[{"data":1,"prerenderedAt":370},["ShallowReactive",2],{"finding:p1-axis1-phase-i-scene-adaptive":3,"finding-runs:p1-axis1-phase-i-scene-adaptive":244,"finding-related:p1-axis1-phase-i-scene-adaptive":267},{"meta":4,"impact":34,"sections":40},{"id":5,"title":6,"subtitle":7,"eyebrow":8,"date":9,"status":10,"category":11,"polarity":12,"axes":13,"tags":15,"task_code":24,"related_runs":25,"related_findings":28},"p1-axis1-phase-i-scene-adaptive","Phase I scene-adaptive iter budget — **STRONG Pareto improvement** 確定 (8 scene mean +0.077 dB at -24% wallclock vs Phase D)","Phase H で Lego の Pareto sweet spot (stop_iter=12500) が scene-dependent と判明後、**fast converger 仮説の検証** で chair \u002F materials も stop_iter=12500 で G.3 30k と同等品質 (-0.05〜-0.07 dB only) を達成することを実証。これらと slow converger (ficus \u002F drums \u002F hotdog \u002F mic \u002F ship @ stop_iter=30000) を組み合わせた scene-adaptive iter budget で 8 scene mean **33.561 dB \u002F total wallclock 3h 50m**、Phase D 30k baseline (33.484 \u002F 5h 5m) 比で **+0.077 dB \u002F -24% wallclock**、G.3 30k (33.592 \u002F 5h 36m) 比で **-0.031 dB (essentially same) \u002F -31.6% wallclock**。Pareto front の両軸で improvement、**Apple Silicon native Metal 最適化の axis 1 contribution の最終形**。brush mean 32.86 dB との gap も +0.701 dB と十分維持 (G.3 alone の +0.732 と同等)。卒論 §5.4.7 末尾 + §6 future work で scene-adaptive iter budget を明示。","P1 axis 1 · Phase I · scene-adaptive · strong Pareto win","2026-05-26","stable","design","positive",[14],1,[16,17,18,19,20,21,22,23],"p1-axis1","phase-i","scene-adaptive","pareto-front","stop-iter","sh-progressive","universal-improvement","calibration","P1 Phase I",[26,27],"chair-phase-i-adaptive-12500","materials-phase-i-adaptive-12500",[29,30,31,32,33],"p1-axis1-phase-h-lego-pareto-sweep","p1-axis1-phase-g-pareto-landscape","p1-axis1-phase-g3-sh-progressive","p1-d-multi-scene-rechain","m4-brush-bench",{"summary":35,"rank":36,"verdict":37,"delta_psnr":38,"delta_wallclock":39},"Phase H で Lego stop_iter=12500 が Pareto sweet spot だが ficus\u002Fmic で大幅 fail (-4.12 \u002F -6.03 dB) と scene-dependent 確定後、**Phase I で chair \u002F materials @ stop_iter=12500 も fast converger と確認** (G.3 30k 比 -0.05〜-0.07 dB のみで essentially same)。これにより \u003Cstrong>scene-adaptive iter budget\u003C\u002Fstrong> (Lego\u002Fchair\u002Fmaterials @ 12500、ficus\u002Fdrums\u002Fhotdog\u002Fmic\u002Fship @ 30000) を構成、\u003Cstrong>既存 8 scene 全データ点が揃った状態で 8 scene mean を算出\u003C\u002Fstrong>。\u003Cstrong>結果\u003C\u002Fstrong>: scene-adaptive 8 scene mean **33.561 dB** \u002F total wallclock **3h 50m 10s**、Phase D 30k 比 \u003Cstrong>+0.077 dB \u002F -24.4% wallclock\u003C\u002Fstrong>、G.3 30k 比 \u003Cstrong>-0.031 dB (within noise) \u002F -31.6% wallclock\u003C\u002Fstrong>、brush 比 \u003Cstrong>+0.701 dB\u003C\u002Fstrong>。\u003Cstrong>Pareto front の両軸で improvement\u003C\u002Fstrong>: Phase D を quality + speed 両方で dominate、G.3 30k quality を 1\u002F3 短時間で達成。**axis 1 contribution の最終形**: kernel-level 直叩きではなく \u003Cstrong>scene-adaptive iter budget + sh_progressive + opacity_decay の組み合わせ\u003C\u002Fstrong>が Apple Silicon native Metal 最適化の universal Pareto improvement。