Ship MiniMax v2 identity refs and Qwen sidecar edits.

Keep v1 as the default video graph, relay image edits to the Qwen Lightning box, and keep /api/health up if one Comfy host is unset.

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

pages/index.vue

@@ -6,7 +6,7 @@
           <div class="flex h-10 w-10 items-center justify-center rounded-xl bg-amber-400 text-zinc-950 font-display font-extrabold">A</div>
           <div>
             <h1 class="font-display text-xl font-bold leading-none">{{ instanceName }}</h1>
-            <p class="text-[11px] uppercase tracking-[0.22em] text-zinc-500">MiniMax H3 · ComfyUI relay</p>
+            <p class="text-[11px] uppercase tracking-[0.22em] text-zinc-500">MiniMax H3 · Qwen Edit · ComfyUI</p>
           </div>
         </div>
         <div class="flex items-center gap-3 text-sm">
@@ -17,6 +17,14 @@
             <span class="h-2 w-2 rounded-full" :class="comfyBadge.dot" />
             {{ comfyBadge.label }}
           </span>
+          <span
+            class="hidden lg:inline-flex items-center gap-2 rounded-full border px-3 py-1"
+            :class="imageComfyBadge.className"
+            :title="imageComfyHost || 'Image edit sidecar'"
+          >
+            <span class="h-2 w-2 rounded-full" :class="imageComfyBadge.dot" />
+            {{ imageComfyBadge.label }}
+          </span>
           <span class="hidden md:block text-zinc-400">{{ userName }}</span>
           <button type="button" class="rounded-full border border-white/10 px-3 py-1 text-zinc-300 hover:border-amber-300/50 hover:text-white" @click="settingsOpen = true">
             Settings
@@ -29,8 +37,28 @@
     <main class="mx-auto grid max-w-7xl gap-6 px-4 py-6 sm:px-6 lg:grid-cols-[minmax(0,1fr)_minmax(0,1fr)]">
       <section class="glass rounded-3xl p-5 sm:p-6 space-y-5">
         <div>
+          <div class="mb-3 flex flex-wrap gap-2">
+            <button
+              type="button"
+              class="rounded-full border px-3 py-1 text-xs font-medium"
+              :class="studioMode === 'video' ? 'border-amber-300/70 bg-amber-400/10 text-amber-100' : 'border-white/10 text-zinc-400 hover:text-white'"
+              @click="studioMode = 'video'"
+            >
+              Video
+            </button>
+            <button
+              type="button"
+              class="rounded-full border px-3 py-1 text-xs font-medium"
+              :class="studioMode === 'edit' ? 'border-amber-300/70 bg-amber-400/10 text-amber-100' : 'border-white/10 text-zinc-400 hover:text-white'"
+              @click="studioMode = 'edit'"
+            >
+              Image edit
+            </button>
+          </div>
           <h2 class="font-display text-2xl font-bold">Input</h2>
-          <p class="text-sm text-zinc-400">Drop a still, describe the motion, then relay the job to the local desktop.</p>
+          <p class="text-sm text-zinc-400">{{ studioMode === 'edit'
+            ? 'Drop a still and describe the edit. This relays to the sidecar Qwen Comfy box once the workflow is mapped.'
+            : 'Drop a still, describe the motion, then relay the job to the local desktop.' }}</p>
         </div>
 
         <div
@@ -39,7 +67,7 @@
           @dragover.prevent="dragging = true"
           @dragleave.prevent="dragging = false"
           @drop.prevent="onDrop"
-          @click="pickerOpen = true"
+          @click="openPicker('main')"
         >
           <div v-if="preview && !hideInputPreview" class="relative">
             <img :src="preview" alt="Source still" class="max-h-72 w-full rounded-xl object-contain bg-black/40">
@@ -59,8 +87,8 @@
 
