of 2, a 1770 x 905 window and one async load slowed; the owner reported it twice, and the fix was published 81 min after the report.
|
|
56
|
+
- **Applies when:** Every game whose owner or players report from a setup the team does not test by default, above all games with saved settings, cheats, return visits or streamed audio.
|
|
57
|
+
- **Verify:** Re-run the owner's reported fault on the recipe before the fix: it reproduces. Run the recipe on the fixed build three times in one page: it is gone each time.
|
|
58
|
+
|
|
59
|
+
### Main is what the owner plays
|
|
60
|
+
- **Invariant:** The build the owner plays changes only by defaults that were seen right. New looks and behaviours go in off behind a flag, and a default flips only in a commit whose files match, byte for byte, a build that was booted plain, first and alone, and whose capture someone opened at full size beside a take-back control. Reports say "on main and seen", "on main, unseen" or "behind a flag".
|
|
61
|
+
- **Technique:** Put every visible change behind a URL flag and keep the old behaviour as a take-back flag. Build the default from an archive of the commit plus your own files, never from a shared working tree, and hash the built files against the commit. Boot it plain (no flag, no cheat, storage cleared, a real key) before any flagged page, with no console errors and no failed requests, and say in the commit message that this is the tree that was booted. Boot the fallback renderer too ("The fallback renderer is booted every time a default changes" in `opengame-stability`). Every round, capture main with no flags at the owner's first views. Anything found worse in play goes back to its last good state at once, and the report says so.
|
|
62
|
+
- **Reference configuration:** LC's preview rebuilt from main every 4 minutes. Five broken visuals (a black frame on raising the sights, a pink magazine, an orange ball for a head) shipped from passing unit tests: "Tests prove logic; only a picture proves a picture." Two owner corrections sent in the same minute both came from main while the team judged a flagged build. LC defaults that skipped the ritual reached the owner: a setting pushed by another lead's sync before its film was read, carried items glowing flat green for 35 minutes from "a film of arms only", a cant applied to 34 rifles with 2 seen (one wrist went from 39 to 81 degrees), and a thumb standing straight up for 30 minutes ("I HAD NOT OPENED THOSE FRAMES"). An OS weak-machine default drew a black sky on every map for three builds after a checkpoint committed it mid-test.
|
|
63
|
+
- **Applies when:** Any live preview or continuous deploy, and any project where several agents commit to the branch the owner plays.
|
|
64
|
+
- **Verify:** Every visible default since the last update maps to a capture path and a take-back flag, and the take-back restores the old behaviour in the served build.
|
|
65
|
+
|
|
66
|
+
### The build under test is the build that was served
|
|
67
|
+
- **Invariant:** Every result names the build it ran on, and the files the browser loaded match that build's receipt.
|
|
68
|
+
- **Technique:** When a build is cut, write a receipt: source commit, build id and a hash of every shipped file. Have the game report its build id in the state contract. Before testing, and after every deploy, hash the files the server returns and compare them with the receipt; any mismatch fails the run and names the file. Rebuild every target (desktop, phone, threaded and not) from the same source each round.
|
|
69
|
+
- **Reference configuration:** OC's R162 served a wasm that did not match its receipt, and the launcher around the check hid the failure. Its iOS bundle had last been built on 17 Oct and shipped stale every round until R160. A claim that two builds matched came from a stale binary copied after a killed build, and shader edits went untested for 3-8 builds because the test build never recompiled them. OS once shipped 7 stale cooked files (a sleeve served at 169,900 bytes against 175,240 cooked) because 90 build snapshots were hard-linked to the live cook.
|
|
70
|
+
- **Applies when:** Every game with a build step, a CDN, a cache or more than one build target.
|
|
71
|
+
- **Verify:** Change one shipped file after the receipt is written: the check fails and names that file.
|
|
72
|
+
|
|
73
|
+
### The change is seen running in a built round
|
|
74
|
+
- **Invariant:** A fix counts as done once it is seen or heard in a build from the merged source, reached through real input, with a log line or state value showing its code path ran.
|
|
75
|
+
- **Technique:** Give each change a marker it emits when its path runs (a counter, a log line, a state field). After the round is built, play to it through the real menu and capture it. A change whose marker never fired is not done, however the code reads, and a change too small to see at play size does not count: make it bolder, or say it does not read. Finish endings with real deaths and real wins, with god mode off.
|
|
76
|
+
- **Reference configuration:** OC had about 25 reports in a row say none of the work had been seen in a build, and complaints stayed open 4-5 rounds because fixes landed minutes after the snapshot. Visuals gated on a health value hid the menu chains for 6 rounds; cutscenes set their camera too late in the frame and were "probably never shown"; after a "no proof captures" speed rule, the upright-corpse fix shipped on a path that ran zero times.
|
|
77
|
+
- **Applies when:** Every change reported as fixed, most of all visual and audio changes and anything behind a condition.
|
|
78
|
+
- **Verify:** The report links the capture and the marker count from the built round.
|
|
79
|
+
|
|
80
|
+
### An item closes only when someone else sees it live, in the owner's conditions
|
|
81
|
+
- **Invariant:** An owner's item is closed only on evidence from the exact released build, gathered by someone who did not build the fix, in the owner's own conditions (map, weather, input device, refresh rate, render scale, saved settings), measuring the symptom the owner described rather than the suspected cause.
|
|
82
|
+
- **Technique:** Before new work each round, a verifier takes every item marked built, captures it on the live build with real mouse, keys and pad at the owner's refresh rate, and closes it with a frame path or reopens it with a reason. Source checks and byte checks guard wiring and are never cited as proof. A predecessor's "done" enters a new register as "verify".
|
|
83
|
+
- **Reference configuration:** OS's sweep of one release reopened 23-24 of 126-127 built items (18-19%), and an earlier sweep reopened 25 or more, including items accepted as "exactly right". One OS item closed at 0 footstep cues while the owner heard an ambience bed generated with "gravel crunching" in its prompt. A device fix passed a scripted capture at 0.03 px; the owner: "killstreak devices still shake/shift a lot when u use them". The test had no mouse look.
|
|
84
|
+
- **Applies when:** Any backlog with a verify state and more than one builder.
|
|
85
|
+
- **Verify:** Ten sampled done rows each cite a capture from the released build showing that row under the owner's named conditions.
|
|
86
|
+
|
|
87
|
+
### A pass on the literal complaint is not a pass on the bar
|
|
88
|
+
- **Invariant:** An owner's note closes when a hostile reader judges the quality under it, not when the named symptom is gone, and its check bounds what looks natural, not only what is impossible.
|
|
89
|
+
- **Technique:** Re-read the owner's words beside the fix and ask what the owner will say next. Write the note's bar in the owner's terms (human, readable, as loud as, as dark as) and test that, not the narrowest symptom. A joint-limit check passes anything possible; add the natural range. Close with the owner's quote beside a capture at the owner's own view.
|
|
90
|
+
- **Reference configuration:** LC reported "0 joint faults over 1,825 arm-states" and a playtest "seen fixed", then the owner: "hands also still look fucked up holding the gun". Wrists at 67-78 degrees had passed a limit of about 95, and the natural band of about 20 degrees came too late. The LC notes that came back were those fixed on a path the owner does not play, fixed literally (a menu effect returned three times), tuned to scripted players, or made default while known to be wrong from the owner's view.
|
|
91
|
+
- **Applies when:** Closing any note the owner raised.
|
|
92
|
+
- **Verify:** The closing note quotes the owner and shows a capture at the owner's view, and a reader who did not do the fix marks it against the quote.
|
|
93
|
+
|
|
94
|
+
### Every fix names the neighbour it could break
|
|
95
|
+
- **Invariant:** A fix ships only when its paired condition also holds, on the same build, through every caller of any shared function it touched, in both directions of failure.
|
|
96
|
+
- **Technique:** Write the pair into the brief: faster boot and every item still drawn; fewer audio dropouts and no added latency; a darker night and the lit room still playable; louder rain and the shot still on top. Re-audit the previous fix's area each round. For a look tuned by critique, accept a change only if it moves toward the reference value, not only away from the last complaint. Smoke through every caller of a shared function, not one.
|
|
97
|
+
- **Reference configuration:** About 25 OS fixes caused the next defect in 5 days: a buffer against audio dropouts built a backlog of about 690 ms ("gun shot sounds are delayed now?... in latest build actualyl it seems all audio"), texture release crashed decals, a boot deferral left the menu at "mode cards 0/6", a zoom binding removed aim ("also remember u need ot be able to eds with E as well! why did we remove that?"), and a shared function smoked through one of its three callers crashed a release. Judges moved an OS scope lens from 648 to 961 to 886 px. On LC a lowered pistol lost its hands in the owner's wider window, rain raised to be heard rose over the gunshot (a +17 dB raise put a drop at 0 dBFS), and a z-fighting mend built a slat wall that bodies walked through, on main for 8 hours.
|
|
98
|
+
- **Applies when:** Every fix, above all performance, memory, audio, lighting and shared-function changes.
|
|
99
|
+
- **Verify:** The proof sheet shows both conditions before and after on one build, plus a smoke through each caller.
|
|
100
|
+
|
|
101
|
+
### Hide one thing at a time before mending anything
|
|
102
|
+
- **Invariant:** Before a visual defect is mended, the team knows which object, material term, light or layer draws it, shown by the defect going away with exactly that one thing hidden.
|
|
103
|
+
- **Technique:** On a frozen frame, hide groups, then meshes, then zero material terms or lights one at a time, reading each hide back from the page and proving the hide tool on one frame first. Move objects to a hidden layer rather than setting a visibility flag other code writes back. Ask "drawn and unlit?" before "missing?". Strip a suspect material to a plain dark rough grey to test the light, then paint the suspect part a flat colour to confirm. A judge's name for a defect is a symptom, not a cause.
|
|
104
|
+
- **Reference configuration:** On LC a shoulder "hole" mended three times was drawn, unlit cloth; pale magazines blamed on texture metalness were the lens fill light; "z-fighting" was sub-pixel snow noise ("nine guesses before that were wrong"); hiding took 2 short jobs where guessing had taken 4 jobs and 9 trials. An LC probe stripped to 3% albedo stayed pale after about 3 h of material restyling, which proved the auto-exposure. OH's judges named the wrong layer nine times ("a 'mouth' was a snowflake"), and OH's rule of proof at play size, with layers hidden first, broke an 8.43 plateau.
