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 / 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 / drums / hotdog / mic / ship @ stop_iter=30000) を組み合わせた scene-adaptive iter budget で 8 scene mean **33.561 dB / total wallclock 3h 50m**、Phase D 30k baseline (33.484 / 5h 5m) 比で **+0.077 dB / -24% wallclock**、G.3 30k (33.592 / 5h 36m) 比で **-0.031 dB (essentially same) / -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。
Phase H で Lego stop_iter=12500 が Pareto sweet spot だが ficus/mic で大幅 fail と scene-dependent 確定後、Phase I で chair / materials @ stop_iter=12500 も fast converger と確認。これと既存 G.3 30k slow converger 5 scene を組み合わせた scene-adaptive iter budget で 8 scene mean 33.561 dB / total 3h50m、Phase D を両軸で dominate (+0.077 dB / -24% wallclock)、G.3 30k quality を 31.6% 短時間で達成。axis 1 universal Pareto improvement の最終形。
scene-adaptive iter budget = axis 1 universal Pareto improvement 確定。8 scene mean **33.561 dB** at **3h50m total wallclock** で:
(1) Phase D 30k baseline (33.484 / 5h05m) を両軸で dominate: +0.077 dB quality AND -24.4% wallclock。これは Pareto front を strict に押し上げた universal improvement。
(2) G.3 alone 30k (33.592 / 5h36m) と quality essentially same (-0.031 dB) で 31.6% 短時間。fast converger 3 scene (Lego/chair/materials) は stop_iter=12500 で saturate、slow converger 5 scene (others) は stop_iter=30000 で full convergence。
(3) brush 32.86 比 +0.701 dB (G.3 alone の +0.732 と同等)、quality margin 維持。
**axis 1 contribution の framing**: 「kernel-level Apple 特化最適化」(Phase F で全 falsified) ではなく 「scene-adaptive iter budget + sh_progressive + opacity_decay の universal recipe combination」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。
| scene | config | stop_iter | PSNR (val) | wallclock | vs Phase D PSNR | vs Phase D wall |
|---|---|---|---|---|---|---|
| Lego | Phase H @ 12500 | 12,500 | **36.259** | 11m 48s | **+0.153 ✓** | -72% |
| chair | Phase I @ 12500 | 12,500 | **35.880** | 18m 35s | **+0.070 ✓** | -69% |
| materials | Phase I @ 12500 | 12,500 | **29.973** | 10m 59s | **+0.073 ✓** | -60% |
| ficus | Phase G.3 @ 30000 | 30,000 | 34.281 | 21m 40s | +0.061 ✓ | -0% |
| drums | Phase G.3 @ 30000 | 30,000 | 27.217 | 1h 5m 36s | +0.019 ≈ | +6% |
| hotdog | Phase G.3 @ 30000 | 30,000 | 37.374 | 28m 17s | +0.039 ✓ | +19% |
| mic | Phase G.3 @ 30000 | 30,000 | **36.624** | 33m 29s | **+0.249 ✓** | +4% |
| ship | Phase G.3 @ 30000 | 30,000 | 30.877 | 39m 46s | -0.049 ≈ | +10% |
| **mean (8 scene)** | scene-adaptive | - | **33.561** | **3h 50m 10s** | **+0.077** | **-24.4%** |
| config | 8 scene wall | mean PSNR | vs Phase D | vs brush | vs G.3 30k | verdict |
|---|---|---|---|---|---|---|
| brush (paper) | ~1h 12m | 32.86 | -0.624 | baseline | - | reference |
| Phase D 30k baseline | 5h 05m | 33.484 | baseline | +0.624 | - | previous universal |
| G.1 stop15k | 1h 57m | 32.103 | -1.381 | -0.757 | -1.489 | **scene-dependent fail** |
| G.1+G.3 stacked (15k+sh) | 2h 07m | 31.998 | -1.486 | -0.862 | -1.594 | **Lego sweet spot only** |
| G.3 alone 30k | 5h 36m | 33.592 | +0.107 | +0.732 | baseline | universal quality (no speed) |
| **scene-adaptive (Phase I)** | **3h 50m** | **33.561** | **+0.077 ✓** | **+0.701 ✓** | -0.031 ≈ | **STRICT Pareto improvement ✓** |
| category | scenes | 推奨 stop_iter | 理由 | wallclock saving |
|---|---|---|---|---|
| **fast converger** | Lego, chair, materials | 12,500 | 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% |
| **slow converger** | ficus, drums, hotdog, mic, ship | 30,000 | 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 |
p1-axis1-phase-h-lego-pareto-sweepp1-axis1-phase-g-pareto-landscapep1-axis1-phase-g3-sh-progressivep1-d-multi-scene-rechainm4-brush-benchfinal-ablation-tableこの finding が観測された / 言及している実験 run。