         <div>
           <div class="mb-2 flex flex-wrap items-center justify-between gap-2">
-            <label class="text-sm font-medium text-zinc-300">Motion & scene prompt</label>
-            <label class="flex cursor-pointer items-center gap-2 text-xs text-zinc-400">
+            <label class="text-sm font-medium text-zinc-300">{{ studioMode === 'edit' ? 'Edit prompt' : 'Motion & scene prompt' }}</label>
+            <label v-if="studioMode === 'video'" class="flex cursor-pointer items-center gap-2 text-xs text-zinc-400">
               <span class="relative inline-flex h-5 w-9 shrink-0 items-center">
                 <input v-model="shotScriptMode" type="checkbox" class="peer sr-only" role="switch">
                 <span class="absolute inset-0 rounded-full bg-zinc-700 transition peer-checked:bg-amber-400" />
@@ -69,18 +97,20 @@
               Shot script
             </label>
           </div>
-          <p v-if="shotScriptMode" class="mb-2 text-xs text-zinc-500">
+          <p v-if="shotScriptMode && studioMode === 'video'" class="mb-2 text-xs text-zinc-500">
             Paste a block. Text before the first marker is shot 1. Write <span class="font-medium text-zinc-300">shot 2</span>, <span class="font-medium text-zinc-300">shot 3</span>, and so on to queue extensions. Every shot uses the duration slider.
           </p>
           <textarea
             v-model="prompt"
             :rows="shotScriptMode ? 12 : 6"
             class="w-full rounded-2xl border border-white/10 bg-zinc-950/70 px-4 py-3 text-sm outline-none ring-amber-300/40 placeholder:text-zinc-600 focus:ring-2"
-            :placeholder="shotScriptMode
+            :placeholder="studioMode === 'edit'
+              ? 'Make the lighting warmer. Replace the background with a rainy street at night. Keep the face unchanged.'
+              : (shotScriptMode
               ? 'A woman walks into the village.\n\nshot 2\nShe crosses the stone bridge at dusk.\n\nshot 3\nShe opens the shop door and steps inside.'
-              : 'Camera slowly pushes in. Soft daylight, fabric ripples, dust motes in the air. Audio: quiet room tone, distant birds.'"
+              : 'Camera slowly pushes in. Soft daylight, fabric ripples, dust motes in the air. Audio: quiet room tone, distant birds.')"
           />
-          <div v-if="shotScriptMode" class="mt-3 space-y-2">
+          <div v-if="shotScriptMode && studioMode === 'video'" class="mt-3 space-y-2">
             <p v-if="!parsedShots.length" class="text-xs text-zinc-500">Add a first shot, then <span class="text-zinc-300">shot 2</span> for an extension.</p>
             <div
               v-for="(shot, index) in parsedShots"
@@ -96,7 +126,7 @@
           </div>
         </div>
 
-        <div v-if="!shotScriptMode" class="space-y-3">
+        <div v-if="studioMode === 'video' && !shotScriptMode" class="space-y-3">
           <div
             v-for="(item, index) in extensionQueue"
             :key="item.id"
@@ -171,7 +201,7 @@
                 <span class="absolute left-0.5 top-0.5 h-5 w-5 rounded-full bg-white transition peer-checked:translate-x-5" />
               </span>
             </label>
-            <label class="flex cursor-pointer items-center justify-between gap-3 rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3 text-sm">
+            <label v-if="studioMode === 'video'" class="flex cursor-pointer items-center justify-between gap-3 rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3 text-sm">
               <span>
                 <span class="block font-medium text-zinc-200">Generate sound</span>
                 <span class="mt-0.5 block text-xs text-zinc-500">Off skips the audio VAE so Comfy encodes a silent clip.</span>
@@ -182,7 +212,7 @@
                 <span class="absolute left-0.5 top-0.5 h-5 w-5 rounded-full bg-white transition peer-checked:translate-x-5" />
               </span>
             </label>
-            <label class="flex cursor-pointer items-center justify-between gap-3 rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3 text-sm">
+            <label v-if="studioMode === 'video'" class="flex cursor-pointer items-center justify-between gap-3 rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3 text-sm">
               <span>
                 <span class="block font-medium text-zinc-200">Autoplay</span>
                 <span class="mt-0.5 block text-xs text-zinc-500">Off leaves finished clips paused until you press play.</span>
@@ -197,7 +227,79 @@
           </div>
         </div>
 