|
|
105
|
+
- **Applies when:** Every visual defect whose cause is not certain, in any renderer.
|
|
106
|
+
- **Verify:** The defect report names the hidden thing and includes two frames, with it and without it.
|
|
107
|
+
|
|
108
|
+
### Check what a capture shows before anyone judges it
|
|
109
|
+
- **Invariant:** Every frame or clip used as evidence is confirmed to show the build, screen, camera and moment it is labelled with, with the player alive, in frame and in control, and every asset loaded, before it is scored or compared.
|
|
110
|
+
- **Technique:** Stamp each capture with the build id, the screen name, the camera mode, the player's health and the loaded-asset count read from the state contract. Run a validator that rejects any capture where those disagree with its label, where a loading, death, title or pause screen shows, or where the HUD has not settled. Report how many captures it rejected and why. How a capture stamps and proves itself is in `opengame-capture`.
|
|
111
|
+
- **Reference configuration:** OC scored 15 death screens as a gameplay baseline, and its first parity round was game-over frames that still scored 14-19 on the metric. A "shoulder view" was first person, a "gun" shot was the title screen, "profile" shots showed the main menu, "after" frames were taken before models loaded, and a folder named R121 held R86; coverage reported all valid on those frames. OC waits 300 ms for the UI to settle. OS's first round had 6 of 10 maps at 1 HP and put the victory screen in all three scope slots; four "gameplay" stills aimed at the sky, five judges scored a tree canopy, and 13 of one map's 39 checklist cameras sat inside walls, roofs or cliff tops.
|
|
112
|
+
- **Applies when:** Every capture that a person or a script will score, compare or cite as proof.
|
|
113
|
+
- **Verify:** Plant a death frame, a loading frame and a frame from an older build in a capture set: the validator rejects all of them by name.
|
|
114
|
+
|
|
115
|
+
### A check proves its scenario before its result
|
|
116
|
+
- **Invariant:** A capture, test or bot run proves nothing until it has shown its subject was in frame, lit, alive and in the intended state, that every value it set reads back from the page, and that a failure it reports happens through real play and not only inside the script.
|
|
117
|
+
- **Technique:** Each scripted check asserts its preconditions (alive, phase is play, subject in view and inside the light, the right counter path) and stops with the reason when one fails. Read back every value a script sets before measuring (position, health, camera distance, loadout) and stop if it differs. Review test scripts like game code, and fail fast when the run-start marker is missing. Drive bots with real input and count what they did (shots, hits, kills, time at the target) from game telemetry. Use staging (god mode, an empty map, a jump to any round) to reach a moment, and judge feel only on unstaged runs.
|
|
118
|
+
- **Reference configuration:** In one OC range most "failures" were test-script errors: a fixture placed 10 m inside a base, the player sent back to spawn, script overruns read as crashes, and a regex that matched only audio cues. One OC bot hit the boss's weak points for 8 of 489 s; another fired 240 shots at the sky; OC's pass bar became 546 shots and 17 kills, counted from telemetry. LC re-shot about a dozen captures (an empty grey frame, a tilted camera, the wrong spawn, a dead player whose audio was read on the results screen). Three LC leads' "0.3 m" stills were taken at 0.63-0.71 m, "the blow took 34 and he stood" because a set health was never read back, and a wall-treading count stood its test zombies outside every district the flow field reached.
|
|
119
|
+
- **Applies when:** Every automated playtest, bot run, scripted capture or eval.
|
|
120
|
+
- **Verify:** Break a precondition on purpose: the check fails rather than passes. A bot run whose counts fall under the bar fails as a test-script failure.
|
|
121
|
+
|
|
122
|
+
### Gates fail closed
|
|
123
|
+
- **Invariant:** A gate that fails, times out, crashes, did not run or judged nothing reports failure (or incomplete, which never reads as pass), and its exit status says so.
|
|
124
|
+
- **Technique:** Propagate exit status through every wrapper (no trailing command that resets it, no pipe that drops it). Treat a timeout, a missing validator, a missing result file and skipped tests as failure. Require the event a check judges (the respawn, the hit, the screen id) before it can pass. Read "it ran" markers written by the game, never lines the test script prints itself. A step passes only when its exit status and the receipt it wrote this run agree; any disagreement is incomplete. Any count a tool prints that could reveal lost content (stand-ins, dropped colliders, skipped items) gets a bar and fails the gate when it rises. Inject one known failure into each gate when it is written and whenever it changes.
|
|
125
|
+
- **Reference configuration:** OC's gate wrappers ended in an `echo` and exited 0 after a failure; failed acceptance checks still wrote PASS. Timeouts counted as done, a missing validator returned pass, and frame-count QA passed a film cut with 402 frozen frames. OS's UI check read a menu backdrop as a modal and checked nothing on any menu for half a day, and one OS tour exited 1 while its receipt said ok. An OS map had no wall collision for 4 days under a green gate while its log printed "2052 stand-in boxes ... no longer collide": "nothing read the number". LC pushed "36 unit tests pass" through a pipe with 2 failed, and conditional skips printed "2 skipped" and exited 0.
|
|
126
|
+
- **Applies when:** Every script, gate and wrapper whose result anyone acts on.
|
|
127
|
+
- **Verify:** Each gate's history shows one deliberate failing run with a non-zero exit, and a run with the judged event removed exits incomplete.
|
|
128
|
+
|
|
129
|
+
### Every check can fail, and is proven again whenever it changes
|
|
130
|
+
- **Invariant:** Every automated check has been seen to fail on a defect it is meant to catch, reports how many things it examined, is proven again whenever its filters or thresholds change, and is never loosened to pass.
|
|
131
|
+
- **Technique:** Plant the defect (a floating crate, a missing file, a busy-wait in the frame, a synthetic speck, the old build) in the same job as the check, and compare the count with a picture you opened. A new gate runs warn-only until it fails the planted case and passes several healthy runs, then hardens. On a filter change, plant a defect below the new tolerance and log what each filter drops. List a changed threshold with its old value, new value and reason, and never change a check and the thing it guards in one pass. A check that examined nothing fails. Audit with a measure that does not read what the fix changed ("Every check proves itself first" in `opengame-scene-audit`).
|
|
132
|
+
- **Reference configuration:** DK's coplanar audit measured 0 m² on every prop until a planted defect showed its clipping was inverted. AW plants a defect for every map check in every map, and its plants found two bugs in the checks. OC's corpse-spin metric improved with heavier corpses while the corpses looked worse. OS's integrity audit reached zero by adding filters ("the zeros came from loosening the audit rather than changing the map") and was accepted only once planted holes were all flagged; OS hardened new checks after 5 of 5 healthy runs and a failing negative control. LC counters that lied: "hands up" read 0.0% while the film showed it in 6 of 12 frames; a white-glove detector at 244 levels could never see a fault at 200; a probe of the page's copy of uniforms found 0 of 28,452 lost writes where the GPU lost 220. LC changed four mix checks in the same pass as the mix, and two went back once listed. OS's summary of its lying checks: "the check reused the fix's inputs, and the capture path differed from the player's"; its plant grounding pass and its check both read the same invisible card corners, so the log said "0 still floating" while about 3,600 plants hung.
|
|
133
|
+
- **Applies when:** Every audit, gate and test the game relies on.
|
|
134
|
+
- **Verify:** Each check's output names its plant and its examined count, and a recorded run on known-bad input that reads non-zero is rerun whenever the check changes.
|
|
135
|
+
|
|
136
|
+
### The measuring tool does not change the game
|
|
137
|
+
- **Invariant:** No code path in the game reads whether a capture, debug or test hook is present, except hooks that declare themselves and change only what they declare, and every reported time comes from a run in which nothing reads the picture back and nothing profiles the page.
|
|
138
|
+
- **Technique:** Grep the game for every read of a hook object and route each through one explicit declaration (a film flag, a stepped clock). Split each sitting into a picture run (frames and counts only) and a bare run (the page's own timing rows, long tasks, landing times), with the machine's load beside it. Name the tool's own cost before blaming the game: a long task inside a canvas read-back belongs to the tool. A profiled boot tells you whose time it is, never how long. Diff the game's own decisions (pacing rule, tier, clock source) per frame with and without the test kit. `opengame-performance` ("Timings come from a bare run") has the timing half.
|
|
139
|
+
- **Reference configuration:** LC's engine skipped its frame-pacing rule whenever an after-frame hook was attached, and the capture kit attached one to every page: four director sittings and three leads saw 56-60 fps while the shipped default drew 4 fps (35 frames in 8 s) for the first 8 s of every run, for an hour after the owner was told it was fixed. An LC blocker ("New run stands 1.3-3.2 s") was the kit's own read-back ("Of a press that stood 2,263 ms, 2,249 were the playtest kit's own frame reader"); bare, the press took 232-354 ms, and another "game" long task was 5,031 of 5,048 ms inside `toBlob`. OS's headed virtual display added about 46 ms a frame (headless on the GPU took one map from 9.9 to 22.3 fps), and a CDP profile put menu audio at about ten times its real cost.
|
|
140
|
+
- **Applies when:** Every game with a capture or test kit that installs anything in the page, and every time measured through captures, above all on a shared GPU.
|
|
141
|
+
- **Verify:** Boot the same build with and without the test kit and diff the game's decisions per frame: identical. Re-run a timed claim with read-back off: if the number moves by more than its spread, the earlier number is void.
|
|
142
|
+
|
|
143
|
+
### Prove the instrument before believing its number
|
|
144
|
+
- **Invariant:** A number counts only once the run is shown to have exercised what it claims (tier, renderer, cache state, frame source, presentation path, machine load) and the instrument's own noise is smaller than the change claimed. A disagreement between two runs of one check is a fact about the runs before it is a fact about the game.
