@shop-os/modelkit 0.2.26 → 0.3.0

package/dist/catalog.js

@@ -1293,6 +1293,11 @@ var NEGATIVE_PROMPT = "blur, distort, and low quality";
 var FRAMES_EXCLUDE_REFERENCES = { framesExcludeReferences: true };
 var KLING_PROMPT = { referenceSyntax: "numbered", maxChars: 2500 };
 var SEEDANCE_PROMPT = { referenceSyntax: "numbered", maxChars: 3e4 };
+var SEEDANCE_2_AUDIO = {
+  audioSeconds: { min: 2, max: 15, total: 15 },
+  audioFormats: ["wav", "mp3"],
+  audioMaxBytes: 15 * 1024 * 1024
+};
 var H3_PROMPT = { referenceSyntax: "numbered", maxChars: 5e4 };
 var OMNI_PROMPT = { referenceSyntax: "caption", maxChars: 5e4 };
 var BYTEDANCE_LOGO = "https://d31vvfpk2ppdyq.cloudfront.net/assets/image/3bba5d81-4e57-4e71-911b-cdfa05acca2c_compressed.png";
@@ -1429,7 +1434,7 @@ var VIDEO_MODELS = [
       description: "Google's conversational video model with native audio, human support, and natural language editing. Aspect ratio only \u2014 resolution and duration are model-determined.",
       logo: GOOGLE_LOGO2
     },
-    capabilities: { acceptsSourceVideo: true }
+    capabilities: { acceptsSourceVideo: true, videoEdit: true }
   },
   {
     id: "gemini-omni-1.1-flash",
@@ -1454,7 +1459,7 @@ var VIDEO_MODELS = [
       // with a video task, so those calls carry the prompt alone).
       // First-to-last-frame interpolation: `<LAST_FRAME>` goes with `<FIRST_FRAME>`.
       supports: ["videoExtension", "lastFrame"],
-      extension: { segmentSeconds: 10, maxTotalSeconds: 40 },
+      extension: { segmentSeconds: 10, maxTotalSeconds: 40, returns: "joined" },
       prompt: OMNI_PROMPT,
       audio: "always"
     },
@@ -1466,7 +1471,11 @@ var VIDEO_MODELS = [
       logo: GOOGLE_LOGO2,
       description: "Google Gemini Omni Flash 1.1 conversational video generation, editing and extension; text, image, and up to 10-second video input, 3-10 second output at up to 4K and 24 FPS, extendable to 40 seconds."
     },
-    capabilities: { acceptsSourceVideo: true, lastFrameRequiresFirstFrame: true }
+    capabilities: {
+      acceptsSourceVideo: true,
+      videoEdit: true,
+      lastFrameRequiresFirstFrame: true
+    }
   },
   {
     // vibe-triggers and Space Builder drop the ratio when a frame is wired; Cowork always sends it.
@@ -1488,7 +1497,8 @@ var VIDEO_MODELS = [
         videos: 3,
         audios: 3,
         audioOnly: false,
-        videoSeconds: { min: 2, max: 15, total: 15 }
+        videoSeconds: { min: 2, max: 15, total: 15 },
+        ...SEEDANCE_2_AUDIO
       },
       sourceVideoSeconds: { min: 2, max: 15, inReferenceTotal: true },
       ratioIgnoredWith: ["firstFrame", "sourceVideo"],
@@ -1510,7 +1520,7 @@ var VIDEO_MODELS = [
       description: "High-quality AI video generation with reference image conditioning. Supports 4-15 second clips with human preprocessing.",
       logo: BYTEDANCE_LOGO
     },
-    capabilities: { acceptsSourceVideo: true, ...FRAMES_EXCLUDE_REFERENCES }
+    capabilities: { acceptsSourceVideo: true, videoEdit: false, ...FRAMES_EXCLUDE_REFERENCES }
   },
   {
     id: "seedance-2.0-fast",
@@ -1530,7 +1540,8 @@ var VIDEO_MODELS = [
         videos: 3,
         audios: 3,
         audioOnly: false,
-        videoSeconds: { min: 2, max: 15, total: 15 }
+        videoSeconds: { min: 2, max: 15, total: 15 },
+        ...SEEDANCE_2_AUDIO
       },
       sourceVideoSeconds: { min: 2, max: 15, inReferenceTotal: true },
       ratioIgnoredWith: ["firstFrame", "sourceVideo"],
@@ -1552,7 +1563,7 @@ var VIDEO_MODELS = [
       logo: BYTEDANCE_LOGO,
       description: "Speed- and cost-optimized ByteDance Seedance 2.0 Fast; text plus image, video, and audio references to generate or edit 4-15 second clips."