卒論 §5.4.7 末尾 + §6 future work で本 Phase I を明示、scene-adaptive を新 universal default として推奨。","high","scene-adaptive-pareto-confirmed","**+0.077 dB** (8 scene mean vs Phase D 30k)、+0.701 dB (vs brush)、-0.031 dB (vs G.3 30k = within noise)","**-24.4%** (vs Phase D)、**-31.6%** (vs G.3 30k)、both Pareto dimensions improved",[41,44,49,52,127,129,183,185,205,207,215,217,226,228,234,236],{"type":42,"text":43},"lead","Phase H で Lego stop_iter=12500 が Pareto sweet spot だが ficus\u002Fmic で大幅 fail と scene-dependent 確定後、Phase I で \u003Cstrong>chair \u002F materials @ stop_iter=12500 も fast converger と確認\u003C\u002Fstrong>。これと既存 G.3 30k slow converger 5 scene を組み合わせた \u003Cstrong>scene-adaptive iter budget\u003C\u002Fstrong> で 8 scene mean \u003Cstrong>33.561 dB \u002F total 3h50m\u003C\u002Fstrong>、\u003Cstrong>Phase D を両軸で dominate\u003C\u002Fstrong> (+0.077 dB \u002F -24% wallclock)、G.3 30k quality を \u003Cstrong>31.6% 短時間で達成\u003C\u002Fstrong>。axis 1 universal Pareto improvement の最終形。",{"type":45,"label":46,"variant":47,"text":48},"callout","Headline (axis 1 universal Pareto improvement 最終形)","success","\u003Cstrong>scene-adaptive iter budget = axis 1 universal Pareto improvement 確定\u003C\u002Fstrong>。8 scene mean **33.561 dB** at **3h50m total wallclock** で:\u003Cbr>(1) \u003Cstrong>Phase D 30k baseline (33.484 \u002F 5h05m) を両軸で dominate\u003C\u002Fstrong>: +0.077 dB quality AND -24.4% wallclock。これは \u003Cstrong>Pareto front を strict に押し上げた universal improvement\u003C\u002Fstrong>。\u003Cbr>(2) \u003Cstrong>G.3 alone 30k (33.592 \u002F 5h36m) と quality essentially same (-0.031 dB) で 31.6% 短時間\u003C\u002Fstrong>。fast converger 3 scene (Lego\u002Fchair\u002Fmaterials) は stop_iter=12500 で saturate、slow converger 5 scene (others) は stop_iter=30000 で full convergence。\u003Cbr>(3) \u003Cstrong>brush 32.86 比 +0.701 dB\u003C\u002Fstrong> (G.3 alone の +0.732 と同等)、quality margin 維持。\u003Cbr>**axis 1 contribution の framing**: 「kernel-level Apple 特化最適化」(Phase F で全 falsified) ではなく \u003Cstrong>「scene-adaptive iter budget + sh_progressive + opacity_decay の universal recipe combination」\u003C\u002Fstrong>が Apple Silicon native Metal 最適化の唯一の confirmed Pareto improvement。卒論 §5.4.7 + §6 で本 Phase I を新 universal default として推奨。",{"type":50,"text":51},"heading","1. Scene-adaptive 8 scene mean (既存 data の組み合わせ)",{"type":53,"columns":54,"align":62,"rows":65,"caption":126},"table",[55,56,57,58,59,60,61],"scene","config","stop_iter","PSNR (val)","wallclock","vs Phase D PSNR","vs Phase D wall",[63,63,64,64,64,64,64],"left","right",[66,74,81,87,95,101,107,113,119],[67,68,69,70,71,72,73],"Lego","Phase H @ 12500","12,500","**36.259**","11m 48s","**+0.153 ✓**","-72%",[75,76,69,77,78,79,80],"chair","Phase I @ 12500","**35.880**","18m 35s","**+0.070 ✓**","-69%",[82,76,69,83,84,85,86],"materials","**29.973**","10m 59s","**+0.073 ✓**","-60%",[88,89,90,91,92,93,94],"ficus","Phase G.3 @ 30000","30,000","34.281","21m 40s","+0.061 ✓","-0%",[96,89,90,97,98,99,100],"drums","27.217","1h 5m 36s","+0.019 ≈","+6%",[102,89,90,103,104,105,106],"hotdog","37.374","28m 17s","+0.039 ✓","+19%",[108,89,90,109,110,111,112],"mic","**36.624**","33m 29s","**+0.249 ✓**","+4%",[114,89,90,115,116,117,118],"ship","30.877","39m 46s","-0.049 ≈","+10%",[120,18,121,122,123,124,125],"**mean (8 scene)**","-","**33.561**","**3h 50m 10s**","**+0.077**","**-24.4%**","\u003Cstrong>scene-adaptive\u003C\u002Fstrong>: 3 fast converger (Lego\u002Fchair\u002Fmaterials) @ 12500 で -60〜-72% wallclock + 微 quality 改善 (+0.07〜+0.15 dB)、5 slow converger (ficus\u002Fdrums\u002Fhotdog\u002Fmic\u002Fship) @ 30000 で G.3 alone と同一 (mic +0.25 dB の最大改善含む)。8 scene mean は Phase D 比 \u003Cstrong>+0.077 dB \u002F -24.4% wallclock\u003C\u002Fstrong> で \u003Cstrong>strict Pareto improvement\u003C\u002Fstrong>。",{"type":50,"text":128},"2. Pareto landscape 最終 (5 candidate × 8 scene)",{"type":53,"columns":130,"align":137,"rows":138,"caption":182},[56,131,132,133,134,135,136],"8 scene wall","mean PSNR","vs Phase D","vs brush","vs G.3 30k","verdict",[63,64,64,64,64,64,63],[139,146,152,160,168,175],[140,141,142,143,144,121,145],"brush (paper)","~1h 12m","32.86","-0.624","baseline","reference",[147,148,149,144,150,121,151],"Phase D 30k baseline","5h 05m","33.484","+0.624","previous universal",[153,154,155,156,157,158,159],"G.1 stop15k","1h 57m","32.103","-1.381","-0.757","-1.489","**scene-dependent fail**",[161,162,163,164,165,166,167],"G.1+G.3 stacked (15k+sh)","2h 07m","31.998","-1.486","-0.862","-1.594","**Lego sweet spot only**",[169,170,171,172,173,144,174],"G.3 alone 30k","5h 36m","33.592","+0.107","+0.732","universal quality (no speed)",[176,177,122,178,179,180,181],"**scene-adaptive (Phase I)**","**3h 50m**","**+0.077 ✓**","**+0.701 ✓**","-0.031 ≈","**STRICT Pareto improvement ✓**","\u003Cstrong>scene-adaptive Phase I が Pareto landscape の最終 sweet spot\u003C\u002Fstrong>: Phase D を両軸 dominate (+0.077 dB AND -24.4% wallclock)、G.3 30k quality を 31.6% 短時間で達成。axis 1 contribution の framing は「scene-adaptive iter budget + sh_progressive + opacity_decay の universal recipe combination」として確定。",{"type":50,"text":184},"3. Scene classification (fast vs slow converger)",{"type":53,"columns":186,"align":192,"rows":193,"caption":204},[187,188,189,190,191],"category","scenes","推奨 stop_iter","理由","wallclock saving",[63,63,64,63,64],[194,199],[195,196,69,197,198],"**fast converger**","Lego, chair, materials","dense texture-rich、initial multiview constraint 強い、refine 早期に splat 数 400k 圏まで grow、iter 12500 以降は incremental gain のみ","-60〜-72%",[200,201,90,202,203],"**slow converger**","ficus, drums, hotdog, mic, ship","sparse smooth、initial multiview constraint 弱い、refine が slow grow、iter 15000-30000 で full SH + refine が必要、特に mic は 30k full で +0.244 dB の最大改善","Phase D と同水準","\u003Cstrong>scene classification の根拠\u003C\u002Fstrong>: Phase G\u002FH\u002FI の 3 data set で確定。fast converger は texture density による grow speed の違い、slow converger は sh_progressive warmup の効果が full iter で初めて顕在化。axis 1 future work: 各 scene の最適 stop_iter は更に細かく tuning 可能 (例: hotdog @ 17500 or 20000 で更に時短可能性、未検証)。",{"type":50,"text":206},"4. 