|
|
145
|
+
- **Technique:** Have the game log the mode it actually took and copy that line into every receipt. Count frames the engine actually ran or presented, not animation-frame callbacks. Run an A/A for the noise floor, compare only same-tree twins in one capture mode, and interleave A and B inside one boot when boots swing. Cold means a fresh profile and an empty GPU driver shader cache. When results disagree, compare runtime versions, machine load, throttle actually achieved, browser and GPU before changing code, and run heavy suites one at a time.
|
|
146
|
+
- **Reference configuration:** Eight honest reports in OpenGame's own repository disagreed for two days because they ran on two Node versions. AW measured a requested 4x throttle at 1.56x on a saturated machine. OC held a build as a regression when the cause was other browsers on the machine at 131%, 107% and 55% CPU, a background-priority daemon cut headless Chrome from 105 to 5.6 fps for about 4.5 h, and OC rejects a perf run when competing processes use more than about half the cores. OS's weak-tier runs used the full renderer for about two days because a device-memory shim never reached the game's worker; a "fresh profile" boot of 5-9 s was warm, and true cold was 45.6 s; 10,297 frame callbacks held 5,053 engine frames, voiding every earlier fps figure; a windowed A/B shipped a 46% bot regression as a 13% gain for three releases against a 20% A/A floor. LC's multiplayer pages read 56.9-57.3 fps from callbacks and drew 17.4-18.5, and busy-machine "wins" of 26.2 against 32.5 ms were 18.4 against 18.2 on a quiet card. "The instrument was the bug more often than the code" (OS).
|
|
147
|
+
- **Applies when:** Every performance, load, memory, flicker or audio number, every scan that reports zero, and every time results disagree.
|
|
148
|
+
- **Verify:** Each receipt quotes the mode line, cache state, frame source and machine load for each run compared, and the A/A spread is under half the claimed change.
|
|
149
|
+
|
|
150
|
+
### A phone pass comes from a real phone
|
|
151
|
+
- **Invariant:** A claim that the game works on phones is backed by a run on a real phone in its real browser, and emulation results are labelled as emulation.
|
|
152
|
+
- **Technique:** Put a mainstream phone in the loop from the first playable build. Cold-load it several times in a row through the real menu, play a full match, and record memory, frame rate, audio and any reload. Desktop emulation of a phone checks layout and touch input and nothing else.
|
|
153
|
+
- **Reference configuration:** Every OC phone gate passed in Chrome's phone emulation while real iPhones crash-looped during load after launch. Emulation also hid missing in-game audio and a real frame rate of 10-20 fps.
|
|
154
|
+
- **Applies when:** Every game that claims phone support.
|
|
155
|
+
- **Verify:** The phone report names the device model, OS version and browser, and links a recording from the device.
|
|
156
|
+
|
|
157
|
+
### Input moves the player, not just the screen
|
|
158
|
+
- **Invariant:** Pressing a movement or action key changes the game's state, not only the pixels.
|
|
159
|
+
- **Technique:** Read the player's position (or the game's equivalent) before and after holding a key for a fixed time through real key events, and check the change against the expected speed. Do the same for the main action and its consequence.
|
|
160
|
+
- **Reference configuration:** A rule shared across DK, MC, AW and ST. The failure it prevents: a camera that bobs or sways on input makes a frame-difference test pass while the player is stuck against geometry.
|
|
161
|
+
- **Applies when:** Every game with player control.
|
|
162
|
+
- **Verify:** Hold forward through real key events and see the position change by roughly speed times duration; hold it against a wall and see it stop.
|
|
163
|
+
|
|
164
|
+
### Timing is proven with real input, and failure paths with injected faults
|
|
165
|
+
- **Invariant:** Timing and failure paths are proven by real events and real failures: real key presses through the browser's input path, measured on the game's own clock, and every fault a player's machine or network can produce injected on purpose, each ending in a message and a way out.
|
|
166
|
+
- **Technique:** Press real keys while adding stalls and CPU throttling, and read the judged time from each key's own event time against the game's clock (the audio clock in a rhythm game). Inject the failures: failed and slow responses, a missing level or chart file, no WebGL or WebGPU, storage that throws, a hidden tab, a lost GPU device, and many rounds in a row with heap counts. Run a seeded careless-player fuzzer (keys, chords, odd text, resize, blur, pointer lock, network drops) and shrink each finding to its shortest repeat.
|
|
167
|
+
- **Reference configuration:** OH's real presses landed at +1 ms mean with p95 under 5 ms and no misses under stalls. OH's launch QA (real presses, thirteen injected faults, every song and level) found the count-in off the beat on every song, a missing chart replaced without a word by a guessed one, two lanes sharing a key, touch-only screens stuck in emulation, a blank page without WebGL2 and an 8.7 MB-a-song heap rise. OS's soak (every map through the real picker, a full match, injected corruption and GPU loss) caught crashes on the 4th and 8th map and an out-of-memory on the 20th, and its fuzzer found a key-combination trap the owner had hit and a black screen after a 1 x 1 resize. `opengame-stability` has the soak's budgets.
|
|
168
|
+
- **Applies when:** Before any launch, and whenever timing or a failure path changes.
|
|
169
|
+
- **Verify:** Each injected fault shows a message and a way out, and the real-input timing report gives the mean and p95 against the game's clock.
|
|
170
|
+
|
|
171
|
+
### A state contract to assert against
|
|
172
|
+
- **Invariant:** A test can read what the game believes (phase, player position and health, score, what is loaded, which mode it took) without reaching into its internals.
|
|
173
|
+
- **Technique:** In development builds, expose a small read-only snapshot such as `window.__og.state` with the few values tests need, built by the game from its real state each frame. Have the game log the tier, renderer and clock it actually chose. Never let tests write to it.
|
|
174
|
+
- **Reference configuration:** AW reads the loaded map's identity before measuring after a failed load fell back without an error to a stand-in arena that measured faster than the real map. OS's checks that held up read facts the game logged on purpose (log lines, a dump of every UI draw command, debug render views); the pixel-only checks were the ones that cried wolf.
|
|
175
|
+
- **Applies when:** Every game that will be tested by a script or an agent.
|
|
176
|
+
- **Verify:** The contract reports the real loaded level and phase, and a planted failure (block a required asset) shows up in it.
|
|
177
|
+
|
|
178
|
+
### Nothing built is left uncalled
|
|
179
|
+
- **Invariant:** Every module, component and asset in the game is reachable from where the game starts.
|
|
180
|
+
- **Technique:** Grep for each new module's name and each generated file's name in the source; follow imports from the entry point. Ban optional calls (`?.()`) at seams between parts of the game, because a call that does nothing, with no error, when its target is missing hides exactly the half-finished handover that breaks games.
|
|
181
|
+
- **Reference configuration:** OpenGame's own platform found a play counter imported nowhere (plays were never counted) and art uploaded to quarantine that was never promoted (every cover 404'd), each with every test green. AW shipped 35 sound files no weapon reached.
|
|
182
|
+
- **Applies when:** Every game, at every handover between two pieces of work.
|
|
183
|
+
- **Verify:** A script lists files under the game's source and assets that nothing references, and the list is empty or explained.
|
|
184
|
+
|
|
185
|
+
### One check loads the real thing the real way
|
|
186
|
+
- **Invariant:** At least one check exercises each dependency the way production does: the real loader, the real network library, the real audio engine, the real input path.
|
|
187
|
+
- **Technique:** Alongside fast tests with stand-ins, keep one test that loads every shipped model through the game's own loader, one that makes the game's real network calls in a real browser, and one that calls every public method of the audio engine on a live context. Run the same scripted walk in the test kit and in the production build.
|
|
188
|
+
- **Reference configuration:** A networking feature once shipped dead because every validator injected its own stand-in and the real script tag was missing. AW's audio had five green checks while a deleted constant threw on the first footstep. LC's input was dead in the real game because nothing called its input poll, while the test kit polled for itself, and a cross-stream deadlock left a hand off in 2,003 of 2,003 frames because each side's unit test used a stub for the other.
|
|
189
|
+
- **Applies when:** Every game with external libraries, assets, networking or more than one team.
|
|
190
|
+
- **Verify:** Delete the real dependency on purpose (remove the script tag, rename a model): the real-path check fails, and the stand-in tests do not notice.
|
|
191
|
+
|
|
192
|
+
### Judge motion and sound as sequences, and say what nobody could perceive
|
|
193
|
+
- **Invariant:** Animation, camera feel, flicker, transitions, timing and the mix are judged from recordings over time on the game clock, never from a single still or a number, and every report says which of them nobody on the team could perceive.
|
|
194
|
+
- **Technique:** Film a fixed sequence (still, turn, strafe, walk, aim) at a fixed simulation step per frame, so before and after film identical poses, and review it at stride 1 and full size on the delivered output; mask screen-fixed overlays from any motion score. Record the mix during the busiest moment and listen at a consistent level. An agent cannot hear, and cannot watch a film at speed (it reads frames in order), so write "nobody has heard this" or "read from frames, not watched at speed" wherever it applies, and give the owner named A/B clips with what to listen or look for. How to film it is in `opengame-capture`.
|
|
195
|
+
- **Reference configuration:** ST records the mix during its playtests and treats peak and loudness readings as never counting as a listening verdict. MC's foot sliding and AW's slideshow remote players were invisible in stills, and OC's wonder-weapon sound v3 was pushed to a loudness number without anyone listening; the owner rejected it on first listen as distorted. OS's still panels sat at 3-4 across 135 builds while the owner's notes were about motion; a static z-fight gate stayed green through 8 days of flicker complaints; free-running flicker captures varied 10-30% against about 5% on the game clock, and unmasked, the reticle supplied most of a "-54%" gain. LC wrote "NOBODY HAS HEARD" 72 times in its audio documents: nine audio leads set every level by meter, and the owner, "the first person to listen to the game", found no rain, a female screech and a harsh flamethrower. "I CANNOT WATCH A FILM AT SPEED" (an LC lead); 72 reload films were made and none was watched by a lead. LC then gave the owner 41 named A/B audio pairs.
|
|
196
|
+
- **Applies when:** Every game with motion or sound.
