@shop-os/modelkit 0.14.0 → 0.15.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (14) hide show
  1. package/CHANGELOG.md +9 -0
  2. package/dist/catalog.cjs +34 -1
  3. package/dist/catalog.cjs.map +1 -1
  4. package/dist/catalog.d.cts +31 -1
  5. package/dist/catalog.d.ts +31 -1
  6. package/dist/catalog.js +34 -1
  7. package/dist/catalog.js.map +1 -1
  8. package/dist/index.cjs +35 -1
  9. package/dist/index.cjs.map +1 -1
  10. package/dist/index.d.cts +31 -1
  11. package/dist/index.d.ts +31 -1
  12. package/dist/index.js +34 -1
  13. package/dist/index.js.map +1 -1
  14. package/package.json +1 -1
package/dist/index.d.cts CHANGED
@@ -601,6 +601,11 @@ type SupportedOption = "seed" | "quality" | "negativePrompt" | "cfgScale" | "gen
601
601
  | "insertWindow"
602
602
  /** Mask rows: `ImageRequest.points` and `ImageRequest.boxes`. */
603
603
  | "points" | "boxes";
604
+ /** Smallest frame a provider takes for an input video, in pixels (fal: "Min width: 700px"). */
605
+ interface VideoPixelLimits {
606
+ minWidth?: number;
607
+ minHeight?: number;
608
+ }
604
609
  interface VideoSecondsLimits {
605
610
  /** Minimum length of each reference video, in seconds. */
606
611
  min?: number;
@@ -622,6 +627,8 @@ interface ReferenceLimits {
622
627
  audioOnly?: boolean;
623
628
  /** Per-video length caps (min/max each) and combined total for reference videos. */
624
629
  videoSeconds?: VideoSecondsLimits;
630
+ /** Smallest frame each reference video may have (Kling 4.0: 700 px). Kits before 0.15.0 ignore it. */
631
+ videoPixels?: VideoPixelLimits;
625
632
  /** Per-clip length caps (min/max each) and combined total for reference audio, in seconds. */
626
633
  audioSeconds?: {
627
634
  min?: number;
@@ -751,6 +758,8 @@ interface ModelConstraints {
751
758
  languages?: LanguageOption[];
752
759
  /** Audio-output rows whose provider lets the caller pick a file format; `defaults.outputFormat` is one. */
753
760
  outputFormats?: string[];
761
+ /** Smallest frame the source video may have (Kling 4.0 Edit: 700 px). Kits before 0.15.0 ignore it. */
762
+ sourceVideoPixels?: VideoPixelLimits;
754
763
  /** Length limits for the source video the row edits or extends. */
755
764
  sourceVideoSeconds?: Omit<VideoSecondsLimits, "total"> & {
756
765
  /** The provider sends the source as a reference video, so it counts toward `references.videoSeconds.total` (Seedance). */
@@ -1253,6 +1262,10 @@ declare const CatalogModelSchemaBase: z.ZodObject<{
1253
1262
  total: z.ZodOptional<z.ZodNumber>;
1254
1263
  editMin: z.ZodOptional<z.ZodNumber>;
1255
1264
  }, z.core.$loose>>;
1265
+ videoPixels: z.ZodOptional<z.ZodObject<{
1266
+ minWidth: z.ZodOptional<z.ZodNumber>;
1267
+ minHeight: z.ZodOptional<z.ZodNumber>;
1268
+ }, z.core.$loose>>;
1256
1269
  audioSeconds: z.ZodOptional<z.ZodObject<{
1257
1270
  min: z.ZodOptional<z.ZodNumber>;
1258
1271
  max: z.ZodOptional<z.ZodNumber>;
@@ -1330,6 +1343,10 @@ declare const CatalogModelSchemaBase: z.ZodObject<{
1330
1343
  name: z.ZodString;
1331
1344
  }, z.core.$loose>>>;
1332
1345
  outputFormats: z.ZodOptional<z.ZodArray<z.ZodString>>;
1346
+ sourceVideoPixels: z.ZodOptional<z.ZodObject<{
1347
+ minWidth: z.ZodOptional<z.ZodNumber>;
1348
+ minHeight: z.ZodOptional<z.ZodNumber>;
1349
+ }, z.core.$loose>>;
1333
1350
  sourceVideoSeconds: z.ZodOptional<z.ZodObject<{
1334
1351
  max: z.ZodOptional<z.ZodNumber>;
1335
1352
  min: z.ZodOptional<z.ZodNumber>;
@@ -1858,6 +1875,19 @@ declare function videoLengthViolation(model: CatalogModel, input: {
1858
1875
  sourceVideo?: number;
1859
1876
  taskType?: VideoRequest["taskType"];
1860
1877
  }): string | undefined;
1878
+ type MeasuredFrame = {
1879
+ width?: number;
1880
+ height?: number;
1881
+ } | undefined;
1882
+ /**
1883
+ * First frame-size violation for the videos a request sends, as a user-facing message, from
1884
+ * their measured pixel sizes (`references.videoPixels`, `sourceVideoPixels`). An unmeasured
1885
+ * video passes: the provider still has the last word.