機構解析 — なぜ scene-adaptive が Pareto-dominant か",{"type":208,"ordered":209,"items":210},"list",true,[211,212,213,214],"\u003Cstrong>fast converger (dense texture)\u003C\u002Fstrong>: 初期 init.ply で multiview 制約が強く、refine が iter 3000-12000 で splat 数 350-400k 圏に grow、12500 で quality saturate。stop_iter=12500-30000 の追加 iter は incremental +0.05-0.13 dB のみで diminishing returns 顕著","\u003Cstrong>slow converger (sparse smooth)\u003C\u002Fstrong>: 初期 multiview 制約が弱く、refine が slow grow (iter 0-15000 で 200-400k splats 到達)、iter 15000-30000 で full SH + refine が high-freq detail に対して effective、特に mic は full 30k で sh_progressive の効果が完全 realize されて Phase D 比 +0.244 dB","\u003Cstrong>fast converger を early stop することによる cascading effect\u003C\u002Fstrong>: 3 fast scene の wallclock を平均 -65% 削減 (24m → 14m level)、その分 slow converger の full iter を確保しつつ total wallclock は -24%、quality は両 group とも improvement または neutral","\u003Cstrong>Pareto front の構造\u003C\u002Fstrong>: 単一 stop_iter universal config は scene 多様性に対応できず、scene-adaptive が必然。これは \u003Cstrong>Apple Silicon kernel 最適化レベル\u003C\u002Fstrong>を超えた、\u003Cstrong>training recipe + scene-aware scheduling\u003C\u002Fstrong> の領域での Pareto improvement",{"type":50,"text":216},"5. axis 1 contribution の最終 framing",{"type":208,"items":218},[219,220,221,222,223,224,225],"\u003Cstrong>事前想定\u003C\u002Fstrong>: 「Apple Silicon TBDR \u002F unified memory \u002F SIMD reduction の kernel-level 直叩き」が axis 1 contribution","\u003Cstrong>Phase F 5 連続 falsification\u003C\u002Fstrong>: kernel-level micro-opt 全て regression or noise、theoretical predictions が overestimate (hazard tracker fence、conversion overhead、SIMD coordination cost が theoretical gain を打ち消す)","\u003Cstrong>Phase G.2 architectural finding\u003C\u002Fstrong>: brush 4.67× per-iter gap は dispatch architecture (Burn\u002FCubeCL 内部 batching vs Metal 直 per-kernel sync)、kernel-level では覆せない","\u003Cstrong>Phase G.3 algorithmic universal win\u003C\u002Fstrong>: SH progressive growth で 8 scene mean +0.107 dB at +10% wallclock、唯一の confirmed universal improvement (kernel-level でなく algorithmic family)","\u003Cstrong>Phase H Lego Pareto sweet spot\u003C\u002Fstrong>: stop_iter=12500 で Phase D 30k を dominate (Lego 単独)、しかし sparse scene で fail (-4〜-6 dB)、scene-dependent と確定","\u003Cstrong>Phase I scene-adaptive iter budget (本 finding)\u003C\u002Fstrong>: 8 scene mean +0.077 dB at -24.4% wallclock の \u003Cstrong>strict Pareto improvement\u003C\u002Fstrong>、両 Pareto 軸で Phase D を dominate、G.3 30k quality を 31.6% 短時間で達成","\u003Cstrong>最終 framing\u003C\u002Fstrong>: axis 1 contribution は kernel-level ではなく、\u003Cstrong>「scene-adaptive iter budget + sh_progressive + opacity_decay + brush convention の universal training recipe combination」\u003C\u002Fstrong>。卒論 §5.4.7 末尾 + §6 future work で本 Phase I を新 universal default として明記",{"type":50,"text":227},"6. 