Add MiniMax and PinkCherry LTX as pickable video engines, with still or text start.

Shot 1 can run text-to-video; later shots still extend from the last frame. Recommend output is also normalized into shot-script cards.

Co-authored-by: Cursor <cursoragent@cursor.com>

utils/parseRecommend.ts

@@ -1,3 +1,5 @@
+import { parseShotScript } from './parseShots'
+
 export function parseEditPasses(text: string): string[] | null {
   const src = String(text || '').replace(/\r\n/g, '\n')
   const marker = /^[ \t]*---pass\s+(\d+)---[ \t]*$/gim
@@ -21,3 +23,40 @@ export function parseEditPasses(text: string): string[] | null {
 export function looksLikeShotScript(text: string) {
   return /^[ \t]*shot\s+[2-9]\d*\b/im.test(String(text || ''))
 }
+
+function stripFences(text: string) {
+  return String(text || '')
+    .replace(/^\s*```(?:[\w-]+)?\s*\n?/, '')
+    .replace(/\n?```\s*$/, '')
+    .trim()
+}
+
+function splitInlineShotMarkers(text: string) {
+  return String(text || '')
+    .replace(/\r\n/g, '\n')
+    .replace(/(^|[^A-Za-z])shot\s+(\d+)\s*[:.)\-]?\s*/gi, '$1\n\nshot $2\n')
+    .replace(/\n{3,}/g, '\n\n')
+    .trim()
+}
+
+function formatShotBody(body: string) {
+  let text = String(body || '').trim()
+  text = text.replace(/^\s*[:.\-–)]\s*/, '')
+  text = text.replace(/\s*\[SHOT CONFIGURATION\]\s*/gi, '\n\n[SHOT CONFIGURATION]\n')
+  text = text.replace(/\s*\[SUBJECT DIRECTION(?:\s*&\s*ACTION)?\]\s*/gi, '\n\n[SUBJECT DIRECTION & ACTION]\n')
+  text = text.replace(/\s*\[AUDIO CUES\]\s*/gi, '\n\n[AUDIO CUES]\n')
+  text = text.replace(/[ \t]+\n/g, '\n').replace(/\n{3,}/g, '\n\n').trim()
+  return text
+}
+
+export function normalizeShotScript(text: string) {
+  const src = splitInlineShotMarkers(stripFences(text))
+  const shots = parseShotScript(src)
+  if (!shots.length) return src
+  if (shots.length === 1 && !/^[ \t]*shot\s+\d+\b/im.test(src)) {
+    return formatShotBody(shots[0].prompt)
+  }
+  return shots
+    .map(shot => `shot ${shot.n}\n${formatShotBody(shot.prompt)}`.trim())
+    .join('\n\n')
+}

utils/videoModels.ts

@@ -0,0 +1,56 @@
+export type VideoEngineId = 'minimax' | 'ltx'
+export type VideoStartId = 'still' | 'text'
+export type MiniMaxGraphId = 'v1' | 'v2'
+export type VideoWorkflowId = 'v1' | 'v2' | 't2v' | 'ltx' | 'ltx-t2v'
+
+export function parseVideoWorkflow(raw: unknown): VideoWorkflowId {
+  const value = String(raw || '').trim()
+  if (value === 'v2') return 'v2'
+  if (value === 't2v') return 't2v'
+  if (value === 'ltx' || value === 'ltx-i2v') return 'ltx'
+  if (value === 'ltx-t2v') return 'ltx-t2v'
+  return 'v1'
+}
+
+export function videoEngineOf(id: VideoWorkflowId): VideoEngineId {
+  return id.startsWith('ltx') ? 'ltx' : 'minimax'
+}
+
+export function videoStartOf(id: VideoWorkflowId): VideoStartId {
+  return id === 't2v' || id === 'ltx-t2v' ? 'text' : 'still'
+}
+
+export function isLtxWorkflow(id: VideoWorkflowId) {
+  return videoEngineOf(id) === 'ltx'
+}
+
+export function isTextToVideo(id: VideoWorkflowId) {
+  return videoStartOf(id) === 'text'
+}
+
+export function minimaxGraphOf(id: VideoWorkflowId): MiniMaxGraphId {
+  return id === 'v2' ? 'v2' : 'v1'
+}
+
+export function workflowForExtension(id: VideoWorkflowId): VideoWorkflowId {
+  if (id === 't2v') return 'v1'
+  if (id === 'ltx-t2v') return 'ltx'
+  return id
+}
+
+export function composeVideoWorkflow(engine: VideoEngineId, start: VideoStartId, graph: MiniMaxGraphId = 'v1'): VideoWorkflowId {
+  if (engine === 'ltx') return start === 'text' ? 'ltx-t2v' : 'ltx'
+  if (start === 'text') return 't2v'
+  return graph === 'v2' ? 'v2' : 'v1'
+}
+
+export function defaultVideoSteps(turbo: boolean, workflow: string = 'v1') {
+  const id = parseVideoWorkflow(workflow)
+  if (!turbo) return isLtxWorkflow(id) ? 12 : 20
+  if (isLtxWorkflow(id)) return 8
+  return id === 'v2' ? 6 : 8
+}
+
+export const LTX_UNET = 'PinkCherry_FineTune_int8_v1_8_LTX23.safetensors'
+export const LTX_DISTILLED_LORA = 'ltx-2.3-22b-distilled-lora-384-1.1.safetensors'
+export const LTX_NEGATIVE = 'blurry, low quality, jitter, deformed, static frame, freeze, watermark, bad anatomy, text, subtitles'

utils/videoSteps.ts

@@ -1,4 +0,0 @@
-export function defaultVideoSteps(turbo: boolean, workflow = 'v1') {
-  if (!turbo) return 20
-  return workflow === 'v2' ? 6 : 8
-}