1886
+ */
1887
+ declare function videoSizeViolation(model: CatalogModel, input: {
1888
+ referenceVideos?: MeasuredFrame[];
1889
+ sourceVideo?: MeasuredFrame;
1890
+ }): string | undefined;
1861
1891
  declare function resolveImageOptions(model: CatalogModel, request: ImageRequest, onUnsupported?: OnUnsupported): ResolveResult<ImageRequest>;
1862
1892
  declare function resolveVideoOptions(model: CatalogModel, request: VideoRequest, onUnsupported?: OnUnsupported): ResolveResult<VideoRequest>;
1863
1893
  declare function resolveAudioOptions(model: CatalogModel, request: AudioRequest, onUnsupported?: OnUnsupported): ResolveResult<AudioRequest>;
@@ -2263,4 +2293,4 @@ declare function describeThrown(cause: unknown): ThrownErrorInfo;
2263
2293
  /** Add the AI SDK `GeneratedFile` accessors to adapter image bytes. base64 is computed on first read. */
2264
2294
  declare function toGeneratedImage(image: ImageBytes): GeneratedImage;
2265
2295
 
2266
- export { ADAPTER_IDS, ADAPTER_KEYS, ARRAY_INPUT_SPEC_KEYS, AUDIO_INPUT_SPEC_KEYS, type Adapter, type AdapterContext, type AdapterEndpoints, type AdapterId, type AdapterRegistry, type AiTextProviderOptions, type ArrayInputSpecKey, type AspectRatio, type AudioNodeMethods, type AudioOutput, type AudioRequest, type AudioResult, type CallContext, Catalog, type CatalogAuth, type CatalogCache, type CatalogCapabilities, type CatalogDisplay, type CatalogEnvelope, type CatalogModel, CatalogModelSchema, CatalogModelSchemaBase, type CatalogOrigin, CatalogSchema, type CatalogSource, type CatalogTextModel, DEFAULT_ADAPTERS, type DataOnlyRow, type DataOnlyVerdict, type DefaultObjectOutput, type DubbingOutcome, type DubbingRequest, type DubbingResult, type Exposure, ExposureSchema, type FalWebhookBody, type GeneratedImage, IMAGE_INPUT_SPEC_KEYS, INPUT_SPEC_SUPPORTS_KEYS, type ImageBytes, type ImageCallOptions, type ImageCallResult, type ImageQuality, type ImageRequest, type ImageResult, type ImageSafetySetting, type ImageSize, type InputSpec, type InputSpecDimensions, type InputSpecIssue, type InputSpecKey, InputSpecSchema, type JsonObject, JsonObjectSchema, type JsonPrimitive, type JsonValue, JsonValueSchema, type KeyTier, type LanguageModelHandle, type LanguageOption, type Lifecycle, type LipsyncRequest, type LipsyncResult, type LiveCatalogState, MODEL3D_IMAGE_SLOTS, MODEL3D_MODELS, MODELS, type MaskBox, type MaskPoint, type MediaJobCallContext, type MediaNodeDefaults, type MediaOutput, type MediaPart, type MediaRef, type Model3dCall, type Model3dConstraints, type Model3dEndpoints, type Model3dFamily, type Model3dInputRefusal, type Model3dPricedInputs, type Model3dPricedOptions, type Model3dRequest, type Model3dResult, type Model3dStatus, type Model3dSubmission, type Model3dTexture, type Model3dViews, type Model3dWired, type ModelConstraints, type ModelKit, type ModelKitConfig, ModelKitError, type ModelKitErrorCode, type ModelKitErrorInit, type ModelType, type ModelkitGenerateObjectOptions, type ModelkitGenerateTextOptions, type ModelkitStreamObjectOptions, type ModelkitStreamTextOptions, type ObjectOutput, type ObjectResult, type OnUnsupported, type OutputFormat, type PricingKind, type PrimeCatalogResult, type PromptConventions, type ProviderErrorBody, type ProviderKeyName, type ProviderKeys, type ProviderOptionValue, type ProviderOptions, type QuarantineResult, type ReferenceLimits, type ReferenceSyntax, type RequiredInput, type RequiredKeyName, type ResolveResult, type Routing, SURFACES, type SpeechRequest, type SubtitleOptions, type SubtitleTextStyle, type