     },
-    capabilities: { acceptsSourceVideo: true, ...FRAMES_EXCLUDE_REFERENCES }
+    capabilities: { acceptsSourceVideo: true, videoEdit: false, ...FRAMES_EXCLUDE_REFERENCES }
   },
   {
     // omni_reference_task_type edit/extend: adaptive ratio, edit sends duration -1.
@@ -1572,12 +1583,20 @@ var VIDEO_MODELS = [
         images: 30,
         videos: 10,
         audios: 10,
-        videoSeconds: { min: 2, max: 30, total: 30, editMin: 4 }
+        videoSeconds: { min: 2, max: 30, total: 30, editMin: 4 },
+        // BytePlus: 2-30 s per clip, 30 s in all, wav/mp3, 15 MB each.
+        audioSeconds: { min: 2, max: 30, total: 30 },
+        audioFormats: ["wav", "mp3"],
+        audioMaxBytes: 15 * 1024 * 1024
       },
       sourceVideoSeconds: { min: 2, max: 30, editMin: 4, inReferenceTotal: true },
       ratioIgnoredWith: ["firstFrame", "sourceVideo"],
       // No seed: 2.x echoes it but does not honour it (two seeded runs differed, 2026-10-01).
-      supports: ["generateAudio", "lastFrame", "watermark", "taskType"],
+      supports: ["generateAudio", "lastFrame", "watermark", "taskType", "videoExtension"],
+      // Extend (probe 2026-10-02, 480p): returns ONLY the added footage, continuing from
+      // the source's last frame at the source's framing and fps but at the requested
+      // resolution, so the caller joins the clips. Forward only in practice.
+      extension: { segmentSeconds: 30, maxTotalSeconds: 60, returns: "new" },
       prompt: SEEDANCE_PROMPT,
       audio: "optional"
     },
@@ -1594,7 +1613,7 @@ var VIDEO_MODELS = [
       logo: BYTEDANCE_LOGO,
       description: "ByteDance Seedance 2.5 for long-form storytelling and omni-reference generation/editing; up to 50 reference assets and 4-30 second clips at up to 1080p."
     },
-    capabilities: { acceptsSourceVideo: true, ...FRAMES_EXCLUDE_REFERENCES }
+    capabilities: { acceptsSourceVideo: true, videoEdit: true, ...FRAMES_EXCLUDE_REFERENCES }
   },
   {
     id: "grok-imagine",
@@ -1958,11 +1977,116 @@ var VIDEO_MODELS = [
   }
 ];
 
+// src/catalog/video-tools.ts
+var TOPAZ_LOGO = "https://cdn.jsdelivr.net/npm/@lobehub/icons-static-png@1.97.0/light/topazlabs.png";
+var TOPAZ_SOURCE_SECONDS = { max: 300 };
+var TOPAZ_USAGE = { unit: "seconds" };
+var VIDEO_TOOL_MODELS = [
+  {
+    id: "topaz-precision",
+    aliases: [],
+    label: "Topaz Precision",
+    type: "video-tool",
+    adapter: "fal",
+    providerModelId: "topaz/upscale/video/precision",
+    lifecycle: "active",
+    exposure: ["cowork"],
+    constraints: {
+      supports: [],
+      sourceVideoSeconds: TOPAZ_SOURCE_SECONDS,
+      tool: { kind: "upscale", upscaleFactor: { min: 1, max: 4 }, maxOutputLongEdge: 3840 }
+    },
+    defaults: {},
+    providerOptions: { fal: { model: "Proteus" } },
+    usage: TOPAZ_USAGE,
+    pricingKind: "video_per_second",
+    display: {
+      provider: "topaz",
+      logo: TOPAZ_LOGO,
+      description: "Topaz Proteus upscaling: sharper, cleaner footage up to 4K while staying faithful to the source.",
+      sortOrder: 20
+    }
+  },
+  {
+    id: "topaz-starlight",
+    aliases: [],
+    label: "Topaz Starlight",
+    type: "video-tool",
+    adapter: "fal",
+    providerModelId: "topaz/upscale/video/generative",
+    lifecycle: "active",
+    exposure: ["cowork"],
+    constraints: {
+      supports: [],
