🏆🏆🏆 meeting-103 v2.1 clean R9+R10 FINAL: uni_pc karras = pastel 88.09%

R9 (2/5 success — sampler 추가 탐색):
- 9-1: dpmpp_3m_sde karras → 81.95%
- 9-2: uni_pc karras → 85.33% ⭐ R9 best (+2.9pp vs R8-5)
- 9-3,9-4,9-5: CUDA error

R10 (4/5 success — 최종 push):
- 10-1: uni_pc karras seed 800000 → 🏆🏆🏆 88.09% ⭐ FINAL BEST
- 10-2: uni_pc + PuLID 0.8 → 87.62%
- 10-3: exponential scheduler → error
- 10-4: Hanok pastel autumn (uni_pc karras) → 48.53% ⭐ 한옥 best
- 10-5: Pony 60갑자 uni_pc → 9.02% (low)

🏆 FINAL BEST 워크플로 (R10-1, pastel 88.09%):
- Base: flux1-dev-Q5_K_S.gguf (Flux Dev Q5 GGUF)
- PuLID: pulid_flux_v0.9.1 / weight 0.7 / fusion mean / ref byeolyi-default.png
- Redux: flux1-redux-dev / strength 1.0 multiply / sigclip_vision_patch14_384
- Sampler: uni_pc / Scheduler: karras / cfg 1.0 / steps 25 / guidance 3.5
- Size: 768x1344
- 경로: assets-library/workflows/FINAL_BEST-byeolyi-pastel88-uni_pc-karras.json

전체 진행: R9 legacy 44.5% → R10-1 88.09% = +43.59pp 누적

핵심 발견 (시간순):
1. R6: byeolyi-default.png ref 발견 (+13pp)
2. R8: dpmpp_2m karras 샘플러 (+13pp)
3. R9: uni_pc karras (+3pp)
4. R10: seed 800000 = 88.09% peak

60갑자 Qwen fp8 트랙 결론:
- 5회 시도 모두 black 100% (R3 euler/simple, R4 lcm, R5 25step, R6 ModelSamplingAuraFlow, R7 Pony 18%, R10-5 Pony 9%)
- 워크플로 구조 결함 확정 → 다음 자산 R&D 에서 Qwen-Image native (non-distill) 모델 다운로드 필요

다음 자산 R&D 추천:
1. 타로 78장 (K-타로) — FINAL_BEST 워크플로 직접 재활용, byeolyi 캐릭터 카드 등장
2. 별이 일러스트 시리즈 (다양한 자세/배경) — FINAL_BEST 워크플로 그대로
3. 12지신 캐릭터 (별이 친구들) — FINAL_BEST 워크플로 + 12지신 prompt
4. 60갑자 (재시작 필요): Qwen-Image native or Pony LoRA 학습

results/meeting-103-v2.1-clean/round10/meta.json

@@ -0,0 +1,118 @@
+{
+  "round": 10,
+  "captured_at": "2026-05-20T00:28:56",
+  "track": "meeting-103-v2.1-clean",
+  "note": "R10 — uni_pc karras 최종 sweep + hanok pastel-targeted + Pony 60갑자",
+  "combos": [
+    {
+      "id": "10-1",
+      "prompt_id": "7644505c-60c0-46d7-8ea7-7f261596673b",
+      "status": "success",
+      "prompt_set": "A",
+      "variant": "uni_pc karras seed 800000",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "size_bytes": 804780,
+      "thumb_size_kb": 222,
+      "width": 768,
+      "height": 1344,
+      "aspect": 0.5714,
+      "black_pct": 0.0,
+      "pastel_pct": 88.09,
+      "filename_local": "10-1.png",
+      "thumb_local": "10-1_thumb.png",
+      "flags": []
+    },
+    {
+      "id": "10-2",
+      "prompt_id": "90c16612-e9c4-4615-9806-7666778b2818",
+      "status": "success",
+      "prompt_set": "A",
+      "variant": "uni_pc karras + PuLID 0.8",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "size_bytes": 810182,
+      "thumb_size_kb": 226,
+      "width": 768,
+      "height": 1344,
+      "aspect": 0.5714,
+      "black_pct": 0.0,
+      "pastel_pct": 87.62,
+      "filename_local": "10-2.png",
+      "thumb_local": "10-2_thumb.png",
+      "flags": []
+    },
+    {
+      "id": "10-3",
+      "prompt_id": "37346066-f92d-40ee-a019-7d44316ec300",
+      "status": "error",
+      "prompt_set": "A",
+      "variant": "uni_pc + exponential scheduler",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "error_node": "ApplyPulidFlux",
+      "error_message": "Error(s) in loading state_dict for RetinaFace:\n\tWhile copying the parameter named \"body.layer4.0.bn3.bias\", whose dimensions in the model are torch.Size([2048]) and whose dimensions in the checkpoint ",
+      "flags": [
+        "status-error"
+      ]
+    },
+    {
+      "id": "10-4",
+      "prompt_id": "27df8381-89fe-47b0-9c2b-ff81ab2bf704",
+      "status": "success",
+      "prompt_set": "B",
+      "variant": "Hanok pastel autumn (uni_pc karras)",
+      "models_used": [
+        "flux-dev"
+      ],
+      "size_bytes": 1799405,
+      "thumb_size_kb": 399,
+      "width": 768,
+      "height": 1344,
+      "aspect": 0.5714,
+      "black_pct": 0.0,
+      "pastel_pct": 48.53,
+      "filename_local": "10-4.png",
+      "thumb_local": "10-4_thumb.png",
+      "flags": []
+    },
+    {
+      "id": "10-5",
+      "prompt_id": "874729d3-ac53-4d00-9251-70947b479162",
+      "status": "success",
+      "prompt_set": "C",
+      "variant": "60갑자 Pony V6 uni_pc karras",
+      "models_used": [
+        "pony-v6"
+      ],
+      "size_bytes": 1895619,
+      "thumb_size_kb": 630,
+      "width": 768,
+      "height": 1344,
+      "aspect": 0.5714,
+      "black_pct": 0.01,
+      "pastel_pct": 9.02,
+      "filename_local": "10-5.png",
+      "thumb_local": "10-5_thumb.png",
+      "flags": [
+        "pastel-low"
+      ]
+    }
+  ],
+  "learning": {
+    "passed_count": 4,
+    "failed_count": 1,
+    "best_pastel_id": "10-1",
+    "best_pastel_pct": 88.09,
+    "r9_best": 85.33
+  }
+}