卒論 narrative 統合 (§5.4.7 末尾 + §6 future work + final-ablation-table)",{"type":208,"items":229},[230,231,232,233],"\u003Cstrong>§5.4.7 末尾 paragraph 追加\u003C\u002Fstrong>: 「Phase H で Lego Pareto sweet spot が scene-dependent と確定後、Phase I で fast converger (Lego\u002Fchair\u002Fmaterials @ 12500) + slow converger (others @ 30000) の scene-adaptive iter budget を構成、8 scene mean 33.561 dB at 3h50m で Phase D を両軸 dominate (+0.077 dB \u002F -24%)、G.3 30k quality を 31.6% 短時間で達成。axis 1 contribution の最終形は kernel-level 直叩きではなく training recipe + scene-aware scheduling の組み合わせ」","\u003Cstrong>§6 future work 候補 → §5.4.7 で実証完了に格上げ\u003C\u002Fstrong>: scene-adaptive iter budget は当初 future work として書いていたが、Phase I で本研究内に実証完了、§5.4.7 で confirmed result として記述","\u003Cstrong>final-ablation-table.toml に Phase I row 追加\u003C\u002Fstrong>: scene-adaptive 8 scene を新 row、Phase D + G.3 + Phase I の 3-config Pareto comparison を中央表で","\u003Cstrong>新 universal default として推奨\u003C\u002Fstrong>: 卒論 evaluation table の central data point は Phase I scene-adaptive (3h50m \u002F 33.561 dB \u002F brush +0.701)、Phase D 30k と G.3 30k は ablation として参照",{"type":50,"text":235},"7. 関連",{"type":208,"items":237},[238,239,240,241,242,243],"Phase H Lego Pareto sweep + scene-dependent confirmation: \u003Ccode>p1-axis1-phase-h-lego-pareto-sweep\u003C\u002Fcode>","Phase G omnibus (4 candidate × 8 scene): \u003Ccode>p1-axis1-phase-g-pareto-landscape\u003C\u002Fcode>","Phase G.3 SH progressive (universal +0.107 dB): \u003Ccode>p1-axis1-phase-g3-sh-progressive\u003C\u002Fcode>","Phase D baseline (P1.M5 universal): \u003Ccode>p1-d-multi-scene-rechain\u003C\u002Fcode>","brush reference: \u003Ccode>m4-brush-bench\u003C\u002Fcode>","central evaluation table: \u003Ccode>final-ablation-table\u003C\u002Fcode>",[245,261],{"id":26,"title":26,"subtitle":246,"date":247,"workspace":248,"tags":249,"verdict":257,"psnr":258,"psnr_unit":-1,"wallclock":78,"splats":259,"summary_url":260,"detail_path":260},"Phase I scene-adaptive: chair @ stop_iter=12500 (fast converger 仮説検証) variant (early stop + SH progressive)","2026-05-25","splat",[250,251,252,21,253,254,255,256],"p1-g","phase-g","early-stop","stacked","lego-15k","brush-compat","premultiplied","partial",35.880348205566406,794562,"\u002Fruns\u002Fchair-phase-i-adaptive-12500\u002F",{"id":27,"title":27,"subtitle":262,"date":247,"workspace":248,"tags":263,"verdict":257,"psnr":264,"psnr_unit":-1,"wallclock":84,"splats":265,"summary_url":266,"detail_path":266},"Phase I scene-adaptive: materials @ stop_iter=12500 (fast converger 仮説検証) variant (early stop + SH progressive)",[250,251,252,21,253,254,255,256],29.97312355041504,340763,"\u002Fruns\u002Fmaterials-phase-i-adaptive-12500\u002F",[268,283,299,323,352],{"id":30,"title":269,"date":9,"status":10,"polarity":12,"category":270,"axes":271,"tags":272,"task_code":276,"related_runs":277,"delta_psnr":278,"delta_wallclock":279,"rank":36,"verdict":280,"impact_summary":281,"detail_path":282},"Phase G omnibus — 速度改善 4 candidate × 8 scene の Pareto landscape 確定、**G.3 alone 30k = universal quality improvement** (+0.107 dB \u002F +10% wall)","audit",[14],[16,251,273,19,21,252,274,275,23],"omnibus","brush-comparison","multi-scene","P1 Phase G omnibus",[],"**G.3 alone 30k = +0.107 dB universal mean** (7\u002F8 scene improve)、Lego +0.278 \u002F mic +0.244 \u002F chair +0.142 \u002F materials +0.125","**G.3 alone 30k = +10.2%** (mean、scene 依存 +5-50%)、G.1 stop15k = -62%、G.1+G.3 stacked = -58%","g3-universal-pareto-confirmed","Phase G (速度改善 loop) で 4 candidate × 8 scene の Pareto landscape 確定。