|
|
197
|
+
- **Verify:** The report links a recording for every motion and audio claim, a planted shimmer is flagged by the clip review, and every audio claim says whether a person listened.
|
|
@@ -0,0 +1,74 @@
|
|
|
1
|
+
import { createRequire } from "node:module";
|
|
2
|
+
import { readFile, mkdir } from "node:fs/promises";
|
|
3
|
+
import { dirname, resolve } from "node:path";
|
|
4
|
+
import { fileURLToPath } from "node:url";
|
|
5
|
+
|
|
6
|
+
export function assessEvidence({ before, after, framesMs, pageErrors }, criteria) {
|
|
7
|
+
const coordinate = value => Array.isArray(value) && value.length > 0 && value.every(Number.isFinite);
|
|
8
|
+
if (!coordinate(before) || !coordinate(after) || before.length !== after.length || !framesMs.length || !framesMs.every(value => Number.isFinite(value) && value > 0)) throw new Error("Missing position or frame-time evidence.");
|
|
9
|
+
if (!Number.isFinite(criteria.movementMin) || criteria.movementMin <= 0 || !Number.isFinite(criteria.frameP95MsMax) || criteria.frameP95MsMax <= 0) throw new Error("Provide positive game-specific movement and frame-time limits.");
|
|
10
|
+
const distance = Math.hypot(...after.map((value, index) => value - before[index]));
|
|
11
|
+
const sorted = [...framesMs].sort((a, b) => a - b);
|
|
12
|
+
const frameP95Ms = sorted[Math.ceil(sorted.length * 0.95) - 1];
|
|
13
|
+
return { ok: distance >= criteria.movementMin && frameP95Ms <= criteria.frameP95MsMax && pageErrors.length === 0,
|
|
14
|
+
distance, frameP95Ms, frameSamples: framesMs.length, pageErrors: [...pageErrors], criteria };
|
|
15
|
+
}
|
|
16
|
+
|
|
17
|
+
/** Drive player input and measure the game's position contract, rather than inferring
|
|
18
|
+
* movement from changed pixels. The host chooses its own tolerances and soak budget. */
|
|
19
|
+
export async function browserEvidence(config) {
|
|
20
|
+
const url = new URL(config.url);
|
|
21
|
+
if (!["http:", "https:"].includes(url.protocol) || url.username || url.password) throw new Error("Use an HTTP(S) game URL without credentials.");
|
|
22
|
+
const require = createRequire(resolve(process.cwd(), "package.json"));
|
|
23
|
+
const { chromium } = require("playwright");
|
|
24
|
+
const holdMs = config.holdMs ?? 500;
|
|
25
|
+
const soakMs = config.soakMs ?? 1000;
|
|
26
|
+
const throttle = config.cpuThrottle ?? 1;
|
|
27
|
+
if (![holdMs, soakMs, throttle].every(value => Number.isFinite(value) && value > 0) || !Array.isArray(config.statePath) || !Array.isArray(config.positionPath) || !config.playSelector || !config.key || !config.screenshot) throw new Error("Specify statePath, positionPath, playSelector, key, screenshot and positive measurement budgets.");
|
|
28
|
+
const browser = await chromium.launch({ headless: true, args: ["--mute-audio"] });
|
|
29
|
+
const errors = [];
|
|
30
|
+
try {
|
|
31
|
+
const context = await browser.newContext({ viewport: config.viewport ?? { width: 1280, height: 720 }, serviceWorkers: "block" });
|
|
32
|
+
const page = await context.newPage();
|
|
33
|
+
page.on("pageerror", error => errors.push(error.message));
|
|
34
|
+
const session = await context.newCDPSession(page);
|
|
35
|
+
await session.send("Network.enable");
|
|
36
|
+
await session.send("Network.setCacheDisabled", { cacheDisabled: true });
|
|
37
|
+
await session.send("Emulation.setCPUThrottlingRate", { rate: throttle });
|
|
38
|
+
const start = performance.now();
|
|
39
|
+
await page.goto(url.href, { waitUntil: "domcontentloaded", timeout: config.loadTimeoutMs ?? 30000 });
|
|
40
|
+
await page.locator(config.playSelector).click();
|
|
41
|
+
await page.waitForFunction(({ statePath, positionPath }) => {
|
|
42
|
+
const read = (value, path) => path.reduce((node, key) => node?.[key], value);
|
|
43
|
+
const position = read(read(window, statePath), positionPath);
|
|
44
|
+
return Array.isArray(position) && position.length > 0 && position.every(Number.isFinite);
|
|
45
|
+
}, { statePath: config.statePath, positionPath: config.positionPath });
|
|
46
|
+
const position = () => page.evaluate(({ statePath, positionPath }) => {
|
|
47
|
+
const read = (value, path) => path.reduce((node, key) => node?.[key], value);
|
|
48
|
+
return read(read(window, statePath), positionPath);
|
|
49
|
+
}, { statePath: config.statePath, positionPath: config.positionPath });
|
|
50
|
+
const before = await position();
|
|
51
|
+
await page.keyboard.down(config.key);
|
|
52
|
+
await page.waitForTimeout(holdMs);
|
|
53
|
+
await page.keyboard.up(config.key);
|
|
54
|
+
const after = await position();
|
|
55
|
+
const framesMs = await page.evaluate(duration => new Promise(resolve => {
|
|
56
|
+
const frames = []; let previous = null; let stopped = false;
|
|
57
|
+
const tick = now => { if (stopped) return; if (previous !== null) frames.push(now - previous); previous = now; requestAnimationFrame(tick); };
|
|
58
|
+
requestAnimationFrame(tick);
|
|
59
|
+
setTimeout(() => { stopped = true; resolve(frames); }, duration);
|
|
60
|
+
}), soakMs);
|
|
61
|
+
await mkdir(dirname(resolve(config.screenshot)), { recursive: true });
|
|
62
|
+
await page.screenshot({ path: resolve(config.screenshot), fullPage: true });
|
|
63
|
+
return { ...assessEvidence({ before, after, framesMs, pageErrors: errors }, config.criteria), coldLoadToCaptureMs: performance.now() - start,
|
|
64
|
+
holdMs, soakMs, cpuThrottle: throttle, screenshot: resolve(config.screenshot) };
|
|
65
|
+
} finally { await browser.close(); }
|
|
66
|
+
}
|
|
67
|
+
|
|
68
|
+
if (process.argv[1] === fileURLToPath(import.meta.url)) {
|
|
69
|
+
try {
|
|
70
|
+
const result = await browserEvidence(JSON.parse(await readFile(process.argv[2], "utf8")));
|
|
71
|
+
console.log(JSON.stringify(result, null, 2));
|
|
72
|
+
if (!result.ok) process.exitCode = 1;
|
|
73
|
+
} catch (error) { console.error(error.message); process.exitCode = 1; }
|
|
74
|
+
}
|
|
@@ -0,0 +1,179 @@
|
|
|
1
|
+
> Carried from `opengame-weapons` 1.1.0. Read it when: Build first-person weapons and view models that hold up next to a shipped shooter. Covers guns built from a reference, size and lens distance, one whole weapon in every view, a still gun, scope-in from the press, a real sight picture, reloads climbed gun by gun, hands that keep one size and never clip, hit zones, camos, upgrades and box odds. Use when the gun looks like a toy, shakes or clips, reloads look fake, scoping feels slow, hits do not kill, or add a weapon.
|
|
2
|
+
|
|
3
|
+
|
|
4
|
+
# Weapons
|
|
5
|
+
|
|
6
|
+
In a first-person shooter the gun is on screen for the whole game, so every flaw in it
|
|
7
|
+
is seen thousands of times. OC's owner raised gun fidelity about 30 times, its wonder
|
|
8
|
+
weapon about 22 times and hands about 15 times. OS's owner asked about hands 14 times
|
|
9
|
+
and scoping about 11. LC's owner wrote about the body, hands, holds and reloads about 19
|
|
10
|
+
times, and at the end 0 of 73 reloads passed the team's own bar. Five habits cost most of
|
|
11
|
+
those rounds. Agents built a known gun from an icon, fixed one gun when sixty had the
|
|
12
|
+
defect, checked size by eye, showed a cheap or empty hand while the real model loaded,
|
|
13
|
+
and judged motion from stills instead of at speed in the player's own picture.
|
|
14
|
+
|
|
15
|
+
Rules carry five tags; `opengame-quality-bar` explains them and the game codes. The
|
|
16
|
+
three.js view model pass, depth handling, coverage masks and model loading are in
|
|
17
|
+
[references/threejs.md](references/threejs.md); scopes and sight pictures are in
|
|
18
|
+
[references/sights-and-scopes.md](references/sights-and-scopes.md); reload staging and
|
|
19
|
+
hand contact are in [references/reloads-and-hands.md](references/reloads-and-hands.md).
|
|
20
|
+
Recoil springs, shake, hit feedback and how much a held gun may move are in
|
|
21
|
+
`opengame-game-feel`; bodies, arms and joint limits are in `opengame-characters`.
|
|
22
|
+
|
|
23
|
+
**One defect is a defect in the whole class.** When the owner flags one gun, sight, camo
|
|
24
|
+
or pickup, sweep every gun on every path before reporting it fixed. The rule and OC's
|
|
25
|
+
numbers are in `opengame-quality-bar`.
|
|
26
|
+
|
|
27
|
+
**An approved look is frozen.** A scope, camo or hand look the owner has approved is the
|
|
28
|
+
target no later pass may make worse, and a regression is fixed by restoring it exactly
|
|
29
|
+
first ("Approved quality is frozen" in `opengame-quality-bar`). OS's approved scope was
|
|
30
|
+
broken twice by "parity" passes until an exact revert ("it used to be perfect this scope").
|
|
31
|
+
|
|
32
|
+
## Rules
|
|
33
|
+
|
|
34
|
+
### Known weapons are built from their reference
|
|
35
|
+
- **Invariant:** A real firearm, vehicle or prop, or an item from the creator's own spec, matches its reference's proportions and silhouette from the side, top and back, as seen from the camera players know it from, before anyone works on surface detail; it is built from the reference, not from an icon, a texture sheet or memory, and the reference files stay out of the build. An item another studio designed is never recreated: its reference sets only the reads players expect (size class, silhouette type, how it handles), and the design, name and files are original.