SupportedOption, type Surface, SurfaceSchema, type TextMethods, type TextModelInstance, type TranscriptWord, type TranscriptionRequest, type TranscriptionResult, type UpscaleOptions, type UsageEvent, type UsageModality, type UsageUnit, VIDEO_INPUT_SPEC_KEYS, VIDEO_TOOL_MODELS, type VideoCallJob, type VideoCallOptions, type VideoJob, type VideoModel, type VideoRequest, type VideoResolution, VideoResolutionSchema, type VideoResult, type VideoSecondsLimits, type VideoToolCallContext, type VideoToolConstraints, type VideoToolKind, type VideoToolOutputFormat, type VideoToolOutputSpec, type VideoToolPlan, type VideoToolRequest, type VideoToolResult, type VideoToolSource, type VideoToolStatus, type VideoToolSubmission, type VoiceCandidate, type VoiceChangeRequest, type VoiceCloneRequest, type VoiceCloneResult, type VoiceCloneState, type VoiceDeleteRequest, type VoiceDesignRequest, type VoiceDesignResult, type VoiceGender, type VoiceOption, type VoiceSample, type Warning, adapterServesType, aiSdkTextAdapter, arkAdapter, aspectRatioToSize, classifyHttpError, compareCatalogOrder, createDefaultModelKit, createModelKit, dataOnlyCapable, describe, describeThrown, elevenLabsAdapter, generateImage as experimental_generateImage, falAdapter, falWebhookBodySchema, fileCatalogCache, fishAdapter, generateAudio, generateImage, generateObject, generateText, generateVideo, googleGenaiAdapter, googleImageAdapter, hasAwsCredentials, higgsfieldAdapter, isArrayInputSpecKey, isInputSpecKey, isListed, isListedOn, isModelKitError, keysFromEnv, languageModel, liveCatalog, liveState, lumaAdapter, lumaReframeTooTall, mergeTextProviderOptions, missingTierKeys, model3dCall, model3dConstraints, model3dInputRefusal, model3dPricingSku, model3dResolution, model3dSeedFits, model3dSkus, model3dTexture, nearestAspectRatio, normalizeExposure, openaiImageAdapter, passthroughTextModel, planVideoTool, pollUntil, prepareMediaPart, primeCatalog, prodiaAdapter, readErrorBody, replicateAdapter, resetLiveCatalog, resolutionTier, resolveAudioOptions, resolveDubbingOptions, resolveImageOptions, resolveKeys, resolveLipsyncOptions, resolveSpeechOptions, resolveTranscriptionOptions, resolveVideoOptions, resolveVoiceChangeOptions, resolveVoiceCloneOptions, resolveVoiceDesignOptions, shouldRetry, sleep, sortCatalogRows, streamObject, streamText, takesTargetAspectRatio, textUsageEvent, toGeneratedImage, validateInputSpec, videoLengthViolation, videoToolConstraints, withRetry, xaiVideoAdapter };
2296
+ export { ADAPTER_IDS, ADAPTER_KEYS, ARRAY_INPUT_SPEC_KEYS, AUDIO_INPUT_SPEC_KEYS, type Adapter, type AdapterContext, type AdapterEndpoints, type AdapterId, type AdapterRegistry, type AiTextProviderOptions, type ArrayInputSpecKey, type AspectRatio, type AudioNodeMethods, type AudioOutput, type AudioRequest, type AudioResult, type CallContext, Catalog, type CatalogAuth, type CatalogCache, type CatalogCapabilities, type CatalogDisplay, type CatalogEnvelope, type CatalogModel, CatalogModelSchema, CatalogModelSchemaBase, type CatalogOrigin, CatalogSchema, type CatalogSource, type CatalogTextModel, DEFAULT_ADAPTERS, type DataOnlyRow, type DataOnlyVerdict, type DefaultObjectOutput, type DubbingOutcome, type DubbingRequest, type DubbingResult, type Exposure, ExposureSchema, type FalWebhookBody, type GeneratedImage, IMAGE_INPUT_SPEC_KEYS, INPUT_SPEC_SUPPORTS_KEYS, type ImageBytes, type ImageCallOptions, type ImageCallResult, type ImageQuality, type ImageRequest, type ImageResult, type ImageSafetySetting, type ImageSize, type InputSpec, type InputSpecDimensions, type