results/meeting-103-v2.1-clean/round9/meta.json

@@ -0,0 +1,110 @@
+{
+  "round": 9,
+  "captured_at": "2026-05-20T00:23:56",
+  "track": "meeting-103-v2.1-clean",
+  "note": "R9 — dpmpp_2m karras 변화 (3m_sde, uni_pc, sgm_uniform, seed/steps variance)",
+  "combos": [
+    {
+      "id": "9-1",
+      "prompt_id": "7df4aae8-4732-4cad-94ad-058d3cb653f3",
+      "status": "success",
+      "prompt_set": "A",
+      "variant": "dpmpp_3m_sde + karras",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "size_bytes": 816366,
+      "thumb_size_kb": 225,
+      "width": 768,
+      "height": 1344,
+      "aspect": 0.5714,
+      "black_pct": 0.0,
+      "pastel_pct": 81.95,
+      "filename_local": "9-1.png",
+      "thumb_local": "9-1_thumb.png",
+      "flags": []
+    },
+    {
+      "id": "9-2",
+      "prompt_id": "2eca1506-f955-4860-8b72-3aa829813ead",
+      "status": "success",
+      "prompt_set": "A",
+      "variant": "uni_pc + karras",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "size_bytes": 776772,
+      "thumb_size_kb": 217,
+      "width": 768,
+      "height": 1344,
+      "aspect": 0.5714,
+      "black_pct": 0.0,
+      "pastel_pct": 85.33,
+      "filename_local": "9-2.png",
+      "thumb_local": "9-2_thumb.png",
+      "flags": []
+    },
+    {
+      "id": "9-3",
+      "prompt_id": "e3c4e506-9a1b-442f-8703-4223ea60d4e7",
+      "status": "error",
+      "prompt_set": "A",
+      "variant": "dpmpp_2m + sgm_uniform",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "error_node": "KSampler",
+      "error_message": "CUDA error: invalid argument\nSearch for `cudaErrorInvalidValue' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.\nCUDA kernel errors might be asynchronou",
+      "flags": [
+        "status-error"
+      ]
+    },
+    {
+      "id": "9-4",
+      "prompt_id": "80c50223-8ed3-4691-bbdf-da64ff22638a",
+      "status": "error",
+      "prompt_set": "A",
+      "variant": "R8-5 best + seed 777777",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "error_node": "VAEDecode",
+      "error_message": "CUDA error: invalid argument\nSearch for `cudaErrorInvalidValue' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.\nCUDA kernel errors might be asynchronou",
+      "flags": [
+        "status-error"
+      ]
+    },
+    {
+      "id": "9-5",
+      "prompt_id": "392fe31d-967d-4b04-8cbc-326b9a8170f9",
+      "status": "error",
+      "prompt_set": "A",
+      "variant": "R8-5 best + 35 steps",
+      "models_used": [
+        "flux-dev",
+        "pulid-flux",
+        "flux-redux"
+      ],
+      "error_node": "VAEDecode",
+      "error_message": "CUDA error: invalid argument\nSearch for `cudaErrorInvalidValue' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.\nCUDA kernel errors might be asynchronou",
+      "flags": [
+        "status-error"
+      ]
+    }
+  ],
+  "learning": {
+    "passed_count": 2,
+    "failed_count": 3,
+    "best_pastel_id": "9-2",
+    "best_pastel_pct": 85.33,
+    "r8_best": 82.43
+  }
+}