|
|
36
|
+
- **Technique:** Get published dimensions, the creator's own model or clear photographs first. Render it beside the candidate under one camera and score the silhouette overlap from side, top and back; treat that score as a floor for the outline only, and check eyepieces, lenses and small parts on a side-by-side render, because an outline score passes them blind. Write a materials list and a reload beat list from the reference, and recover its handling data (magazine, reload time, cadence, projectile speed, splash). Keep reference files outside the build folder, hash-check the shipped files against them, ship original names, and strip maker names and markings from textures as well as file names.
|
|
37
|
+
- **Reference configuration:** OC built its signature wonder weapon for about 14 builds from "a texture sheet and a 64-pixel icon, never the actual model", and the owner rejected all of them in one message ("ur not even fucking CLOSE"). It took about 30 builds over 3 days. Six owner images and measured proportions fixed it: side outline overlap 0.947 to 0.954 against a bar of 0.95, first-person framing from directly behind, and triangles held at or below the heaviest loadout gun (24.9k; the first attempt was 47k). A written materials list (worn enamel, chipped edges, brass patina, a gauge with a needle, rivets, a decal, grime) and handling data (magazine, reload time, cadence, projectile speed, splash) came before the next build. A hash check found 0 of 44 private reference files among 12,987 shipped, while trademarked names were still in the build at the launch audit. OS built about 20 rifles from published dimensions behind an overlap gate (side 0.92, top 0.88); all 23 runs passed while judges named thick eyepieces and flat lenses; its owner judged against the reference the owner had named, not photographs of the real rifle, and the owner's taste beat reference fidelity where the two differed (a level hip pose over the source's odd one); seven rifles still shipped maker names and were rebaked. An agent's own note: "more surface detail alone won't address your criticism".
|
|
38
|
+
- **Applies when:** Any weapon, vehicle or prop built to a reference the audience knows, a real object or the creator's own spec. Ship original names, designs and markings, never another studio's model, design, names or textures.
|
|
39
|
+
- **Verify:** Each round, a sheet of reference and build side by side from the same side, top, back and first-person cameras, with the overlap score for each view and a line on each small part.
|
|
40
|
+
|
|
41
|
+
### Generated meshes are for large forms; held weapons come from a source that can be photoreal
|
|
42
|
+
- **Invariant:** Small exact parts (sights, screws, levers, knurling, ejection ports) are modelled by script, parametrically or from licensed artist models, and no weapon enters the game before a studio render beside its reference: a photograph of the real thing, or the reference model or concept for an invented one.
|
|
43
|
+
- **Technique:** Search licensed artist models (CC0, or CC BY with a credit) before generating a held weapon. Generate only large forms, from side pictures you have looked at before paying for the mesh, with magazines and moving parts made separately. Grade generated materials before use: metalness only where steel is worn bare, roughness kept in a real band, normals softened. Check the mesh on a millimetre grid, cut moving parts out as whole objects, and refuse an install whose ambient occlusion bake is near black. Judge at the views the camera actually sees, then on the owner's path. Spend only with the person's own account and their yes on the cost (`opengame-assets`).
|
|
44
|
+
- **Reference configuration:** LC paid about $1.05 a generated weapon ($0.18 for side pictures, $0.70 for the mesh), $96.68 for 72, and 0 of 73 were graded photoreal. An M1911 rebuilt from five generated parts cost $6.60 and three hours and still "smears levers, knurling and screws ... its maps are painted light, not material". Defects included mirrored pictures, fused rounds, flat chambers, no ejection port and cocking handles cut into fixed bodies on 18 weapons; licensed artists' models were "crisp where the generated mesh was soft". Blued steel read right at metalness 0.2 to 0.5 and roughness 0.45 to 0.7, and generated maps were graded to roughness 0.55 to 0.67 with normals at half strength. OS refused installs with AO under 40 of 255 after one bake averaged 0.6 ("The renders still looked right only because the flat paint colours land on the correct faces"), and bevels over 4 mm folded into "a hollow black channel".
|
|
45
|
+
- **Applies when:** Weapons, vehicles, machines and any hard-surface object seen at arm's length.
|
|
46
|
+
- **Verify:** A studio, reference and in-game sheet per weapon with a line-by-line verdict.
|
|
47
|
+
|
|
48
|
+
### View model size is measured per class, in every pose, on several screens
|
|
49
|
+
- **Invariant:** Every gun of a class covers about the same share of the screen at hip, aimed, sprinting and drawing, on wide, standard and phone screens, checked against a frame from the genre's reference game; every gun rests level at hip; the gun is always in hand and clear of the HUD.
|
|
50
|
+
- **Technique:** Render the weapon layer alone to a mask and count its pixels in each pose at each screen size. Keep a table of target shares per class. Fit hip placement once for the whole roster against annotated reference frames, with a fixed view model lens so the FOV slider does not resize the gun, and gate pitch and roll. Change one shared scale, then rerun every gun. Place holds near the bottom of the picture against the edge computed from the live aspect, because a fixed horizontal field of view crops the bottom of a wide window. Also check the muzzle position and near-plane cuts on long rifles at phone aspect.
|
|
51
|
+
- **Reference configuration:** OC hip targets, as a share of a 1280 x 720 frame: pistol about 6%, SMG about 8%, rifle about 9%, LMG about 9.5%, wonder pistol 7 to 9%. The wonder weapon shipped at about 40% and went to 5 to 7% in a hotfix. One fix swung the hip guns from "very small" to "so fucking massive", and the size had been checked against CoD by one agent's eye. Within a class OC holds the hip footprint within about 2x (rifles had spread 3.9x and went to 2.1x); hip yaw above about 13 degrees reads as sideways (one gun sat at 21.5). OC's guns are drawn at 1.5x life size against an arm of about 0.52 m, so a true shoulder weld was impossible; fix the scale on day one. OS fitted hip placement to 41 annotated reference stills under a fixed 68 degree view model lens, so FOV 65 to 120 moves the gun under 0.8%: 17 of 20 rifles within 3% of target, roll within 1.2 degrees against a 2 degree gate, and pitch from 6.5 to 14 degrees down to 1.8 to 2.2 after five notes in 24 h ("so many guns are tilted up... not just m40a3"). LC's pistol rests at 4.5 to 5% of the picture's width, 18 cm from the lens; a rest tuned at 1280 x 720 lost both hands off the bottom of the owner's 1790 x 860 window ("u cant even see ur hands now?").
|
|
52
|
+
- **Applies when:** First-person games with held weapons or tools.
|
|
53
|
+
- **Verify:** The coverage table for every gun in all four poses at 1280 x 720, 1920 x 1080, 844 x 390 and the widest window the players use, each gun inside its class band, with pitch and roll per gun.
|
|
54
|
+
|
|
55
|
+
### Nothing of the view model comes near the lens, on any frame
|
|
56
|
+
- **Invariant:** In every state and every change between two, no weapon, glove or sleeve comes nearer the camera than its floor, covers more than its share of the picture, or shows its inside, at every window shape the players use.
|
|
57
|
+
- **Technique:** In a headless test that steps the real view code, project every vertex of the arms, gloves and weapon into the player's picture each frame on a coarse grid and fail on distance and cover. When it fails, change the pose; never add a reactive push-away, which jolts and runs away. Give the view model its own depth floor (a near plane or a depth remap) under its closest part so a long gun never shows its inside. While aimed, the weapon's eye relief places the sight instead, and only the sight itself may come that close.
|
|
58
|
+
- **Reference configuration:** LC's floors are weapon 12 cm and glove and sleeve 9 cm (pistol 16), checked on a 120 x 58 grid of the owner's 1790 x 860, 100 degree picture. It fails when anything nearer than its floor covers over a sixth of the picture for more than three frames, a sleeve over a twelfth for six frames, or over a third outside a reload. It found a glove 2.1 cm from the glass at a steep look and a one-handed crank pushing a glove through the lens. A reactive guard jolted 37 times and ran away 35 cm. LC's aimed apertures sit 6 to 15 cm from the eye under their own eye relief; the floors govern every other state. OC found 19 of 61 guns with mesh within 0.15 m of the eye, 9 of them through it. OS's view-model depth remap (an effective 1 cm near plane) with a 10.5 cm eyepiece floor showed no gun's inside on any frame over 20 rifles x 17 actions.
|
|
59
|
+
- **Applies when:** First-person games with visible arms or a held weapon.
|
|
60
|
+
- **Verify:** The per-frame test over every state at 30, 60 and 144 fps plus an injected long frame, and one film at the widest supported aspect.
|
|
61
|
+
|
|
62
|
+
### A still player sees a still gun
|
|
63
|
+
- **Invariant:** The held weapon, its sights and screens, and anything attached to the camera are single-imaged and still when the player is still; under real mouse look at the player's refresh rate, camera-attached objects track the view exactly and the weapon moves only by its designed motion, never by a late frame or a smear.
|
|
64
|
+
- **Technique:** Give the view model its own depth band and write real motion vectors for it from the camera's own matrices; keep that band out of velocity dilation and motion blur, and zero its velocity only inside an aimed see-through optic. Pose camera-attached objects after the camera update, in the same pass as the hands. Test at the player's refresh rate with real flicks, never only in a scripted capture with no mouse look. How much the weapon may move on its own, and the designed motion added back, are in `opengame-game-feel`.
|
|
65
|
+
- **Reference configuration:** OS: "even when i dont move the guns barrel and attachments are just like shakign left and right?..." Zeroing the hip velocity stopped that and caused "a dotted sort of hazy outline around the gun", because the temporal upscaler's dilation spread zero onto world pixels at the silhouette. The final path took the ACOG still-hold frame diff from 110 to 4.45 at 120 Hz; 30 fps captures never showed it. A canopy lagged 31 cm and 8.6 degrees at 3 rad/s and a killstreak controller slipped 150 to 480 mm, both to 0 once posed in the hands' pass. A scripted check passed at 0.03 px and the owner reopened it ("killstreak devices still shake/shift a lot"), because the test had no mouse look. LC's muzzle travel standing went from 3 x 9 px to 0.5 x 1.6 at 720p once the rig took the camera's own motion.