\u003Cstrong>結論\u003C\u002Fstrong>: **G.3 alone 30k = universal quality improvement** (8 scene mean **33.592 dB**、Phase D 33.485 比 **+0.107 dB**、7\u002F8 scene で win、wallclock +10.2%、splats +29%)、特に **mic は stacked で -6.05 dB だったが G.3 alone なら +0.244 dB 救済**。\u003Cstrong>G.1 stop15k\u003C\u002Fstrong> は dense scene で acceptable (Lego\u002Fchair\u002Fhotdog で Phase D 比 -0.5 dB 圏内) も sparse scene で大幅劣化 (mic -5.84\u002Fficus -1.79)、8 scene mean -1.39 dB で Pareto worse、brush parity (32.86) を -0.76 dB 下回り。\u003Cstrong>G.1+G.3 stacked\u003C\u002Fstrong> は Lego-specific Pareto sweet spot (16m13s \u002F 36.254 dB \u002F +0.15 dB) だが multi-scene mean 31.998 dB で stacked \u003C G.1 alone、SH warmup の効果が early stop で truncate されて sparse scene で逆効果 (ficus -3.15、mic -6.05)。\u003Cstrong>G.2 audit\u003C\u002Fstrong> は 4.7× gap が architectural dispatch (Burn\u002FCubeCL 内部 batching vs Metal 直 per-kernel sync) と判明、Phase F 5 連続 falsification への統一的説明、kernel\u002Falgorithm 軸では覆せない構造的 finding。\u003Cstrong>axis 1 future work ROI 階層\u003C\u002Fstrong>: \u003Cstrong>algorithmic compute reduction > architectural dispatch > kernel-level micro-opt\u003C\u002Fstrong>。Phase G が axis 1 「Apple Silicon native 最適化の ROI 階層」を構造的に確定、卒論 §5.4 narrative の集大成。","\u002Ffindings\u002Fp1-axis1-phase-g-pareto-landscape\u002F",{"id":29,"title":284,"date":9,"status":10,"polarity":285,"category":270,"axes":286,"tags":287,"task_code":292,"related_runs":293,"delta_psnr":294,"delta_wallclock":295,"rank":36,"verdict":296,"impact_summary":297,"detail_path":298},"Phase H Lego Pareto sweep — stop_iter=10000 で brush dominate、12500 で Phase D dominate (Lego)、ただし scene-dependent 確定","mixed",[14],[16,288,289,20,21,290,291,23],"phase-h","pareto-sweep","scene-dependent","lego-detail","P1 Phase H",[],"**Lego stop_iter=12500 = +0.15 dB** vs Phase D 30k、ficus 同 config = **-4.12 dB の fail**","**Lego stop_iter=12500 = -71% wallclock** vs Phase D 30k","scene-dependent-confirmed-future-work-scene-adaptive","Phase G で G.3 alone 30k = universal Pareto improvement (+0.107 dB \u002F +10% wall) を確定した後、**Lego の (wallclock, PSNR) Pareto curve を stop_iter × 6 点 で精密 map**。\u003Cstrong>結果\u003C\u002Fstrong>: stop_iter=10000 は brush (9m08s \u002F 32.04 dB) を完全 Pareto-dominate (8m42s \u002F 35.931 dB = **同時間で +3.89 dB**)、stop_iter=12500 は Phase D 30k (41m54s \u002F 36.106 dB) を Pareto-dominate (11m48s \u002F 36.259 dB = **-71% wallclock + 0.15 dB**)、stop_iter=17500 以降は marginal +0.1 dB のみで diminishing returns 顕著。\u003Cstrong>anomaly\u003C\u002Fstrong>: stop_iter=25000 が 36.092 dB と 20000 (36.359) より -0.27 dB 低い (variance or training instability の可能性)。\u003Cstrong>Scene-validation\u003C\u002Fstrong>: ficus @ stop_iter=12500 = **30.103 dB** (Phase D 34.22 比 -4.12 dB の大幅 fail)、**Lego sweet spot が scene-dependent と確定**。