feat(video-pipeline): 添加可配置的转场策略引擎和TTS语速支持

- 引入转场策略系统（`getTransition`），支持 `fixed`、`director`、`rhythm` 三种模式 - 根据账号配置文件动态读取转场配置（`loadTransitions`） - 图片和视频轨道分别调用转场策略，替代原有的固定“闪白”转场 - 支持 `byPosition`（hook/body/keypoint/closing）和 `byDirector` 两种高级选择策略 - 图片动画支持 `loop_animation` 与 `in_animation` 解析（“缩放”、“弹入”等组合） - TTS 合成新增 `rate` 字段（源自账号配置 `ttsRate`），默认语速调整为 1.1 - 默认动画类型从 `kenburns-zoom` 改为 `缩放`，适配中文 CapCut
2026-05-01 14:16:08 +08:00
parent 21ff5311c1
commit f5d47ec5db
7 changed files with 118 additions and 15 deletions
--- a/.claude/skills/video-from-script/scripts/capcut_assemble.js
+++ b/.claude/skills/video-from-script/scripts/capcut_assemble.js
@@ -108,6 +108,44 @@ async function uploadToOSS(filePath) {
  return url
 }

+// ============================================================================
+// 转场选择策略
+// ============================================================================
+
+function getTransition(item, index, totalCount, transitionConfig) {
+  // 无配置 → 不加转场
+  if (!transitionConfig) return { name: '', duration: 0 }
+
+  const defaultT = transitionConfig.default || { name: '闪白', duration: 150000 }
+  const strategy = transitionConfig.strategy || 'fixed'
+
+  // 第一个素材不加转场
+  if (index === 0) return { name: '', duration: 0 }
+
+  switch (strategy) {
+    case 'director': {
+      // 按 directorRef 选择转场
+      const ref = (item.directorRef || '').toLowerCase()
+      const byDirector = transitionConfig.byDirector || {}
+      return byDirector[ref] || defaultT
+    }
+
+    case 'rhythm': {
+      // 按位置选择转场（hook / body / keypoint / closing）
+      const rules = transitionConfig.byPosition || {}
+      if (index === 1) return rules.hook || defaultT
+      if (index >= totalCount - 2) return rules.closing || defaultT
+      // 每隔3个 shot 用一个强调转场
+      if (index % 3 === 0) return rules.keypoint || defaultT
+      return rules.body || defaultT
+    }
+
+    case 'fixed':
+    default:
+      return defaultT
+  }
+}
+
 async function batchUploadToOSS(inputDir, files) {
  const urls = {}
  for (const file of files) {
@@ -179,7 +217,7 @@ async function assemble(args) {
    format = '9:16',
    apiKey = '',
    duration = '4',
-    animation = 'kenburns-zoom',
+    animation = '缩放',
    localAudio = 'true',
  } = args

@@ -284,7 +322,7 @@ async function assemble(args) {
  // -- 导入素材 --
  step++; console.log(`[${step}/${totalSteps}] 导入素材...`)
  if (mode === 'images') {
-    await addImages(draftUrl, items, imgUrls, timeline, width, height, animation)
+    await addImages(draftUrl, items, imgUrls, timeline, width, height, animation, transitionConfig)
  } else {
    // 视频模式：调速 → 上传 OSS → 添加到草稿
    // Step 1: ffmpeg 调速（在上传前，避免传两份）
@@ -330,7 +368,7 @@ async function assemble(args) {
        }
      }
    }
-    await addVideos(draftUrl, inputDir, items, timeline, width, height)
+    await addVideos(draftUrl, inputDir, items, timeline, width, height, transitionConfig)
  }

  // -- 添加 TTS 配音 --
@@ -355,6 +393,9 @@ async function assemble(args) {
    console.log(`  字幕风格: ${subtitleStyle.font || '默认'} ${subtitleStyle.inAnimation ? subtitleStyle.inAnimation + '→' + subtitleStyle.outAnimation : ''}`)
  }

+  // -- 读取转场策略 --
+  const transitionConfig = loadTransitions(manifest)
+
  // -- 添加字幕 --
  step++; console.log(`[${step}/${totalSteps}] 添加字幕...`)
  if (subtitles === 'true' && items.some(i => i.script || i.text)) {
@@ -418,23 +459,40 @@ async function assemble(args) {
 // 添加图片（自动上传到 OSS）
 // ============================================================================

-async function addImages(draftUrl, items, imgUrls, timeline, width, height, animation = '') {
+async function addImages(draftUrl, items, imgUrls, timeline, width, height, animation = '', transitionConfig = null) {
  const imageInfos = items.map((item, i) => {
    const url = imgUrls[item.file]
    if (!url) throw new Error(`图片 ${item.file} 未上传成功，无法添加`)
    const tl = timeline[i]
+    const t = getTransition(item, i, items.length, transitionConfig)