+      sourceVideoSeconds: TOPAZ_SOURCE_SECONDS,
+      tool: { kind: "upscale", upscaleFactor: { min: 1, max: 4 }, maxOutputLongEdge: 3840 }
+    },
+    defaults: {},
+    providerOptions: { fal: { model: "Starlight Precise 2.6" } },
+    usage: TOPAZ_USAGE,
+    pricingKind: "video_per_second",
+    display: {
+      provider: "topaz",
+      logo: TOPAZ_LOGO,
+      description: "Topaz Starlight Precise 2.6: generative upscaling that rebuilds fine detail, up to 4K. Slower and costlier than Precision.",
+      sortOrder: 10
+    }
+  },
+  {
+    id: "topaz-apollo",
+    aliases: [],
+    label: "Topaz Apollo",
+    type: "video-tool",
+    adapter: "fal",
+    providerModelId: "topaz/interpolate/video",
+    lifecycle: "active",
+    exposure: ["cowork"],
+    constraints: {
+      supports: [],
+      sourceVideoSeconds: TOPAZ_SOURCE_SECONDS,
+      tool: { kind: "interpolate", fps: { min: 16, max: 120 } }
+    },
+    defaults: {},
+    providerOptions: { fal: { model: "Apollo" } },
+    usage: TOPAZ_USAGE,
+    pricingKind: "video_per_second",
+    display: {
+      provider: "topaz",
+      logo: TOPAZ_LOGO,
+      description: "Topaz Apollo frame interpolation: smoother motion at up to 120 fps, same length as the source."
+    }
+  },
+  {
+    id: "topaz-hyperion-2.5",
+    aliases: [],
+    label: "Topaz Hyperion 2.5",
+    type: "video-tool",
+    adapter: "fal",
+    providerModelId: "topaz/sdr-to-hdr/video",
+    lifecycle: "active",
+    exposure: ["cowork"],
+    constraints: {
+      supports: [],
+      sourceVideoSeconds: TOPAZ_SOURCE_SECONDS,
+      tool: { kind: "hdr", outputFormats: ["h265", "prores"] }
+    },
+    defaults: {},
+    usage: TOPAZ_USAGE,
+    pricingKind: "video_per_second",
+    display: {
+      provider: "topaz",
+      logo: TOPAZ_LOGO,
+      description: "Topaz Hyperion 2.5 SDR to HDR: a 10-bit HDR master in H.265 (HDR10) or ProRes 422 HQ, at the source's size and frame rate."
+    }
+  }
+];
+
 // src/catalog/models.ts
 var MODELS = [
   ...TEXT_MODELS,
   ...IMAGE_MODELS,
   ...VIDEO_MODELS,
+  ...VIDEO_TOOL_MODELS,
   ...OTHER_MODELS
 ];
 
@@ -2268,7 +2392,15 @@ var ADAPTER_IDS = [
   "higgsfield",
   "xai-video"
 ];
-var ModelTypeSchema = z3.enum(["text", "image", "video", "audio", "upscale", "bg-remove"]);
+var ModelTypeSchema = z3.enum([
+  "text",
+  "image",
+  "video",
+  "audio",
+  "upscale",
+  "bg-remove",
+  "video-tool"
+]);
 var LifecycleSchema = z3.enum(["active", "deprecated", "retired"]);
 var SurfaceSchema = z3.enum(["cowork", "space_builder"]);
 var SURFACES2 = SurfaceSchema.options;
@@ -2348,7 +2480,22 @@ var ReferenceLimitsSchema = z3.looseObject({
   audios: z3.number().optional(),
   total: z3.number().optional(),
   audioOnly: z3.boolean().optional(),
-  videoSeconds: VideoSecondsLimitsSchema.optional()
+  videoSeconds: VideoSecondsLimitsSchema.optional(),
+  audioSeconds: z3.looseObject({
+    min: z3.number().optional(),
+    max: z3.number().optional(),
+    total: z3.number().optional()
+  }).optional(),
+  audioFormats: z3.array(z3.string()).optional(),
+  audioMaxBytes: z3.number().positive().optional()
+});
+var RangeSchema = z3.looseObject({ min: z3.number(), max: z3.number() });
+var VideoToolConstraintsSchema = z3.looseObject({
+  kind: z3.enum(["upscale", "interpolate", "hdr"]),
+  upscaleFactor: RangeSchema.optional(),