-        <details class="rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3">
+        <div v-if="studioMode === 'video'" class="space-y-3">
+          <div>
+            <p class="mb-2 text-xs uppercase tracking-wider text-zinc-500">MiniMax graph</p>
+            <div class="grid grid-cols-2 gap-2">
+              <button
+                type="button"
+                class="rounded-xl border px-3 py-2 text-left text-sm"
+                :class="videoWorkflow === 'v1' ? 'border-amber-300 bg-amber-400/10 text-amber-100' : 'border-white/10 hover:border-white/20'"
+                @click="videoWorkflow = 'v1'"
+              >
+                <span class="block font-semibold">v1 · validated</span>
+                <span class="text-xs text-zinc-400">First-frame image to video</span>
+              </button>
+              <button
+                type="button"
+                class="rounded-xl border px-3 py-2 text-left text-sm"
+                :class="videoWorkflow === 'v2' ? 'border-amber-300 bg-amber-400/10 text-amber-100' : 'border-white/10 hover:border-white/20'"
+                @click="videoWorkflow = 'v2'"
+              >
+                <span class="block font-semibold">v2 · identity refs</span>
+                <span class="text-xs text-zinc-400">Optional lock with up to 4 stills</span>
+              </button>
+            </div>
+          </div>
+
+          <div v-if="videoWorkflow === 'v2'" class="rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3 space-y-3">
+            <label class="flex cursor-pointer items-center justify-between gap-3 text-sm">
+              <span>
+                <span class="block font-medium text-zinc-200">Use identity references</span>
+                <span class="mt-0.5 block text-xs text-zinc-500">Off uses first-frame only. On sends extra stills into Ref2VA.</span>
+              </span>
+              <span class="relative inline-flex h-6 w-11 shrink-0 items-center">
+                <input v-model="useIdentityRefs" type="checkbox" class="peer sr-only" role="switch">
+                <span class="absolute inset-0 rounded-full bg-zinc-700 transition peer-checked:bg-amber-400" />
+                <span class="absolute left-0.5 top-0.5 h-5 w-5 rounded-full bg-white transition peer-checked:translate-x-5" />
+              </span>
+            </label>
+            <div v-if="useIdentityRefs" class="grid grid-cols-2 gap-2 sm:grid-cols-4">
+              <div
+                v-for="slot in 4"
+                :key="slot"
+                class="relative min-h-24 overflow-hidden rounded-xl border border-dashed border-white/15 bg-zinc-950/60"
+              >
+                <button
+                  type="button"
+                  class="block w-full text-left"
+                  @click="openPicker(slot - 1)"
+                >
+                  <img
+                    v-if="identityPreviews[slot - 1]"
+                    :src="identityPreviews[slot - 1]"
+                    alt=""
+                    class="h-24 w-full object-cover"
+                  >
+                  <span v-else class="flex h-24 items-center justify-center px-2 text-center text-[11px] text-zinc-500">
+                    Ref {{ slot }}
+                  </span>
+                </button>
+                <button
+                  v-if="identityRefs[slot - 1]"
+                  type="button"
+                  class="absolute right-1 top-1 rounded-full bg-zinc-950/80 px-2 py-0.5 text-[10px] text-zinc-300"
+                  @click="clearIdentityRef(slot - 1)"
+                >
+                  Remove
+                </button>
+              </div>
+            </div>
+            <p v-if="useIdentityRefs" class="text-xs text-zinc-500">Extra stills lock people and objects. The first-frame still above stays the opening frame. Identity refs are not reused on queued extensions.</p>
+          </div>
+        </div>
+
+        <details v-if="studioMode === 'video'" class="rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3">
           <summary class="cursor-pointer text-sm font-medium">Generation settings</summary>
           <div class="mt-4 space-y-4">
             <div>
@@ -296,14 +398,47 @@
           </div>
         </details>
 