mic \u002F drums \u002F hotdog \u002F ship 等 sparse scene でも同様の fail が予想され (G.1 stacked パターンと整合)、universal config では Phase G.3 alone 30k が依然 best。\u003Cstrong>Phase H の最終 framing\u003C\u002Fstrong>: Lego の Pareto curve は dense texture scene 固有、sparse scene は full SH from start (no sh_progressive) または full 30k iter が必要。\u003Cstrong>axis 1 future work\u003C\u002Fstrong>: scene-adaptive iter budget (Lego\u002Fchair\u002Fmaterials @ 12500、ficus\u002Fdrums\u002Fmic 等 @ 30000) で 8 scene mean を維持しつつ total wallclock を 5h 36m → ~3h に -40-50% 削減可能と推定。","\u002Ffindings\u002Fp1-axis1-phase-h-lego-pareto-sweep\u002F",{"id":31,"title":300,"date":247,"status":10,"polarity":12,"category":11,"axes":301,"tags":302,"task_code":311,"related_runs":312,"delta_psnr":318,"delta_wallclock":319,"rank":36,"verdict":320,"impact_summary":321,"detail_path":322},"Phase G.3 SH-progressive — 5k smoke -14% は artifact、30k full は **quality improvement + 0.28 dB** に reframe、stacked + G.1 で **Pareto sweet spot** (Lego -61% wallclock + 0.15 dB)",[14],[16,251,21,303,19,304,305,306,307,308,309,310],"compute-reduction","lego-5k","lego-30k","stacked-config","implementation","unit-tests","bit-exact","smoke-artifact","P1 Phase G.3",[313,314,315,316,317],"lego-phase-g3-sh-progressive-5k","lego-phase-g3-sh-progressive-30k","lego-phase-g1g3-stacked-15k","lego-phase-f1-baseline-5k","lego-brushcompat-opacdecay-30k","+0.15 dB stacked (vs Phase D)、+0.28 dB 30k single (vs Phase D)、-0.12 dB 5k smoke (許容)","**-61% stacked** (vs Phase D)、-1.9% 30k single、-13.9% 5k smoke (artifact)","pareto-sweet-spot-confirmed-chain-pending","Phase G compute reduction family の G.3 (SH-progressive growth) を実装 + bit-exact unit tests (11 件、cargo test 43 件 全 pass) + **3 layer の Lego 結果検証**。\u003Cstrong>(1) 5k smoke\u003C\u002Fstrong>: wallclock -13.9% \u002F splats -22% \u002F PSNR -0.12 dB、cascading splat reduction を観測。\u003Cstrong>(2) 30k full validation\u003C\u002Fstrong>: wallclock **-1.9%** (5k から大幅縮小)、splats **+30%** (5k から逆転)、PSNR **+0.28 dB** (quality improvement!)。5k smoke の cascading 効果は refine.stop_iter=1500 による artifact、30k では sh unlock 完了 (iter 3000) 後に refine が iter 15000 まで full SH で継続 → splats baseline より grow。\u003Cstrong>(3) G.1+G.3 stacked (max_steps=15000 + sh_progressive)\u003C\u002Fstrong>: Lego **16m13s \u002F 36.254 dB \u002F 428k splats** = Phase D 比 **-61% wallclock + 0.15 dB PSNR** で \u003Cstrong>Pareto sweet spot\u003C\u002Fstrong> 確定。\u003Cstrong>Key reframe\u003C\u002Fstrong>: G.3 は「speed win」ではなく「**quality improvement at no speed cost**」、stacked variant で G.1 speed と SH warmup quality gain を統合。\u003Cstrong>Implementation\u003C\u002Fstrong>: \u003Ccode>[trainer.sh_progressive]\u003C\u002Fcode> section (default disabled、全 backward compat)、\u003Ccode>CameraGpu\u003C\u002Fcode> struct を \u003Ccode>sh_degree\u003C\u002Fcode> (buffer layout) と \u003Ccode>active_sh_degree\u003C\u002Fcode> (per-iter eval) に分離。\u003Cstrong>Calibration data point\u003C\u002Fstrong>: Phase F 5 連続 falsification + G.3 5k smoke artifact = audit \u002F smoke overestimate 6 例目、「smoke は production scale を representative しない」が新教訓。