docs: mark Phase 2 modernization items as DONE

Author: SysAdminDoc

MODERNIZATION.md

@@ -112,9 +112,9 @@ All pure Python, no external dependencies. Includes: zoom.py, auto_zoom.py (cv2)
 |---|-------|--------|--------|--------|
 | 1 | auto-editor v30 rewritten in Nim; pip package frozen at v29.3.1 | `auto_edit.py` | Detect native Nim binary, pip fallback | TODO |
 | 2 | Anthropic API version stale (`2023-06-01`) | `llm.py` | Update version string, add latest model IDs | DONE (2026-04-06) — Added Gemini provider, updated default model to llama3.2, OpenAI max_completion_tokens |
-| 3 | CTranslate2 in maintenance mode | `captions_enhanced.py` | Make SeamlessM4T primary, NLLB fallback | TODO |
+| 3 | CTranslate2 in maintenance mode | `captions_enhanced.py` | Make SeamlessM4T primary, NLLB fallback | DONE (2026-04-06) — translate() and translate_segments_auto() with SeamlessM4T-first, NLLB fallback. Route default "auto" |
 | 4 | FFmpeg minterpolate slow + artifacting | `video_ai.py` | Add RIFE neural interpolation as `method="rife"` | TODO |
-| 5 | B-roll gen backends outdated (SVD superseded) | `broll_generate.py` | Add Wan 2.1 as primary, remove SVD | TODO |
+| 5 | B-roll gen backends outdated (SVD superseded) | `broll_generate.py` | Add Wan 2.1 as primary, remove SVD | DONE (2026-04-06) — Wan 2.1 added as primary (1.3B/14B models), auto-selected when available |
 
 ### TIER 2 -- Significant Improvements Available
 
@@ -123,9 +123,9 @@ All pure Python, no external dependencies. Includes: zoom.py, auto_zoom.py (cv2)
 | 6 | Silero VAD v6 (ONNX support) | `silence.py` | Upgrade torch.hub model, add ONNX path | DONE (2026-04-06) — VAD now at v6.2.1, ONNX inference when onnxruntime available |
 | 7 | Kokoro v1.0 ONNX model available | `voice_gen.py` | Verify version, test ONNX inference | DONE (2026-04-06) — Version pin updated to >=0.9 |
 | 8 | DeepFilterNet3 improvements | `audio_pro.py` | Verify v0.5.x installed | DONE (2026-04-06) — Verified current, imports correct |
-| 9 | CodeFormer > GFPGAN for degraded faces | `face_swap.py` | Wire CodeFormer as default enhancer | TODO |
+| 9 | CodeFormer > GFPGAN for degraded faces | `face_swap.py` | Wire CodeFormer as default enhancer | DONE (pre-existing) — Already fully implemented with model="codeformer" + fidelity slider |
 | 10 | Depth Anything checkpoint verification | `depth_effects.py` | Verify latest HuggingFace checkpoints | DONE (2026-04-06) — Docstring corrected to current HF org |
-| 11 | Two-stage scene detection pipeline | `scene_detect.py` | Add `method="hybrid"` (PySceneDetect + TransNetV2) | TODO |
+| 11 | Two-stage scene detection pipeline | `scene_detect.py` | Add `method="hybrid"` (PySceneDetect + TransNetV2) | DONE (2026-04-06) — detect_scenes_hybrid() with 0.5s dedup, TransNetV2 fallback |
 | 12 | .t7 style transfer models are 2017-era | `style_transfer.py` | Research AesPA-Net / InST for temporal consistency | TODO |
 | 13 | Stable Audio Open vs MusicGen | `music_ai.py` | Add as backend option | TODO |
 | 14 | AV1 export preset (40% smaller files) | `export_presets.py` | Add SVT-AV1 + NVENC AV1 presets | DONE (2026-04-06) — Added av1_1080p, av1_4k, hevc_1080p presets + preset int bug fix |