InputSpecIssue, type InputSpecKey, InputSpecSchema, type JsonObject, JsonObjectSchema, type JsonPrimitive, type JsonValue, JsonValueSchema, type KeyTier, type LanguageModelHandle, type LanguageOption, type Lifecycle, type LipsyncRequest, type LipsyncResult, type LiveCatalogState, MODEL3D_IMAGE_SLOTS, MODEL3D_MODELS, MODELS, type MaskBox, type MaskPoint, type MediaJobCallContext, type MediaNodeDefaults, type MediaOutput, type MediaPart, type MediaRef, type Model3dCall, type Model3dConstraints, type Model3dEndpoints, type Model3dFamily, type Model3dInputRefusal, type Model3dPricedInputs, type Model3dPricedOptions, type Model3dRequest, type Model3dResult, type Model3dStatus, type Model3dSubmission, type Model3dTexture, type Model3dViews, type Model3dWired, type ModelConstraints, type ModelKit, type ModelKitConfig, ModelKitError, type ModelKitErrorCode, type ModelKitErrorInit, type ModelType, type ModelkitGenerateObjectOptions, type ModelkitGenerateTextOptions, type ModelkitStreamObjectOptions, type ModelkitStreamTextOptions, type ObjectOutput, type ObjectResult, type OnUnsupported, type OutputFormat, type PricingKind, type PrimeCatalogResult, type PromptConventions, type ProviderErrorBody, type ProviderKeyName, type ProviderKeys, type ProviderOptionValue, type ProviderOptions, type QuarantineResult, type ReferenceLimits, type ReferenceSyntax, type RequiredInput, type RequiredKeyName, type ResolveResult, type Routing, SURFACES, type SpeechRequest, type SubtitleOptions, type SubtitleTextStyle, type SupportedOption, type Surface, SurfaceSchema, type TextMethods, type TextModelInstance, type TranscriptWord, type TranscriptionRequest, type TranscriptionResult, type UpscaleOptions, type UsageEvent, type UsageModality, type UsageUnit, VIDEO_INPUT_SPEC_KEYS, VIDEO_TOOL_MODELS, type VideoCallJob, type VideoCallOptions, type VideoJob, type VideoModel, type VideoPixelLimits, type VideoRequest, type VideoResolution, VideoResolutionSchema, type VideoResult, type VideoSecondsLimits, type VideoToolCallContext, type VideoToolConstraints, type VideoToolKind, type VideoToolOutputFormat, type VideoToolOutputSpec, type VideoToolPlan, type VideoToolRequest, type VideoToolResult, type VideoToolSource, type VideoToolStatus, type VideoToolSubmission, type VoiceCandidate, type VoiceChangeRequest, type VoiceCloneRequest, type VoiceCloneResult, type VoiceCloneState, type VoiceDeleteRequest, type VoiceDesignRequest, type VoiceDesignResult, type VoiceGender, type VoiceOption, type VoiceSample, type Warning, adapterServesType, aiSdkTextAdapter, arkAdapter, aspectRatioToSize, classifyHttpError, compareCatalogOrder, createDefaultModelKit, createModelKit, dataOnlyCapable, describe, describeThrown, elevenLabsAdapter, generateImage as experimental_generateImage, falAdapter, falWebhookBodySchema, fileCatalogCache, fishAdapter, generateAudio, generateImage, generateObject, generateText, generateVideo, googleGenaiAdapter, googleImageAdapter, hasAwsCredentials, higgsfieldAdapter, isArrayInputSpecKey, isInputSpecKey, isListed, isListedOn, isModelKitError, keysFromEnv, languageModel, liveCatalog, liveState, lumaAdapter, lumaReframeTooTall, mergeTextProviderOptions, missingTierKeys, model3dCall, model3dConstraints, model3dInputRefusal, model3dPricingSku, model3dResolution, model3dSeedFits, model3dSkus, model3dTexture, nearestAspectRatio, normalizeExposure, openaiImageAdapter, passthroughTextModel, planVideoTool, pollUntil, prepareMediaPart, primeCatalog, prodiaAdapter, readErrorBody, replicateAdapter, resetLiveCatalog, resolutionTier, resolveAudioOptions, resolveDubbingOptions, resolveImageOptions, resolveKeys, resolveLipsyncOptions, resolveSpeechOptions, resolveTranscriptionOptions, resolveVideoOptions, resolveVoiceChangeOptions, resolveVoiceCloneOptions, resolveVoiceDesignOptions, shouldRetry, sleep, sortCatalogRows, streamObject, streamText, takesTargetAspectRatio, textUsageEvent, toGeneratedImage, validateInputSpec, videoLengthViolation, videoSizeViolation, videoToolConstraints, withRetry, xaiVideoAdapter };