+        <details v-else class="rounded-2xl border border-white/10 bg-zinc-950/40 px-4 py-3">
+          <summary class="cursor-pointer text-sm font-medium">Image edit settings</summary>
+          <div class="mt-4 space-y-4">
+            <p class="text-xs text-zinc-500">
+              Sidecar: <span class="text-zinc-200">{{ imageComfyHost || 'not configured' }}</span>
+              · Qwen Image Edit Lightning (4-step LoRA)
+            </p>
+            <label class="block text-sm">
+              <span class="mb-1 block text-zinc-400">Negative prompt</span>
+              <textarea v-model="editNegative" rows="2" class="w-full rounded-xl border border-white/10 bg-zinc-950 px-3 py-2 text-sm" />
+            </label>
+            <div class="grid gap-4 sm:grid-cols-2">
+              <label class="block text-sm">
+                <span class="mb-1 block text-zinc-400">Steps · {{ editSteps }}</span>
+                <input v-model.number="editSteps" type="range" min="1" max="8" step="1" class="w-full">
+              </label>
+              <label class="block text-sm">
+                <span class="mb-1 block text-zinc-400">CFG · {{ editCfg }}</span>
+                <input v-model.number="editCfg" type="range" min="0.5" max="4" step="0.1" class="w-full">
+              </label>
+              <label class="block text-sm">
+                <span class="mb-1 block text-zinc-400">Denoise · {{ editDenoise }}</span>
+                <input v-model.number="editDenoise" type="range" min="0.05" max="1" step="0.05" class="w-full">
+              </label>
+              <label class="block text-sm">
+                <span class="mb-1 block text-zinc-400">Seed</span>
+                <input v-model="seedInput" class="w-full rounded-xl border border-white/10 bg-zinc-950 px-3 py-2 text-sm" placeholder="random">
+              </label>
+            </div>
+            <p class="text-xs text-zinc-500">Lower denoise keeps more of the original still. Lightning is tuned for 4 steps, CFG 1, denoise around 0.75.</p>
+          </div>
+        </details>
+
         <div class="flex flex-wrap gap-3">
           <button
             type="button"
             class="flex-1 rounded-2xl bg-amber-400 px-5 py-3 font-semibold text-zinc-950 transition hover:bg-amber-300 disabled:cursor-not-allowed disabled:opacity-50"
-            :disabled="busy || !file || !prompt.trim() || !folderId || !queueReady"
-            @click="generate"
+            :disabled="busy || !file || !prompt.trim() || !folderId || (studioMode !== 'edit' && !queueReady)"
+            @click="studioMode === 'edit' ? editImage() : generate()"
           >
-            {{ generateLabel }}
+            {{ studioMode === 'edit' ? (busy ? 'Editing…' : 'Edit image') : generateLabel }}
           </button>
           <button
             v-if="busy"
@@ -375,8 +510,15 @@
             :class="{ 'sensitive-glow': awaitingReveal || lockedSave }"
             :style="{ aspectRatio: `${playerSize.width} / ${playerSize.height}` }"
           >
+            <img
+              v-if="editResultUrl && !lockedSave"
+              :src="editResultUrl"
+              alt="Edited still"
+              class="h-full w-full bg-black object-contain"
+              :class="{ 'opacity-0': awaitingReveal }"
+            >
             <video
-              v-if="videoUrl && !lockedSave"
+              v-else-if="videoUrl && !lockedSave"
               ref="player"
               :src="videoUrl"
               class="h-full w-full bg-black"
@@ -404,12 +546,28 @@
               <span class="text-sm text-amber-100/80">Unlock that folder in the library to view this clip</span>
             </div>
           <div
-            v-else-if="!videoUrl"
+            v-else-if="!videoUrl && !editResultUrl"
             class="flex h-full min-h-[12rem] items-center justify-center text-sm text-zinc-500"
           >
-            Generated MP4 appears here
+            {{ studioMode === 'edit' ? 'Edited still appears here' : 'Generated MP4 appears here' }}
           </div>
           </div>
+          <div v-if="editResultUrl && !lockedSave && !awaitingReveal" class="mt-4 flex flex-wrap gap-3">
+            <a
+              :href="editResultUrl"
+              :download="`${outputDownloadName || 'aigen-edit'}.png`"
+              class="inline-flex rounded-2xl border border-white/15 px-4 py-2 text-sm hover:border-amber-300/50"
+            >
+              Download still
+            </a>
+            <button
+              type="button"
+              class="inline-flex rounded-2xl border border-amber-300/40 px-4 py-2 text-sm text-amber-100 hover:border-amber-300/70"
+              @click="useEditAsInput"
+            >
+              Use as input still
+            </button>
+          </div>
           <div v-if="videoUrl && !lockedSave" class="mt-4 flex flex-wrap gap-3">
             <a
               v-if="!awaitingReveal"
@@ -646,6 +804,7 @@
       :stills="stills"
       :folder-id="folderId"
       :hide-input-preview="hideInputPreview"
+      :heading="pickerHeading"
       @close="pickerOpen = false"
       @updated="applyLibrary"
       @error="toast"
@@ -712,6 +871,7 @@ interface LibraryClip {
   familyId?: string
   parentClipId?: string
   chainIndex?: number
+  workflow?: 'v1' | 'v2'
 }
 
 interface RetryDraft {
@@ -737,6 +897,7 @@ interface RetryDraft {
   samplerName?: string
   scheduler?: string
   extensions?: { prompt: string; duration: number }[]
+  workflow?: 'v1' | 'v2'
 }
 
