Send V2 identity prompts to Reference to Video instead of Image to Video.

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

server/utils/comfy.ts

@@ -163,12 +163,14 @@ export function extractPromptFromHistory(entry: unknown) {
   const prompt = (entry as { prompt?: unknown[] })?.prompt
   const graph = Array.isArray(prompt) ? prompt[2] : null
   if (!graph || typeof graph !== 'object') return ''
+  let imageToVideo = ''
   for (const node of Object.values(graph as Record<string, { class_type?: string; inputs?: Record<string, unknown> }>)) {
-    if (node?.class_type === 'MiniMaxH3ImageToVideo') {
-      return String(node.inputs?.prompt || node.inputs?.text || '').trim()
-    }
+    const text = String(node.inputs?.prompt || node.inputs?.text || '').trim()
+    if (!text) continue
+    if (node?.class_type === 'MiniMaxH3ReferenceToVideo') return text
+    if (node?.class_type === 'MiniMaxH3ImageToVideo') imageToVideo = text
   }
-  return ''
+  return imageToVideo
 }
 
 export function extractClipMetaFromHistory(entry: unknown) {

server/utils/workflow.ts

@@ -73,7 +73,8 @@ export function buildWorkflow(params: GenerateParams) {
   const refs = Array.from({ length: 4 }, (_, index) => String(params.referenceImageNames?.[index] || '').trim())
   const extraPictures = refs.flatMap((name, index) => name ? [index + 2] : [])
   const useRefs = version === 'v2' && params.useIdentityRefs === true
-  const graphPrompt = useRefs ? buildIdentityPrompt(params.prompt, extraPictures) : params.prompt
+  const actionPrompt = params.prompt
+  const graphPrompt = useRefs ? buildIdentityPrompt(actionPrompt, extraPictures) : actionPrompt
 
   for (const [id, node] of Object.entries(graph)) {
     if (IMAGE_CLASSES.has(node.class_type) && 'image' in node.inputs && id === '114') {
@@ -81,19 +82,20 @@ export function buildWorkflow(params: GenerateParams) {
     }
 
     if (PROMPT_CLASSES.has(node.class_type) || node.class_type.includes('Qwen3')) {
-      if ('prompt' in node.inputs) node.inputs.prompt = graphPrompt
-      if ('text' in node.inputs) node.inputs.text = graphPrompt
+      const nodePrompt = promptForNode(node.class_type, useRefs, actionPrompt, graphPrompt)
+      if ('prompt' in node.inputs) node.inputs.prompt = nodePrompt
+      if ('text' in node.inputs) node.inputs.text = nodePrompt
     }
 
     if (node.class_type === 'MiniMaxH3ImageToVideo') {
-      node.inputs.prompt = graphPrompt
+      node.inputs.prompt = useRefs ? '' : actionPrompt
       if (typeof node.inputs.width === 'number') node.inputs.width = width
       if (typeof node.inputs.height === 'number') node.inputs.height = height
       if (typeof node.inputs.length === 'number') node.inputs.length = params.length
     }
 
     if (node.class_type === 'MiniMaxH3ReferenceToVideo') {
-      node.inputs.prompt = graphPrompt
+      node.inputs.prompt = useRefs ? graphPrompt : actionPrompt
       if (!Array.isArray(node.inputs.length)) node.inputs.length = ['174', 1]
     }
 
@@ -206,6 +208,12 @@ export function buildWorkflow(params: GenerateParams) {
   return graph
 }
 
+function promptForNode(classType: string, useRefs: boolean, actionPrompt: string, graphPrompt: string) {
+  if (classType === 'MiniMaxH3ImageToVideo') return useRefs ? '' : actionPrompt
+  if (classType === 'MiniMaxH3ReferenceToVideo') return useRefs ? graphPrompt : actionPrompt
+  return graphPrompt
+}
+
 function applyV2IdentityPath(
   graph: WorkflowGraph,
   params: GenerateParams,