@shop-os/modelkit 0.4.1 → 0.4.3

package/dist/index.cjs

@@ -1497,6 +1497,7 @@ var OMNI_RATIOS = ["16:9", "9:16"];
 var GROK_RATIOS = ["1:1", "16:9", "9:16", "4:3", "3:4", "3:2", "2:3"];
 var H3_RATIOS = ["21:9", "16:9", "4:3", "1:1", "3:4", "9:16"];
 var H3_REFERENCE_RATIOS = ["adaptive", ...H3_RATIOS];
+var WAN_RATIOS = ["adaptive", "16:9", "4:3", "1:1", "3:4", "9:16"];
 var NEGATIVE_PROMPT = "blur, distort, and low quality";
 var FRAMES_EXCLUDE_REFERENCES = { framesExcludeReferences: true };
 var KLING_PROMPT = { referenceSyntax: "numbered", maxChars: 2500 };
@@ -1507,6 +1508,7 @@ var SEEDANCE_2_AUDIO = {
   audioMaxBytes: 15 * 1024 * 1024
 };
 var H3_PROMPT = { referenceSyntax: "numbered", maxChars: 5e4 };
+var WAN_PROMPT = { referenceSyntax: "numbered", maxChars: 2e4 };
 var OMNI_PROMPT = { referenceSyntax: "caption", maxChars: 5e4 };
 var BYTEDANCE_LOGO = "https://d31vvfpk2ppdyq.cloudfront.net/assets/image/3bba5d81-4e57-4e71-911b-cdfa05acca2c_compressed.png";
 var GOOGLE_LOGO2 = "https://d31vvfpk2ppdyq.cloudfront.net/assets/image/192485d2-a76a-4657-9c06-930b2c70badd_compressed.png";
@@ -2112,6 +2114,48 @@ var VIDEO_MODELS = [
     // references instead of the adapter silently dropping the frame.
     capabilities: { ...FRAMES_EXCLUDE_REFERENCES }
   },
+  {
+    // One row for fal's three Wan 3.0 endpoints; the adapter picks text-, image- or
+    // reference-to-video from the inputs. Each takes aspect_ratio, so none is ignored.
+    // https://fal.ai/models/alibaba/wan-3.0/reference-to-video/api
+    id: "wan-3.0",
+    aliases: ["wan-3", "wan3"],
+    label: "Wan 3",
+    type: "video",
+    adapter: "fal",
+    providerModelId: "alibaba/wan-3.0",
+    lifecycle: "active",
+    exposure: ["cowork", "space_builder"],
+    constraints: {
+      aspectRatios: [...WAN_RATIOS],
+      resolutions: ["480p", "720p", "1080p"],
+      durations: { min: 2, max: 30 },
+      durationPresets: [5, 10, 15, 30],
+      references: {
+        images: 10,
+        videos: 5,
+        audios: 5,
+        videoSeconds: { total: 15 },
+        audioSeconds: { total: 15 }
+      },
+      supports: ["seed", "lastFrame", "generateAudio", "safetyChecker"],
+      prompt: WAN_PROMPT,
+      audio: "optional"
+    },
+    defaults: {
+      durationSeconds: 5,
+      resolution: "1080p",
+      aspectRatio: "adaptive",
+      generateAudio: true
+    },
+    usage: { ...VIDEO_USAGE },
+    pricingKind: "video_per_second",
+    display: {
+      provider: "alibaba",
+      description: "Alibaba Wan 3 video model for text-to-video, image-to-video (start/end frame), and reference-to-video from up to 10 images, 5 videos and 5 audio clips, with native audio; 2-30 seconds at up to 1080p."
+    },
+    capabilities: { ...FRAMES_EXCLUDE_REFERENCES, lastFrameRequiresFirstFrame: true }
+  },
   {
     // Duration 3-10s per trigger-workflows mapDuration; the endpoint takes no resolution.