 interface QueuedExtension {
@@ -790,6 +951,11 @@ const hideInputPreview = ref(false)
 const autoplayEnabled = ref(false)
 const JOB_STORE = 'aigen-active-job'
 const busy = ref(false)
+const studioMode = ref('video')
+const imageComfyOk = ref(false)
+const imageComfyBusy = ref(false)
+const imageComfyConfigured = ref(false)
+const imageComfyHost = ref('')
 const progress = ref(0)
 const statusMessage = ref('Waiting for a still.')
 const statusBusy = ref(false)
@@ -804,6 +970,11 @@ const activeChainPlan = ref<{ label: string; prompt: string; duration: number }[
 const expandedFamilies = ref<Record<string, boolean>>({})
 const elapsedMs = ref(0)
 const videoUrl = ref('')
+const editResultUrl = ref('')
+const editSteps = ref(4)
+const editCfg = ref(1)
+const editDenoise = ref(0.75)
+const editNegative = ref('blurry, low quality, deformed, bad anatomy, watermark, text, extra fingers')
 const jobId = ref('')
 const currentClipId = ref('')
 const extendOpen = ref(false)
@@ -838,6 +1009,15 @@ const renameInput = ref<HTMLInputElement | null>(null)
 const unlockPassword = ref('')
 const settingsOpen = ref(false)
 const pickerOpen = ref(false)
+const pickerSlot = ref<'main' | number>('main')
+const videoWorkflow = ref('v1')
+const useIdentityRefs = ref(false)
+const identityRefs = ref<(File | null)[]>([null, null, null, null])
+const identityPreviews = ref(['', '', '', ''])
+const pickerHeading = computed(() => {
+  if (pickerSlot.value === 'main') return 'Choose an image'
+  return `Choose identity ref ${Number(pickerSlot.value) + 1}`
+})
 let timer: ReturnType<typeof setInterval> | null = null
 let source: EventSource | null = null
 let pollTimer: ReturnType<typeof setInterval> | null = null
@@ -875,6 +1055,35 @@ const comfyBadge = computed(() => {
   }
 })
 
+const imageComfyBadge = computed(() => {
+  if (!imageComfyConfigured.value) {
+    return {
+      label: 'Edit sidecar unset',
+      className: 'border-white/10 text-zinc-500',
+      dot: 'bg-zinc-500'
+    }
+  }
+  if (!imageComfyOk.value) {
+    return {
+      label: 'Edit Comfy offline',
+      className: 'border-red-400/30 text-red-300',
+      dot: 'bg-red-400'
+    }
+  }
+  if (imageComfyBusy.value) {
+    return {
+      label: 'Edit Comfy busy',
+      className: 'border-amber-400/40 text-amber-200',
+      dot: 'bg-amber-400'
+    }
+  }
+  return {
+    label: 'Edit Comfy online',
+    className: 'border-emerald-400/30 text-emerald-300',
+    dot: 'bg-emerald-400'
+  }
+})
+
 function snap32(value: number) {
   return Math.max(32, Math.round(value / 32) * 32)
 }
@@ -1060,9 +1269,16 @@ async function loadLibrary() {
   applyLibrary(data)
 }
 
-function applyComfyHealth(health: { comfy?: { ok?: boolean; busy?: boolean } } | null | undefined) {
+function applyComfyHealth(health: {
+  comfy?: { ok?: boolean; busy?: boolean }
+  imageComfy?: { ok?: boolean; busy?: boolean; configured?: boolean; host?: string }
+} | null | undefined) {
   comfyOk.value = Boolean(health?.comfy?.ok)
   comfyBusy.value = Boolean(health?.comfy?.ok && health?.comfy?.busy)
+  imageComfyConfigured.value = Boolean(health?.imageComfy?.configured)
+  imageComfyOk.value = Boolean(health?.imageComfy?.ok)
+  imageComfyBusy.value = Boolean(health?.imageComfy?.ok && health?.imageComfy?.busy)
+  imageComfyHost.value = String(health?.imageComfy?.host || '')
 }
 
 async function pollComfyHealth() {
@@ -1075,6 +1291,8 @@ onMounted(async () => {
     hideInputPreview.value = localStorage.getItem('aigen-hide-input-preview') === 'true'
     withSound.value = localStorage.getItem('aigen-generate-sound') !== 'false'
     autoplayEnabled.value = localStorage.getItem('aigen-autoplay') === 'true'
+    videoWorkflow.value = localStorage.getItem('aigen-video-workflow') === 'v2' ? 'v2' : 'v1'
+    useIdentityRefs.value = localStorage.getItem('aigen-use-identity-refs') === 'true'
   } catch { /* ignore */ }
   applyInstancePrivacy()
   window.addEventListener('pointerdown', onSitePointerDown, true)
@@ -1129,6 +1347,18 @@ watch(autoplayEnabled, (value) => {
   }
 })
 