+  maxOutputLongEdge: z3.number().positive().optional(),
+  fps: RangeSchema.optional(),
+  outputFormats: z3.array(z3.enum(["h265", "prores"])).min(1).optional()
 });
 var ModelConstraintsSchema = z3.looseObject({
   aspectRatios: z3.array(AspectRatioSchema).optional(),
@@ -2363,7 +2510,14 @@ var ModelConstraintsSchema = z3.looseObject({
   supports: z3.array(SupportedOptionSchema).optional(),
   ratioIgnoredWith: z3.array(z3.enum(["firstFrame", "sourceVideo"])).optional(),
   sourceVideoSeconds: VideoSecondsLimitsSchema.omit({ total: true }).extend({ inReferenceTotal: z3.boolean().optional() }).optional(),
-  extension: z3.looseObject({ segmentSeconds: z3.number(), maxTotalSeconds: z3.number() }).optional(),
+  // segmentSeconds/maxTotalSeconds stay required: a kit before 0.3.0 drops a row whose
+  // extension lacks them, so every extension block keeps them.
+  extension: z3.looseObject({
+    segmentSeconds: z3.number(),
+    maxTotalSeconds: z3.number(),
+    returns: z3.enum(["new", "joined"]).optional()
+  }).optional(),
+  tool: VideoToolConstraintsSchema.optional(),
   // Loose on purpose: the read path keeps a row whose value is invalid (the accessors in
   // conventions.ts read it as "not set"); CatalogModelSchema below rejects it on write.
   audio: z3.unknown().optional(),
@@ -2435,6 +2589,7 @@ var CatalogDisplaySchema = z3.looseObject({
 });
 var CatalogCapabilitiesSchema = z3.looseObject({
   acceptsSourceVideo: z3.boolean().optional(),
+  videoEdit: z3.boolean().optional(),
   framesExcludeReferences: z3.boolean().optional(),
   lastFrameRequiresFirstFrame: z3.boolean().optional(),
   reasoning: z3.boolean().optional(),
@@ -2477,6 +2632,13 @@ var CatalogModelSchemaBase = z3.looseObject({
       message: "active text rows need contextWindow"
     });
   }
+  if (row.type === "video-tool" && row.constraints.tool === void 0) {
+    ctx.addIssue({
+      code: "custom",
+      path: ["constraints", "tool"],
+      message: "video-tool rows need constraints.tool"
+    });
+  }
   if (row.replacedBy !== void 0 && row.replacedBy === row.id) {
     ctx.addIssue({
       code: "custom",
@@ -2782,6 +2944,114 @@ function resetLiveCatalog() {
   liveCatalog.replace([...MODELS]);
 }
 
+// src/catalog/video-tools-spec.ts
+import { z as z6 } from "zod";
+var RangeSchema2 = z6.object({ min: z6.number(), max: z6.number() });
+var ToolSchema = z6.object({
+  kind: z6.enum(["upscale", "interpolate", "hdr"]),
+  upscaleFactor: RangeSchema2.optional(),
+  maxOutputLongEdge: z6.number().positive().optional(),
+  fps: RangeSchema2.optional(),
+  outputFormats: z6.array(z6.enum(["h265", "prores"])).min(1).optional()
+});
+function videoToolConstraints(row) {
+  if (row.type !== "video-tool") return void 0;
+  const parsed = ToolSchema.safeParse(row.constraints.tool);
+  if (!parsed.success) return void 0;
+  const { kind, upscaleFactor, maxOutputLongEdge, fps, outputFormats } = parsed.data;
+  return {
+    kind,
+    ...upscaleFactor !== void 0 && { upscaleFactor },
+    ...maxOutputLongEdge !== void 0 && { maxOutputLongEdge },
+    ...fps !== void 0 && { fps },
+    ...outputFormats !== void 0 && { outputFormats }
+  };
+}
+function resolutionTier(width, height) {
+  const shortEdge = Math.min(width, height);
+  if (shortEdge <= 360) return "360p";
+  if (shortEdge <= 480) return "480p";
+  if (shortEdge <= 720) return "720p";