|
|
66
|
+
- **Applies when:** First-person games, above all with temporal anti-aliasing, upscaling or motion blur.
|
|
67
|
+
- **Verify:** Still-hold clips at 120 to 144 Hz with a frame-to-frame muzzle diff under 0.5 px (OS's bar), 200% crops while strafing, and drawn-position traces under real flicks.
|
|
68
|
+
|
|
69
|
+
### Exactly one whole weapon is visible, in every view
|
|
70
|
+
- **Invariant:** The player, bots, menus, killcams and drops always show exactly one complete weapon: never bare hands, floating parts, a mix of stand-in and hero parts, or a generic gun-shaped placeholder. Cards that show weapons appear as a group, and nobody pays for a weapon whose model is not ready.
|
|
71
|
+
- **Technique:** Keep the previous complete weapon in hand until the exact new one has loaded with its materials and its shaders are compiled. Preload the likely next weapons (the visible roster, the box pool, nearby wall buys, menu neighbours) and fetch textures with the model. When nothing is resident, hold the raise off frame or keep the screen dark rather than show a stand-in in front of the eye. A light version is allowed only as a complete copy of the same weapon with its own geometry and anchors, on cards and at distance, and never mixed with hero parts. Gate a purchase on the model being ready. Flag any frame with a whole-weapon miss or a detached part. Ship authored hero models with named parts and sockets (see `opengame-assets`); code-built guns do not pass.
|
|
72
|
+
- **Reference configuration:** OC: one lazy-load change put low-poly guns in 6 places, and a wall gun charged points while its model was missing ("never load any low poly versions of guns ANYWHERE in this game", "NOTHING low poly shows when spinning box"); OC prewarmed its visible roster of 8 guns, about 28 MB. OS's owner ruled on the other half: "we cant have a situatoinw here not all the gun models load..." and "u can see bots running around with no guns". "Guns never fail to show" became its first launch blocker; cards fade in together from light copies of the real rifles (10 to 25 s down to 0.1 to 0.24 s); the class rifle prefetches with the weapon table (13 s of empty gloves to the first spawn frame); the raise waits off frame up to 4 s; a per-frame part check caught a suppressor 0.025 m off its rifle, and a sweep found 0 misses in 5,939 frames. The rule changed because two owners ruled on two halves of it: OC's against a low-detail gun, OS's against an empty hand.
|
|
73
|
+
- **Applies when:** Any game that streams or swaps weapons the player sees, and bots holding them.
|
|
74
|
+
- **Verify:** Throttle the network and cold-boot a full bot lobby, then take every acquisition path and open every menu that shows a gun, stepping the recording at each swap: every frame shows one whole weapon, and the miss log is empty.
|
|
75
|
+
|
|
76
|
+
### The first frame of a run shows the finished weapon, lit, in its hold
|
|
77
|
+
- **Invariant:** On every path into play (first visit, return, retry, any saved cheats or loadout, a purchase), the first frame shown has the chosen weapon finished, lit and in its rest hold, with no raise, upgrade reveal or perk animation playing on it.
|
|
78
|
+
- **Technique:** Settle the pose for some simulation steps before the first draw. Keep the screen dark until the chosen weapons are built. Size finished-material slots for the largest loadout, and prefetch purchasable weapons within a radius so a bought gun is built, mapped and lit before it can be bought. Play an upgrade's reveal only when the weapon is taken from its machine, and never let start-of-run grants queue animations on the held weapon. Test with every setting the owner has saved.
|
|
79
|
+
- **Reference configuration:** LC settles 20 steps at 1/60 s, keeps 12 finished-model material slots (with 8, refits drew plain) and prefetches racks within 30 m. Six granted perks each queued an injection and put the gun aside for 3.5 s, seen only with every cheat saved ("nobody testing with two cheats saw it"). The owner reported the spawn four times, among them "what the fuck is this first view of gun? guns need to automatically look like 2nd", and of a purchase "it looked very weird for a few seconds as if flashlight wasnt on". After: 20 of 20, 22 of 22 and 11 of 11 spawns clean.
|
|
80
|
+
- **Applies when:** Every shooter, and every game with a held item at spawn.
|
|
81
|
+
- **Verify:** 20 spawns per loadout on the owner's path, frames 0 to 40 read, on 5 and 20 Mbit lines.
|
|
82
|
+
|
|
83
|
+
### Every way to get a gun shows the same gun
|
|
84
|
+
- **Invariant:** A gun from the box, a wall, an upgrade, a pickup, the firing range or a menu preview matches the loadout version in model, scale, camo, optic, grip and angle.
|
|
85
|
+
- **Technique:** Route every acquisition through one path that applies the slot's camo and a compatible optic. Audit every gun in every view against every path; a debug grant does not prove a pickup path. Give the firing range every gun, attachment, camo and upgrade.
|
|
86
|
+
- **Reference configuration:** OC's sheet was 61 guns x hip, aimed and third person x base, upgraded and camo, 549 views, plus 152 acquisition cases. Wall guns fill 82% of their niche at no less than 1.5x held size. Pickup camo had reset to the default, and the wonder weapon was missing from the range preview twice. Wonder weapons come only from the box, which meant auditing all 11 code paths that write a weapon slot.
|
|
87
|
+
- **Applies when:** Games with more than one way to get a weapon.
|
|
88
|
+
- **Verify:** One contact sheet per round of every gun on every path, and a list of slot-writing code paths each with its capture.
|
|
89
|
+
|
|
90
|
+
### Weapons in other hands have their own size band
|
|
91
|
+
- **Invariant:** A weapon held by a bot, a teammate or the player's third-person body reads in proportion to that body at every range, and the drawn body follows its aim smoothly while hits stay on the simulation.
|
|
92
|
+
- **Technique:** Scale only the world instance, never through a scale shared with first person, and cap the drawn length to a share of body height. Turn the drawn body toward the aim through a critically damped spring with a capped lag, and resolve shots from the simulation, never from the drawn pose. Re-express first-person actions such as reloads in the holder's own frame, judge where a friend aims by the angle from the watcher's eye, and never let a muzzle rest on a friend.
|
|
93
|
+
- **Reference configuration:** OS drew first-person guns at 1.5x and shared that scale, so a 2.12 m rifle was drawn 3.2 m long in a bot's hands: "ptrs barrel still reads wayy too long when u see a bot holding it ... all snpier should not look egregiously big or lage or long in a bots hands (nor too small)". Its third-person cap is 1.25 m, 0.53 to 0.65 of body height, and a 0.06 s spring capped at 0.6 rad cut one-frame yaw snaps from 1.7% to 0.2% of frames. LC's co-op friend played a first-person reload track that sent his magazine "into his belly" until it was re-expressed in his frame, and "aimed at the watcher" judged in metres put the stock at his collar; no bore rests on a friend within the weapon's length plus 0.33 m and 40 degrees. OC's third-person gun was hidden behind the body in every follow view (`opengame-characters`).
|
|
94
|
+
- **Applies when:** Games where players see others holding weapons: bots, co-op, multiplayer and third person.
|
|
95
|
+
- **Verify:** A sheet of every weapon in a bot's hands at 4, 10 and 30 m, and the drawn yaw's per-frame step through a fight.
|
|
96
|
+
|
|
97
|
+
### Aiming goes through a real sight picture, measured per weapon
|
|
98
|
+
- **Invariant:** When aimed, the rear sight, the front sight and the bore's point at range lie on one line through the picture's centre, tops level, at the weapon's own eye relief; nothing of hand, sleeve or receiver sits on that line or beside the notch; lenses are clear; the gun stays upright; and the aim is calm and reversible mid-raise.
|
|
99
|
+
- **Technique:** Publish each weapon's sight points in data and fit the eye point from the sight meshes' world positions, never from a mount socket. Place the lens at the rear sight minus the bore direction times the eye relief, let the body follow the eye, and keep the firing hand out of the picture on long guns. Narrow the field of view on the sights and slow the look by the same ratio. Ray-test the sight line through each optic. Ease aiming over the gun's own aim time and allow reversal partway. Draw reticles in a shader, visible only from the eye side, behind dark coated glass, with housings that never take camo glow. Build aim rotation from yaw and pitch so no roll creeps in. Test what stands beside the notch as well as above it. Where a true sight picture cannot look right, aim with the gun lowered and a slight zoom.
|
|
100
|
+
- **Reference configuration:** OC aimed from a socket 343 mm off the ring, a 31 px error; a glove covered 35.29% of the central box on one gun ("your glove completely blocks your ADS") and the fix took it to 0.04%; the sight is within 6 px of the crosshair at 1280 x 720 and 844 x 390 (achieved 0.5 to 2 px); lenses on 4 rifles marked transparent blocked 385 of 385 rays; shortest-arc aim rotation rolled guns 25 to 30 degrees; aim times run 0.19 s (pistol) to 0.48 s (LMG), and the view model lens goes from 75 to 58.5 degrees; for the minigun the fallback was a lowered gun with a slight zoom, since CoD's version is hip-only. LC sets eye relief per weapon, 6 to 50 cm across 73 (a handgun at real reach, 43 cm; open sights 30 to 40; apertures 6 to 15), narrows the view 1.35x for long arms and 1.6x for pistols, and holds each rear sight, post and the round at 25 m within 3 px of one line. Earlier, 69 of 73 lined up within 4 px while 56 long guns still had the firing hand in frame at 10 to 19 cm, and the owner: "the ads view / positionions on all guns is extremely bad and not realistic like i just checked the mp40". A finger-over-the-sight check passed because it looked only above the notch floor ("when ur zoomed in with pistol why does ur right pointer finger liek cover ur aim?").
|
|
101
|
+
- **Applies when:** Shooters with aim-down-sights.
|
|
102
|
+
- **Verify:** Aim every gun at desktop, phone and the widest window and capture: measure sight offset in pixels, the covered share of a central box and the hand's distance from the line, and ray-test every lens; then a quarter-speed film of raise, hold, shot and lower.