+watch(videoWorkflow, (value) => {
+  try {
+    localStorage.setItem('aigen-video-workflow', value)
+  } catch { /* ignore */ }
+})
+
+watch(useIdentityRefs, (value) => {
+  try {
+    localStorage.setItem('aigen-use-identity-refs', String(value))
+  } catch { /* ignore */ }
+})
+
 watch(turbo, (on) => {
   if (!cfgTouched.value) cfg.value = on ? CFG_TURBO : CFG_QUALITY
 })
@@ -1221,10 +1451,42 @@ function readFile(next: File) {
   img.src = preview.value
 }
 
+function openPicker(slot: 'main' | number = 'main') {
+  pickerSlot.value = slot
+  pickerOpen.value = true
+}
+
+function setIdentityRef(index: number, next: File) {
+  const prev = identityPreviews.value[index]
+  if (prev) URL.revokeObjectURL(prev)
+  const copies = identityRefs.value.slice()
+  copies[index] = next
+  identityRefs.value = copies
+  const previews = identityPreviews.value.slice()
+  previews[index] = URL.createObjectURL(next)
+  identityPreviews.value = previews
+}
+
+function clearIdentityRef(index: number) {
+  const prev = identityPreviews.value[index]
+  if (prev) URL.revokeObjectURL(prev)
+  const copies = identityRefs.value.slice()
+  copies[index] = null
+  identityRefs.value = copies
+  const previews = identityPreviews.value.slice()
+  previews[index] = ''
+  identityPreviews.value = previews
+}
+
 function onPickedImage(next: File, hidden = false) {
+  const slot = pickerSlot.value
+  pickerOpen.value = false
+  if (slot !== 'main') {
+    setIdentityRef(Number(slot), next)
+    return
+  }
   if (hidden && showPrivacyToggles.value) hideInputPreview.value = true
   readFile(next)
-  pickerOpen.value = false
 }
 
 async function persistDroppedStill(next: File) {
@@ -1364,6 +1626,7 @@ async function loadDraft(draft: RetryDraft) {
   withSound.value = draft.sound !== false
   if (draft.hideInput && showPrivacyToggles.value) hideInputPreview.value = true
   duration.value = clampDuration(Number(draft.duration))
+  if (draft.workflow === 'v2') videoWorkflow.value = 'v2'
   extensionQueue.value = (draft.extensions || []).map(item => ({
     id: crypto.randomUUID(),
     prompt: String(item.prompt || ''),
@@ -1643,6 +1906,73 @@ function removeQueuedExtension(id: string) {
   extensionQueue.value = extensionQueue.value.filter(item => item.id !== id)
 }
 
