fix: 优化架构

engine/adapters/video/__init__.py

@@ -0,0 +1 @@
+

engine/adapters/video/base.py

@@ -0,0 +1,9 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+
+class BaseVideoGen:
+    def generate(self, image_path: str, prompt: dict, output_path: str | Path) -> str:
+        raise NotImplementedError
+

engine/adapters/video/ltx_adapter.py

@@ -0,0 +1,18 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from engine.config import AppConfig
+
+from .base import BaseVideoGen
+
+
+class LTXVideoGen(BaseVideoGen):
+    def __init__(self, cfg: AppConfig):
+        self.cfg = cfg
+
+    def generate(self, image_path: str, prompt: dict, output_path: str | Path) -> str:
+        # Reserved for future: direct image->video generation (LTX / diffusion video).
+        # Current project keeps clip generation via MoviePy for stability.
+        raise NotImplementedError("LTXVideoGen is not implemented yet")
+

engine/adapters/video/moviepy_adapter.py

@@ -0,0 +1,81 @@
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+from moviepy import AudioFileClip, VideoClip
+from PIL import Image
+
+from engine.config import AppConfig
+
+from .base import BaseVideoGen
+
+
+class MoviePyVideoGen(BaseVideoGen):
+    def __init__(self, cfg: AppConfig):
+        self.cfg = cfg
+
+    def generate(self, image_path: str, prompt: dict, output_path: str | Path) -> str:
+        output_path = Path(output_path)
+        output_path.parent.mkdir(parents=True, exist_ok=True)
+
+        # Required prompt fields for shot rendering.
+        duration_s = float(prompt.get("duration_s", 3))
+        fps = int(prompt.get("fps", self.cfg.get("video.mock_fps", 24)))
+        audio_path = prompt.get("audio_path")
+
+        # Clip resolution.
+        size = prompt.get("size")
+        if isinstance(size, (list, tuple)) and len(size) == 2:
+            w, h = int(size[0]), int(size[1])
+        else:
+            mock_size = self.cfg.get("video.mock_size", [1024, 576])
+            w, h = int(mock_size[0]), int(mock_size[1])
+
+        base_img = Image.open(image_path).convert("RGB")
+
+        def make_frame(t: float):
+            progress = float(t) / max(duration_s, 1e-6)
+            progress = max(0.0, min(1.0, progress))
+            scale = 1.0 + 0.03 * progress
+            new_w = max(w, int(w * scale))
+            new_h = max(h, int(h * scale))
+            frame = base_img.resize((new_w, new_h), Image.LANCZOS)
+            left = (new_w - w) // 2
+            top = (new_h - h) // 2
+            frame = frame.crop((left, top, left + w, top + h))
+            return np.array(frame)
+
+        video = VideoClip(make_frame, duration=duration_s, has_constant_size=True)
+
+        # Optional audio.
+        if audio_path and os.path.exists(str(audio_path)):
+            a = AudioFileClip(str(audio_path))
+            video = video.with_audio(a)
+        else:
+            a = None
+
+        try:
+            video.write_videofile(
+                str(output_path),
+                fps=fps,
+                codec="libx264",
+                audio_codec="aac",
+                preset="veryfast",
+                threads=2,
+            )
+        finally:
+            try:
+                video.close()
+            except Exception:
+                pass
+            if a is not None:
+                try:
+                    a.close()
+                except Exception:
+                    pass
+
+        return str(output_path)
+