-    return {
+    // animation 解析：支持 "缩放" (group), "放大" (in), "渐显+缩小" (in+out)
+    const info = {
      image_url: url,
      width,
      height,
      start: tl.start,
      end: tl.end,
      duration: tl.duration,
-      animation: animation || '',
-      transition: i > 0 ? '闪白' : '',
-      transition_duration: 150000,
+      transition: t.name,
+      transition_duration: t.duration,
    }
+
+    if (animation) {
+      const parts = animation.split('+')
+      for (const part of parts) {
+        const name = part.trim()
+        // 组合动画（持续整段）：缩放、三分割 等
+        if (name === '缩放' || name === '缩放 II') {
+          info.loop_animation = name
+        } else {
+          // 默认作为入场动画
+          info.in_animation = name
+        }
+      }
+    }
+
+    return info
  })

  // 单次全量提交，所有图片在同一轨道
@@ -509,9 +567,10 @@ async function adjustVideoSpeed(videoPath, targetDurationSec) {
  })
 }

-async function addVideos(draftUrl, inputDir, items, timeline, width, height) {
+async function addVideos(draftUrl, inputDir, items, timeline, width, height, transitionConfig = null) {
  const videoInfos = items.map((item, i) => {
    const tl = timeline[i]
+    const t = getTransition(item, i, items.length, transitionConfig)
    return {
      video_url: item.videoUrl || (item.video ? path.resolve(inputDir, item.video) : null) || item.url || path.resolve(inputDir, item.file),
      width,
@@ -520,8 +579,8 @@ async function addVideos(draftUrl, inputDir, items, timeline, width, height) {
      end: tl.end,
      duration: tl.duration,
      mask: '',
-      transition: i > 0 ? '闪白' : '',
-      transition_duration: 150000,
+      transition: t.name,
+      transition_duration: t.duration,
      volume: item.volume || 1,
    }
  })
@@ -718,6 +777,18 @@ function loadSubtitleStyle(manifest) {
  } catch { return {} }
 }

+function loadTransitions(manifest) {
+  const account = manifest.account
+  if (!account) return null
+  const scriptDir = __dirname
+  const accountFile = path.join(scriptDir, '..', '..', '..', 'accounts', account, 'account.json')
+  if (!fs.existsSync(accountFile)) return null
+  try {
+    const accountData = JSON.parse(fs.readFileSync(accountFile, 'utf-8'))
+    return accountData.capcut?.transitions || null
+  } catch { return null }
+}
+
 // ============================================================================
 // 添加字幕（支持关键词高亮 + 账号字幕风格 + 分句切分）
 // ============================================================================
--- a/.claude/skills/video-from-script/scripts/lib/cmd-init.js
+++ b/.claude/skills/video-from-script/scripts/lib/cmd-init.js
@@ -93,6 +93,7 @@ function initManifest(options) {
    references,
    ...(accountConfig.ttsVoice ? { ttsVoice: accountConfig.ttsVoice } : {}),
    ...(accountConfig.ttsInstruction ? { ttsInstruction: accountConfig.ttsInstruction } : {}),
+    ...(accountConfig.ttsRate ? { ttsRate: accountConfig.ttsRate } : {}),
    items,
  }

--- a/.claude/skills/video-from-script/scripts/lib/phase-assemble.js
+++ b/.claude/skills/video-from-script/scripts/lib/phase-assemble.js
@@ -23,7 +23,7 @@ async function phaseAssemble(manifest, manifestPath, options) {
    subtitles: mode === 'images' ? 'true' : 'false',
    voiceover: manifest.items.some(it => it.audio) ? 'true' : 'false',
    duration: '4',
-    animation: capcutConfig.animation || 'kenburns-zoom',
+    animation: capcutConfig.animation || '缩放',
  }

  if (capcutConfig.defaultBGM) assembleArgs.bgm = capcutConfig.defaultBGM
--- a/.claude/skills/video-from-script/scripts/lib/phase-tts.js
+++ b/.claude/skills/video-from-script/scripts/lib/phase-tts.js
@@ -30,6 +30,7 @@ async function phaseTts(manifest, manifestPath, options = {}) {
        id: item.id || idx,
        voice: manifest.ttsVoice || undefined,
        instruction: manifest.ttsInstruction || undefined,
+        rate: manifest.ttsRate || undefined,
      })
      item.audio = path.relative(dir, filePath).replace(/\\/g, '/')
      item.audioDuration = Math.round(duration * 1000) / 1000
--- a/.claude/skills/video-from-script/scripts/qwen-tts.js
+++ b/.claude/skills/video-from-script/scripts/qwen-tts.js
@@ -108,7 +108,7 @@ function synthesize(text, options = {}) {
            format: 'mp3',
            sample_rate: 24000,
            volume: 50,
-            rate: 1.0,
+            rate: options.rate || 1.1,
            pitch_rate: 1.0,
            text_type: 'PlainText',
            ...(instruction ? { instruction } : {}),