     id: "kling-o1-v2v",
@@ -4908,7 +4952,7 @@ function isMinimax(providerModelId) {
   return MINIMAX_MODEL_IDS.has(providerModelId.toLowerCase());
 }
 function isFalVideoKnown(providerModelId) {
-  return isKling(providerModelId) || isMinimax(providerModelId);
+  return isKling(providerModelId) || isMinimax(providerModelId) || isWan(providerModelId);
 }
 function minimaxFamily(providerModelId) {
   return H3_MAX_MODEL_IDS.has(providerModelId.toLowerCase()) ? "h3-max" : "h3";
@@ -4925,6 +4969,44 @@ function minimaxBase(req, family) {
   if (req.seed !== void 0) input.seed = req.seed;
   return input;
 }
+var WAN_MODEL_IDS = /* @__PURE__ */ new Set(["alibaba/wan-3.0"]);
+function isWan(providerModelId) {
+  return WAN_MODEL_IDS.has(providerModelId.toLowerCase());
+}
+function wanVideo(req, family) {
+  const input = {
+    prompt: req.prompt.trim(),
+    duration: req.durationSeconds ?? 5,
+    resolution: req.resolution ?? "1080p",
+    aspect_ratio: req.aspectRatio ?? "adaptive",
+    audio: req.generateAudio ?? true
+  };
+  if (req.safetyChecker !== void 0) input.enable_safety_checker = req.safetyChecker;
+  if (req.seed !== void 0) input.seed = req.seed;
+  const images = (req.referenceImages ?? []).map(refToUrl);
+  const videos = (req.referenceVideos ?? []).map(refToUrl);
+  const audios = (req.referenceAudios ?? []).map(refToUrl);
+  if (images.length + videos.length + audios.length > 0) {
+    if (images.length > 0) input.reference_image_urls = images;
+    if (videos.length > 0) input.reference_video_urls = videos;
+    if (audios.length > 0) input.reference_audio_urls = audios;
+    return { endpoint: `${family}/reference-to-video`, input };
+  }
+  if (req.firstFrame) {
+    input.start_image_url = refToUrl(req.firstFrame);
+    if (req.lastFrame) input.end_image_url = refToUrl(req.lastFrame);
+    return { endpoint: `${family}/image-to-video`, input };
+  }
+  if (req.lastFrame) {
+    throw new ModelKitError({
+      code: "missing_input",
+      message: `${family} takes a last frame only together with a first frame`,
+      provider: "fal",
+      retryable: false
+    });
+  }
+  return { endpoint: `${family}/text-to-video`, input };
+}
 var KLING_MODEL_IDS = /* @__PURE__ */ new Set([
   "fal-ai/kling-video/o1/video-to-video/reference"
 ]);
@@ -4980,10 +5062,17 @@ async function generateFalVideo(req, ctx, deps) {
     }
     return await submitQueueJob(ctx, req, deps, KLING_O1_ENDPOINT, input);
   }
+  if (isWan(providerModelId)) {
+    const { endpoint: endpoint2, input } = wanVideo(req, providerModelId.toLowerCase());
+    if (ctx.routing.fal === "cloudflare-passthrough") {
+      return await passthroughVideoJob(ctx, req, endpoint2, input);
+    }
+    return await submitQueueJob(ctx, req, deps, endpoint2, input);
+  }
   if (!isMinimax(providerModelId)) {
     throw new ModelKitError({
       code: "unsupported_option",
-      message: `fal has no input mapper for "${providerModelId}"; add the id to KLING_MODEL_IDS or MINIMAX_MODEL_IDS`,
+      message: `fal has no input mapper for "${providerModelId}"; add the id to KLING_MODEL_IDS, MINIMAX_MODEL_IDS or WAN_MODEL_IDS`,
       provider: "fal",
       model: ctx.model.id,
       retryable: false
@@ -5668,8 +5757,10 @@ async function generateOmniVideo(req, ctx, createGenAIClient) {
   const apiKey = googleKey(ctx);
   const modelId = ctx.model.id;
   const providerModelId = ctx.model.providerModelId;
-  const duration = req.durationSeconds ?? 10;
   const extend = req.taskType === "extend";
+  const sourceSeconds = extend ? void 0 : req.sourceVideo?.durationSeconds;
+  const editSeconds = sourceSeconds !== void 0 && Number.isFinite(sourceSeconds) && sourceSeconds > 0 ? sourceSeconds : void 0;
+  const duration = req.durationSeconds ?? editSeconds ?? 10;
   const previousId = extend ? req.previousInteractionId : void 0;
   const chained = previousId !== void 0;
   for (const field of [
@@ -8148,7 +8239,7 @@ function dataOnlyCapable(row) {
     }
     case "fal": {
       if (row.type === "video") {
-        return isFalVideoKnown(row.providerModelId) ? { ok: true } : { ok: false, reason: "fal video needs a Kling or MiniMax id" };
+        return isFalVideoKnown(row.providerModelId) ? { ok: true } : { ok: false, reason: "fal video needs a Kling, MiniMax or Wan id" };
       }
       return { ok: true };
     }