|
|
103
|
+
|
|
104
|
+
### Scope-in registers from the press, and nothing forces the player out
|
|
105
|
+
- **Invariant:** Zoom starts on the press frame, with field of view, weapon offset and look sensitivity on one curve; a shot fired once the sight picture shows lands on the reticle; cycling a bolt, reloading or firing never makes the player release and press aim again; and release restores the hip view within a few frames.
|
|
106
|
+
- **Technique:** Drive field of view, offset and sensitivity from one eased progress value. Cut to the scope image at a fixed fraction with the same aperture on both sides of the cut and no crossfade, and fire from the rendered view on the press. A scope is a field-of-view change plus an overlay, never a second world render, and the render scale and targets stay fixed across the cut so nothing is rebuilt. Whether the rifle leaves the eye during a bolt cycle is a style call; either way it returns to the sight on its own while aim is held. Set each optic's magnification to frame a standing target at the game's usual engagement range, and capture scoped frames there, never at a nearby wall. Prove hits with a bench of real inputs at fixed delays and ranges, then leave the feel to the owner's hands.
|
|
107
|
+
- **Reference configuration:** OS shots 100 to 200 ms into a scope-in hit 0 to 2 of 12, "2-2.7° below the crosshair", then 12 of 12 from 100 ms. Lens and hands share 220 ms in and 70 ms out against a quickscope bar of about 0.27 s where the reference data said 0.40 s ("like sleight of hand pro quickscopes on mw2/mw3 from cod on all guns (no perks required"); separate curves read as "left to straight". "u shouldnt force people to zoom out to pull the bolt", after an agent had added the unscope from memory of the reference games; LC, a realistic body-camera game, works the bolt with the rifle off the eye and brings it back. "frame would halt for a few frames righ tbefore zoom in": a resolution change at scope-in re-created 46 to 70 textures, and ADS now costs about 0.4 ms. The bench passed while the owner still saw "a weird frame", a three-frame blur. OS tuned scope fields of view to frame a standing target at each map's signature range (53 to 236 m) and captured at 100 to 300 m, after judges had scored scope shots taken 4 to 23 m away for about 40 builds.
|
|
108
|
+
- **Applies when:** Any scoped or aim-down-sights weapon.
|
|
109
|
+
- **Verify:** Bench hits per 12 at 0 to 250 ms and 50 to 200 m, and a 120 fps strip of scope-in, fire, bolt and release read frame by frame.
|
|
110
|
+
|
|
111
|
+
### Reloads are choreographed beat by beat
|
|
112
|
+
- **Invariant:** Each reload is a sequence of named physical beats specific to its mechanism, readable at a glance, with each beat's time held in one data file that the animation and the sound both read, so every sound lands on its contact.
|
|
113
|
+
- **Technique:** Write a beat spec per mechanism (box magazine, belt, pan, tube, cylinder, energy cells) for empty and partial reloads. Export the timings, trigger each sound from the same table with a short swish before contact, and add a small sound for each part that moves. Make sure every reload state machine exits. Someone listens to every gun's reload before it counts.
|
|
114
|
+
- **Reference configuration:** The owner on OC: "rn u can barely tell they are reloadin gbut we need lke a sick mag drop / mag out and then mag in and like slap the side of the gun", and later that a magnum should eject the shells inside it. OC's wonder weapon spec was latch, chamber opens, spent cells out, fresh cells in one at a time, snap shut, flicker on. Reload sounds had run 503 to 743 ms late (SMG bolt 503, rifle magazine 743, pump 649); afterwards they land within 12 ms of hand contact with a swish 110 ms before. OC's bow had an auto-reload cycle that never exited. LC had reload sounds for ten families, six stages each, timed only on pistols, and nobody had heard one when the owner asked "we should also have reload sfx for all guns too please... idk if we have them?".
|
|
115
|
+
- **Applies when:** Every game with reloading weapons, and any tool with a multi-step use animation.
|
|
116
|
+
- **Verify:** Record a reload with sound, step it frame by frame, and check that each beat's sound starts within the game's tolerance of its contact frame, for every gun.
|
|
117
|
+
|
|
118
|
+
### Each reload is climbed gun by gun, in the player's own picture, at speed
|
|
119
|
+
- **Invariant:** Every weapon's empty and partial reload reads as that weapon's real drill from the player's own camera at full speed: the work is in the picture and lit, a hand is at work through it, every action begins and ends with a hand travelling, nothing pops, floats, flashes or fills the frame, and parts that move are separate and really move.
|
|
120
|
+
- **Technique:** Keep a per-weapon ledger scored on fixed points (drill, picture, hands, wrists, motion, parts, light), each fault with its owner, and climb it one gun at a time. Stage by where the magazine well sits in the picture; cant long arms toward the camera and keep the cant through the stage; time hand paths round the weapon and turn any jump of a target into a timed, eased travel; hand parts over when the hand arrives, not by the clock; show an empty magazine as a draw range on the one carried part. Confirm the mesh's moving parts are separate objects before animating hands. Light the work, not the wall above it. Calibrate the ledger against the owner: when the owner accepts what it fails, move the bar. Staging numbers are in [references/reloads-and-hands.md](references/reloads-and-hands.md).
|
|
121
|
+
- **Reference configuration:** The LC owner's order: "we really need to put each reload animation in first person view under scope and hillclimb to perfection that hands and wrists and arms and what u see on screen is all photorealistic". Its bar was 7 points scored 0 to 10, pass at 9 on each, with a hand at work at least every half second, sleeve and glove together under a sixth of the picture, wrists within 70 degrees bent and 30 to the side, and no finger through metal over 1.5 mm. A canted M16 took 7.7% of the picture against 19 to 20% upright. At the end 72 were filmed and 0 of 73 passed, while the owner accepted all eight pistol reloads ("the pistol is pretty good reload already lets just do all the other guns"), where the tester had marked the M1911 6/7/5/7/4/5/5: the owner reads the whole act at speed. Two leads spent eleven placings reaching for a cocking handle fused into the receiver mesh. OS's clip audit rebuilt logged frames from the real models and found bolt handles rotating into the stock and a reload path 16 cm under the rifle; clipped game time fell from about 24 s to 11.
|
|
122
|
+
- **Applies when:** Shooters with visible reloads, and any multi-step tool action seen in first person.
|
|
123
|
+
- **Verify:** Film every weapon empty and with rounds left, refitted too, at speed and at a quarter speed, in the player's light and window; a person watches them at speed, and the ledger shows each gun's scores.
|
|
124
|
+
|
|
125
|
+
### Hands keep one size and one material, and stay outside the gun on every frame
|
|
126
|
+
- **Invariant:** On every weapon, action and frame, in first and third person, each hand draws at one scale and shares one material and fill light in every state, touches the parts it holds without sinking into them, never reveals the gun's inside, and the support hand stays on the gun.
|
|
127
|
+
- **Technique:** Fit by grip pose, IK target and weapon placement, never by a per-weapon hand scale. Author hand and magazine contact points on each gun model. Over a deterministic stepped capture of every weapon and action, log the drawn glove scale and the signed clip depth every frame, flag every row over the threshold and fix the worst first. Count penetration over every vertex on the game's own rig, not the solver's. After locking the scale, refit every grip, since a constant scale moves contact on every clip. A debug view that paints back faces one colour finds exposed interiors. A simple hand (a thumb and a mitten, see `opengame-characters`) makes the fit tractable where the style allows. Measure third person too.
|
|
128
|
+
- **Reference configuration:** OC flags any phase deeper than 8 mm: 184 contact points over 56 guns, yet at launch 55 of 114 rows over 57 guns were still over it, the worst 20 to 58 mm, and third person was never measured; the support hand stays 2 to 80 mm from the gun ("we cant have that on any fo the guns"); shotgun shell inserts needed new motion. OS: "when reloading the arisaka my hand got a ton smaller? ... cant have that on any animation/gun...", fixed from 1.59x to one scale of exactly 1.1 over 15,382 frames, after a constant scale had pushed one bolt clip from 6 to 29 mm; cells over 6 mm fell only from 56 to 33; a black patch on the mitten through 30% of the bolt cycle on 7 rifles was missed by sampling 8 frames per bolt; fists and armed gloves differed 5x in fill ("the color of hands when holding gun doesnt match"). LC's finger solver passes nothing deeper than 1.5 mm counted over every vertex (sampling every second vertex missed a thumb 1.7 mm into the hammer), and its 2.9 mm on the solver's rig was 7.9 mm on the game's stretched forearm; a pistol thumb 8.9 mm inside the frame went to 0.
|
|
129
|
+
- **Applies when:** Games with visible hands on held items.
|
|
130
|
+
- **Verify:** The scale log reads 0.00% change, the clip table passes for every weapon and action on every frame, and every-frame strips cover the worst three cells.
|
|
131
|
+
|
|
132
|
+
### Handling numbers come from the reference game
|
|
133
|
+
- **Invariant:** Rate of fire, aim time, swap time, draw time, recoil and spray follow the reference for each gun, timings decide a press the same way at any frame rate, and damage is audited as time to kill across the run.
|
|
134
|
+
- **Technique:** Table every gun's handling against its reference, and check every row of weapon data against a published source. Use real fire rates, authored spray patterns, and recoil springs per class scaled per gun. Run cancels and timings on elapsed time. Keep aim assist off the mouse, and fit it to the reference game's measured envelope. Audit time to kill at early, middle and late rounds, and treat an outlier gun as a bug.
|
|
135
|
+
- **Reference configuration:** OC from World at War: aim 0.245 s in and 0.25 s out, swap 0.25 s, first draw from a machine 0.8 s, kick settling in 0.15 to 0.2 s, aimed kick at 30 to 40% of hip, muzzle flash 2 frames, real fire rates (MG42 1,200 rpm, a revolver 120; one revolver had done 1,000 damage at 625 rpm). Per-gun recoil scale 0.55 to 1.8x of its class; upgraded automatics kill a round-30 zombie in 0.65 to 1.0 s of headshots. OS: aim assist on pads only from the reference's envelope (strafe slip 3.9 to 0.1 degrees), LC's on keys and pads only, a per-frame camera trace took recoil overshoot from 16% to 3%, and "no perk should speed up ADS". LC corrected 26 rows of weapon data against a published source. Spring, timing and recoil tables are in [references/handling-numbers.md](references/handling-numbers.md).