+  if (shortEdge <= 1080) return "1080p";
+  if (shortEdge <= 1440) return "2k";
+  return "4k";
+}
+function fail(error) {
+  return { ok: false, error };
+}
+function planVideoTool(row, req) {
+  const tool = videoToolConstraints(row);
+  if (!tool) return fail(`${row.id} is not a video tool`);
+  const { width, height, durationSeconds, fps } = req.source;
+  if (!(width > 0 && height > 0 && durationSeconds > 0)) {
+    return fail("The source video's width, height and duration are required");
+  }
+  const maxSeconds = row.constraints.sourceVideoSeconds?.max;
+  if (maxSeconds !== void 0 && durationSeconds > maxSeconds) {
+    return fail(`${row.label} takes source videos up to ${maxSeconds} seconds.`);
+  }
+  const base = {
+    durationSeconds,
+    ...fps !== void 0 && { fps }
+  };
+  if (tool.kind === "upscale") {
+    const factor = req.upscaleFactor;
+    if (factor === void 0) return fail("An upscale factor is required");
+    const range = tool.upscaleFactor;
+    if (range && (factor < range.min || factor > range.max)) {
+      return fail(`${row.label} upscales by ${range.min}x to ${range.max}x.`);
+    }
+    const outWidth = Math.round(width * factor);
+    const outHeight = Math.round(height * factor);
+    const limit = tool.maxOutputLongEdge;
+    if (limit !== void 0 && Math.max(outWidth, outHeight) > limit) {
+      return fail(
+        `A ${factor}x upscale of a ${width}x${height} video passes the ${limit}px limit.`
+      );
+    }
+    return {
+      ok: true,
+      output: {
+        ...base,
+        width: outWidth,
+        height: outHeight,
+        resolution: resolutionTier(outWidth, outHeight),
+        format: "h264"
+      }
+    };
+  }
+  if (tool.kind === "interpolate") {
+    const target = req.targetFps;
+    if (target === void 0 || !Number.isInteger(target)) {
+      return fail("A whole-number target frame rate is required");
+    }
+    const range = tool.fps;
+    if (range && (target < range.min || target > range.max)) {
+      return fail(`${row.label} outputs ${range.min} to ${range.max} fps.`);
+    }
+    return {
+      ok: true,
+      output: {
+        durationSeconds,
+        fps: target,
+        width,
+        height,
+        resolution: resolutionTier(width, height),
+        format: "h264"
+      }
+    };
+  }
+  const formats = tool.outputFormats ?? ["h265"];
+  const format = req.outputFormat ?? formats[0] ?? "h265";
+  if (!formats.includes(format)) {
+    return fail(`${row.label} writes ${formats.join(" or ")}.`);
+  }
+  return {
+    ok: true,
+    output: { ...base, width, height, resolution: resolutionTier(width, height), format }
+  };
+}
+
 // src/resolve-options.ts
 var SIZE_ORDER = ["512", "1K", "2K", "4K"];
 var RESOLUTION_ORDER = ["360p", "480p", "720p", "768p", "1080p", "2k", "4k"];
@@ -3270,6 +3540,7 @@ export {
   SurfaceSchema,
   TEXT_MODELS,
   VIDEO_MODELS,
+  VIDEO_TOOL_MODELS,
   billedAudio,
   compareCatalogOrder,
   isListed,
@@ -3280,14 +3551,17 @@ export {
   nearestAspectRatio,
   normalizeExposure,
   outputAudio,
+  planVideoTool,
   primeCatalog,
   promptConventions,
   resetLiveCatalog,
+  resolutionTier,
   resolveAudioOptions,
   resolveImageOptions,
   resolveOptions,
   resolveVideoOptions,
   sortCatalogRows,
-  videoLengthViolation
+  videoLengthViolation,
+  videoToolConstraints
 };
 //# sourceMappingURL=catalog.js.map