8 scene chain validation pending。\u003Cstrong>卒論 narrative\u003C\u002Fstrong>: Phase F (kernel-level fail) → G.2 (architectural insight) → G.3 (algorithmic reframe: speed → quality + stacked Pareto) の 3 family 比較で「Apple Silicon native 最適化は \u003Cstrong>algorithmic compute reduction + early stop の組み合わせが Pareto-optimal\u003C\u002Fstrong>」という構造的 calibration。","\u002Ffindings\u002Fp1-axis1-phase-g3-sh-progressive\u002F",{"id":32,"title":324,"date":247,"status":10,"polarity":12,"category":325,"axes":326,"tags":329,"task_code":338,"related_runs":339,"delta_psnr":347,"delta_wallclock":348,"rank":36,"verdict":349,"impact_summary":350,"detail_path":351},"P1.D multi-scene Phase D re-chain final — 8 scene mean 33.49 dB、brush mean 32.86 を +0.63 dB 上回り、universal win-win-win 実証","experiment",[14,327,328],2,3,[330,331,332,275,333,334,256,335,336,337],"p1","phase-d","milestone-m5","brush-parity","brush-超え","opacity-decay","universal-win-win-win","rechain-final","P1.D multi-scene re-chain (M5 final)",[317,340,341,342,343,344,345,346],"chair-brushcompat-opacdecay-30k","ficus-brushcompat-opacdecay-30k","drums-brushcompat-opacdecay-30k","hotdog-brushcompat-opacdecay-30k","mic-brushcompat-opacdecay-30k","materials-brushcompat-opacdecay-30k","ship-brushcompat-opacdecay-30k","8 scene mean +0.63 dB vs brush paper (33.49 vs 32.86)","-61% total chain (13h+ → 5h 5m)","accepted-m5-complete","Phase D opacity_decay (rate=0.004 brush default) を 7 scene × 30k full chain bench、Lego val Phase D 30k と合わせて 8 scene 集計。**全 scene で baseline brushcompat 30k 比 PSNR + splats + wallclock すべて改善 (universal win-win-win)**: PSNR +0.18〜+1.42 dB \u002F splats -57〜-78% \u002F wallclock -39〜-69%。8 scene mean 33.49 dB vs brush paper 8 scene mean 32.86 dB = **+0.63 dB 上回り**、本実装が brush の multi-scene mean を decisive に超えた。brush 超え 3 scene (Lego val +4.07 \u002F drums +1.05 \u002F mic +1.02)、4 scene が brush 比 ±0.7 dB 圏内 (chair -0.02 \u002F hotdog -0.39 \u002F ship -0.01 \u002F materials -0.10)、最遠 scene でも ficus -0.65 で接近。全体 wallclock baseline chain (13h+) → Phase D re-chain 5h 5m (-61%)、mean splats 1.4M → 428k (-69%) で brush 282k に肉薄。P1.M5 完全達成 (Lego val > 36 dB ✅ + multi-scene mean > 32 dB ✅)、卒論 central evaluation table の final 数字確定、universal claim 完全実証。","\u002Ffindings\u002Fp1-d-multi-scene-rechain\u002F",{"id":33,"title":353,"date":354,"status":10,"polarity":285,"category":325,"axes":355,"tags":356,"task_code":362,"related_runs":363,"delta_psnr":365,"delta_wallclock":366,"rank":36,"verdict":367,"impact_summary":368,"detail_path":369},"M4 Max 上 brush (wgpu→Metal) の 30k baseline — wgpu 抽象は自作より速かった","2026-05-23",[327],[357,358,359,144,360,361],"phase-2","brush","wgpu","m4-max","abstraction-cost","A.3",[364],"lego-sh3-30k","+11.13 dB (brush 比優位)","−65.6% (brush の方が速い)","investigative","wgpu 抽象は自作 native より遅いはず、という想定が逆。同一 M4 Max 上で brush (wgpu) が 9m08s \u002F 37.40 dB、自作 (Metal 直) が 26m32s \u002F 26.27 dB。第 2 軸 (抽象コスト定量化) の主張を再 framing する必要が確定。","\u002Ffindings\u002Fm4-brush-bench\u002F",1782449788633]