+async function editImage() {
+  if (!file.value || !prompt.value.trim()) return
+  if (!folderId.value) {
+    toast('Choose a library folder before editing.')
+    return
+  }
+  statusBusy.value = false
+  busy.value = true
+  listenGen += 1
+  stopListen()
+  settledUi = false
+  progress.value = 2
+  videoUrl.value = ''
+  editResultUrl.value = ''
+  currentClipId.value = ''
+  closeExtend()
+  awaitingReveal.value = false
+  concealOutput.value = hideInputPreview.value
+  lockedSave.value = false
+  watchingLibrary.value = false
+  statusMessage.value = 'Checking sidecar ComfyUI...'
+  maxStep.value = editSteps.value
+  startTimer()
+  try {
+    const body = new FormData()
+    body.append('image', file.value)
+    body.append('prompt', prompt.value.trim())
+    body.append('negative', editNegative.value)
+    body.append('steps', String(editSteps.value))
+    body.append('cfg', String(editCfg.value))
+    body.append('denoise', String(editDenoise.value))
+    body.append('seed', seedInput.value || 'random')
+    body.append('folderId', folderId.value)
+    body.append('hideInput', String(showPrivacyToggles.value && hideInputPreview.value))
+    body.append('name', clipName.value.trim())
+    outputDownloadName.value = clipFileName(clipName.value.trim() || prompt.value.split(/[.!?\n]/)[0] || 'aigen-edit')
+    const started = await $fetch<{ jobId: string; steps: number; hideThumbnail?: boolean; folderLocked?: boolean }>('/api/edit', {
+      method: 'POST',
+      body
+    })
+    jobId.value = started.jobId
+    maxStep.value = started.steps
+    persistActiveJob(started.jobId, started.hideThumbnail === true || hideInputPreview.value, started.folderLocked === true)
+    listen(started.jobId, started.hideThumbnail === true || hideInputPreview.value, started.folderLocked === true)
+  } catch (error: any) {
+    settledUi = true
+    const message = error?.data?.statusMessage || error?.statusMessage || error?.message || 'Image edit failed'
+    statusMessage.value = message
+    toast(message)
+    busy.value = false
+    statusBusy.value = false
+    stopTimer()
+    void pollComfyHealth()
+  }
+}
+
+async function useEditAsInput() {
+  if (!editResultUrl.value) return
+  try {
+    const blob = await $fetch<Blob>(editResultUrl.value, { responseType: 'blob' })
+    readFile(new File([blob], `${outputDownloadName.value || 'edited'}.png`, { type: blob.type || 'image/png' }))
+    toast('Edited still is now the input')
+  } catch {
+    toast('Could not load that still as input')
+  }
+}
+
 async function generate() {
   if (!file.value || !prompt.value.trim()) return
   if (!queueReady.value) {
@@ -1653,6 +1983,11 @@ async function generate() {
     toast('Choose a library folder before generating.')
     return
   }
+  const identityFiles = identityRefs.value.filter((item): item is File => Boolean(item))
+  if (videoWorkflow.value === 'v2' && useIdentityRefs.value && !identityFiles.length) {
+    toast('Add at least one identity reference still, or turn references off.')
+    return
+  }
   const shots = shotScriptMode.value ? parsedShots.value : null
   const initialPrompt = (shots?.[0]?.prompt || prompt.value).trim()
   const queued = shots
@@ -1665,6 +2000,7 @@ async function generate() {
   settledUi = false
   progress.value = 1
   videoUrl.value = ''
+  editResultUrl.value = ''
   currentClipId.value = ''
   closeExtend()
   awaitingReveal.value = false
@@ -1700,6 +2036,13 @@ async function generate() {
     body.append('hideInput', String(showPrivacyToggles.value && hideInputPreview.value))
     body.append('sound', String(withSound.value))
     body.append('name', clipName.value.trim())
+    body.append('workflow', videoWorkflow.value)
+    body.append('useIdentityRefs', String(videoWorkflow.value === 'v2' && useIdentityRefs.value))
+    if (videoWorkflow.value === 'v2' && useIdentityRefs.value) {
+      for (const ref of identityFiles) {
+        body.append('refImage', ref)
+      }
+    }
     if (activeDraftId) body.append('draftId', activeDraftId)
     if (queued.length) {
       body.append('extensions', JSON.stringify(queued))
@@ -1780,12 +2123,34 @@ function applyJobEvent(payload: Record<string, any>, hidden: boolean, folderLock
     return
   }
 
-  if (complete && (payload.clipId || payload.filename)) {
+  if (complete && (payload.clipId || payload.filename || payload.stillId)) {
     settledUi = true
     const dest = folders.value.find(folder => folder.id === folderId.value)
     const locked = dest
       ? Boolean(dest.protected && !dest.unlocked)
       : (folderLocked || payload.folderLocked === true)
+    if (payload.stillId) {
+      if (locked) {
+        editResultUrl.value = ''
+        lockedSave.value = true
+        browseFolderId.value = folderId.value
+        statusMessage.value = 'Saved to the locked folder. Unlock it to view.'
+      } else {
+        editResultUrl.value = `/api/library/stills/${payload.stillId}`
+        awaitingReveal.value = Boolean(payload.hideThumbnail || hidden)
+        concealOutput.value = awaitingReveal.value
+        lockedSave.value = false
+        statusMessage.value = 'Edit finished'
+      }
+      busy.value = false
+      statusBusy.value = false
+      stopTimer()
+      stopListen()
+      clearActiveJob()
+      loadLibrary().catch(() => null)
+      void pollComfyHealth()
+      return
+    }
     const keepLibraryView = watchingLibrary.value && Boolean(currentClipId.value) && payload.clipId !== currentClipId.value
     if (keepLibraryView) {
       busy.value = false