|
|
136
|
+
- **Applies when:** Shooters, above all those that imitate a known game.
|
|
137
|
+
- **Verify:** The handling table for every gun beside its reference values, and a time-to-kill run per gun at each audited round.
|
|
138
|
+
|
|
139
|
+
### Hit zones come from the round's path, and lethal zones are an owner ruling
|
|
140
|
+
- **Invariant:** A round aimed at the chest scores as chest from the front, side and back, every zone the owner rules lethal kills with every weapon meant to kill there, and a hit marker shows only on damage that was scored.
|
|
141
|
+
- **Technique:** Classify trunk hits by how far the round's line passes from the capsule's axis, not by its entry point on the surface, and take the head zone from the head as last drawn. Write a lethal-zone table per weapon class, get one owner ruling, and gate it with a scripted sweep over zone, range and angle that asserts zone, damage and kill. Keep one authoritative damage event that the HUD and the marker read, never a second computation. Measure perks through the real hit path.
|
|
142
|
+
- **Reference configuration:** OS took the zone from the entry point, a full radius out on a capsule's side, so "EVERY trunk hit from the front, side or back scores ARMS". Most rifles did 70 to 98 to the trunk and did not kill; only 3 of 20 did, after a table had claimed every filmed rifle did, and only filming a montage found it. Its head capsule sat 0.13 m above the drawn head. The owner: "im also getting too many hitmarkers... when i shoudlnt be" and "chest and above and one shot kills"; the research notes had disagreed three ways on lethal zones. Two of four perks did nothing through the real path, and an armour perk was tuned to +40 so heavy chest hits (142 to 147) still kill, keeping the chest-and-above ruling true. OC's zone multipliers are in [references/handling-numbers.md](references/handling-numbers.md).
|
|
143
|
+
- **Applies when:** Any shooter with zone damage on primitive hitboxes.
|
|
144
|
+
- **Verify:** Per weapon, 10 rounds at mid-chest from front, side and back plus a 24-height sweep, logging zone, damage and kill: only the ruled zones kill.
|
|
145
|
+
|
|
146
|
+
### Each weapon kills in its own way
|
|
147
|
+
- **Invariant:** The body's reaction to a kill matches the weapon: energy weapons drop bodies where they stand, shotguns and explosives throw them, and small arms push them a little.
|
|
148
|
+
- **Technique:** Choose the death response from the weapon that dealt the kill: a directional death clip, then a ragdoll with an impulse set per weapon class. Tag splash damage by what it is, so an energy splash is not treated as an explosion.
|
|
149
|
+
- **Reference configuration:** OC: energy kills rise under 0.15 m and travel under 0.6 m (the wonder weapon at most 0.005 m up and under 0.75 m of slide); a point-blank shotgun throws at 5 to 6 m/s with a sideways part, arm flail, 45 ms of hitstop and stronger shake; a pistol 1.7 m/s; an explosive 4 to 5 m/s out and 3.5 to 4.5 up. The directional death clip plays 0.16 to 0.42 s before the ragdoll. The owner: "ray guns hsouldnt make zombies fly", after the agents had logged the wonder weapon throwing enemies into backflips as a win because its splash was tagged explosive.
|
|
150
|
+
- **Applies when:** Games with several weapon types and visible deaths.
|
|
151
|
+
- **Verify:** Kill the same enemy type at the same range with each weapon class and record it; measure the root's rise and travel against the table.
|
|
152
|
+
|
|
153
|
+
### Camos cover the whole gun, skip its hardware and have a measured cost
|
|
154
|
+
- **Invariant:** Every finish covers all paintable surface of every gun, leaves metal, optics, grips, lens rims and glass to an explicit hardware mask, reads as a real material, never blows out to pure white or black, stays calm in the sight picture, costs a fixed amount over the plain finish, and is never cut to save memory without asking. Which finish is the default is the owner's decision, written where every critic reads it.
|
|
155
|
+
- **Technique:** Mark hardware per weapon with a paint, metal and glow mask, never by material name, and measure coverage over every model and in a rendered view. Bake AO into every gun's material so finishes have something to sit on. Build finishes as material response (sheen by angle, a few clean bolts) on a tileable swatch atlas with cheap shader layers, or as one graph worked in the object's own space and selected only by uniforms, branched so an unworn finish costs about nothing. Cut glow from the sight picture while aiming. Time each finish against interleaved plain-finish windows in one boot.
|
|
156
|
+
- **Reference configuration:** OC: 58 of 60 guns lacked AO, so AO went into the packed material at cavity 0.2 to 0.4 with normals of at least 1024 px; gold read as "lumpy sparkly clay"; lightning became at most two clean bolts; an 11-octave camo cost 1.6 to 1.8 ms, 28% of a combat frame, and went to about 4 octaves against a 0.4 ms target; "do NOT cut down the sick camos we had to save memory". OS: coverage went from "half the gun" to 99.4 to 100% over 540 combinations; a name rule painted grips and scope bodies because of "'enamel' in the polymer name", and about a third of each rifle is hardware; a vivid camo that blew out on every rifle went from 82% to 12% on bare metal; every finish stayed within 0.25 ms of the 0.72 ms plain one, where 3D Voronoi finishes had cost 1.2 to 1.8 ms. LC's finishes all cost the same unworn until branched (0.9 ms worn, about 0 unworn). The default: OC's owner wanted the factory finish in menus and starting gear and kept rejecting a rainbow camo there, while OS's owner made a vivid camo the default ("also make the default spawn in gun for new users m40a3 with prism camo"); judges ranked removing it first or second in every round, a lead removed it, and the owner restored it, about 12 commits in 20 h. The default is the owner's call, not a house look.
|
|
157
|
+
- **Applies when:** Weapon or character skins on a hero object.
|
|
158
|
+
- **Verify:** A coverage sheet of every gun and finish; crops of three weapons in three lightings with a clipped-channel histogram; a per-finish GPU timing table from one boot, worn and unworn.
|
|
159
|
+
|
|
160
|
+
### An upgrade is a staged event, and firing never lights the held gun
|
|
161
|
+
- **Invariant:** An upgrade is a staged moment with its own sound on every upgraded shot; firing lights a held weapon for at most a frame, with no heat glow or colour building on it unless the design calls for one; ammo and status effects are seen in a capture before they count as done.
|
|
162
|
+
- **Technique:** Stage the upgrade as a sequence: intake, charge, return, pickup, inspect, voice line. Budget the held gun's shader cost, drive animated finishes by uniforms, and compile them all before play. Count hot pixels between flashes over a burst. Dump the compiled shader when a finish misbehaves.
|
|
163
|
+
- **Reference configuration:** OC: the upgrade looked "cheap" 3 times, and the spec that followed opens with a knuckle crack before the staged sequence; its status effects were reported "Not checked in a capture". LC: the muzzle flash had lit the held gun six frames of every seven in a burst, and orange pixels between flashes went from 1,615 and 2,597 to 0; an animated finish cost 0.5 ms (1.2 worst), one of five became fit, and a shader variable first declared inside one branch turned every lamp-lit plain weapon green.
|
|
164
|
+
- **Applies when:** Games with upgrade stations, animated finishes or special ammo.
|
|
165
|
+
- **Verify:** Capture an upgrade and an upgraded shot with sound, and film 30 rounds stepped, counting hot pixels between flashes.
|
|
166
|
+
|
|
167
|
+
### Wonder weapons feel special without breaking the frame or the mix
|
|
168
|
+
- **Invariant:** A wonder weapon is as finished as the best loadout gun, stays in its class's size band, is no louder than the other guns, and its effects stay inside the near-lens caps.
|
|
169
|
+
- **Technique:** Hold it to every rule above, then add character through shape, sound design and projectile, not through size, level or flash. A person listens to its sound before it ships.
|
|
170
|
+
- **Reference configuration:** OC's first loud wonder weapon fire sound was rejected as "too aggressive" and "distorted" (the loudness numbers are in `opengame-audio`), and the shipped wonder weapon's in-game peak of -7.4 dBFS stays within about 4 dB of the others. Projectile streaks are at least 56 px, and its hits stopped counting as explosions. Triangles stay at 20 to 25k like loadout guns, and camos had to work on it too.
|
|
171
|
+
- **Applies when:** Games with a signature or legendary weapon.
|
|
172
|
+
- **Verify:** Fire it beside a loadout gun at a wall 1 m away and at range: compare loudness, flash coverage and screen share, and have a person listen to both.
|
|
173
|
+
|
|
174
|
+
### Random rewards have written odds, a pity timer and a show
|
|
175
|
+
- **Invariant:** A random reward (a mystery box, a loot crate, a gacha pull) has written odds tuned against the reference game, a pity timer for its rarest reward, a visible rarity for every result, weights that keep the weakest results rare, and a reveal that is an event.
|
|
176
|
+
- **Technique:** Keep the weights in one table and compute the real odds of each result from it. Show rarity through colour, sound and light. Slow the reveal's ticks toward the result. Simulate many pulls and compare the results with the table.
|
|
177
|
+
- **Reference configuration:** OC: wonder weapon chance 3.4% per pull, guaranteed after 16 to 20 misses; rarity shown by beam colour (white, green, blue, purple, gold); crate weights tiered 12/8/5/4; the chance of the box leaving went to 10/20/30% from 15/30/50; the starter pistol's weight went from 12 to 4. Later tables gave the wonder weapon 1 in 56, then 2 of 446 (about 1 in 102), where World at War is about 1 in 20 to 25. The spin lasts 3.4 s (BO3 4.6 s), the offer 12 s, the move 8 s; ticks slow from about 16 a second to 3 with a music-box jingle. The box costs 950 points.
|
|
178
|
+
- **Applies when:** Games with randomised loot from a container, a drop table or a pull. OC's box is the reference here.
|
|
179
|
+
- **Verify:** Simulate ten thousand pulls from the shipped table and compare the counts with the intended odds, then play a pull and record the spin with sound.
|