package/dist/index.d.ts CHANGED
@@ -601,6 +601,11 @@ type SupportedOption = "seed" | "quality" | "negativePrompt" | "cfgScale" | "gen
601
601
  | "insertWindow"
602
602
  /** Mask rows: `ImageRequest.points` and `ImageRequest.boxes`. */
603
603
  | "points" | "boxes";
604
+ /** Smallest frame a provider takes for an input video, in pixels (fal: "Min width: 700px"). */
605
+ interface VideoPixelLimits {
606
+ minWidth?: number;
607
+ minHeight?: number;
608
+ }
604
609
  interface VideoSecondsLimits {
605
610
  /** Minimum length of each reference video, in seconds. */
606
611
  min?: number;
@@ -622,6 +627,8 @@ interface ReferenceLimits {
622
627
  audioOnly?: boolean;
623
628
  /** Per-video length caps (min/max each) and combined total for reference videos. */
624
629
  videoSeconds?: VideoSecondsLimits;
630
+ /** Smallest frame each reference video may have (Kling 4.0: 700 px). Kits before 0.15.0 ignore it. */
631
+ videoPixels?: VideoPixelLimits;
625
632
  /** Per-clip length caps (min/max each) and combined total for reference audio, in seconds. */
626
633
  audioSeconds?: {
627
634
  min?: number;
@@ -751,6 +758,8 @@ interface ModelConstraints {
751
758
  languages?: LanguageOption[];
752
759
  /** Audio-output rows whose provider lets the caller pick a file format; `defaults.outputFormat` is one. */
753
760
  outputFormats?: string[];
761
+ /** Smallest frame the source video may have (Kling 4.0 Edit: 700 px). Kits before 0.15.0 ignore it. */
762
+ sourceVideoPixels?: VideoPixelLimits;
754
763
  /** Length limits for the source video the row edits or extends. */
755
764
  sourceVideoSeconds?: Omit<VideoSecondsLimits, "total"> & {
756
765
  /** The provider sends the source as a reference video, so it counts toward `references.videoSeconds.total` (Seedance). */
@@ -1253,6 +1262,10 @@ declare const CatalogModelSchemaBase: z.ZodObject<{
1253
1262
  total: z.ZodOptional<z.ZodNumber>;
1254
1263
  editMin: z.ZodOptional<z.ZodNumber>;
1255
1264
  }, z.core.$loose>>;
1265
+ videoPixels: z.ZodOptional<z.ZodObject<{
1266
+ minWidth: z.ZodOptional<z.ZodNumber>;
1267
+ minHeight: z.ZodOptional<z.ZodNumber>;
1268
+ }, z.core.$loose>>;
1256
1269
  audioSeconds: z.ZodOptional<z.ZodObject<{
1257
1270
  min: z.ZodOptional<z.ZodNumber>;
1258
1271
  max: z.ZodOptional<z.ZodNumber>;
@@ -1330,6 +1343,10 @@ declare const CatalogModelSchemaBase: z.ZodObject<{
1330
1343
  name: z.ZodString;
1331
1344
  }, z.core.$loose>>>;
1332
1345
  outputFormats: z.ZodOptional<z.ZodArray<z.ZodString>>;
1346
+ sourceVideoPixels: z.ZodOptional<z.ZodObject<{
1347
+ minWidth: z.ZodOptional<z.ZodNumber>;
1348
+ minHeight: z.ZodOptional<z.ZodNumber>;
1349
+ }, z.core.$loose>>;
1333
1350
  sourceVideoSeconds: z.ZodOptional<z.ZodObject<{
1334
1351
  max: z.ZodOptional<z.ZodNumber>;
1335
1352
  min: z.ZodOptional<z.ZodNumber>;
@@ -1858,6 +1875,19 @@ declare function videoLengthViolation(model: CatalogModel, input: {
1858
1875
  sourceVideo?: number;
1859
1876
  taskType?: VideoRequest["taskType"];
1860
1877
  }): string | undefined;
1878
+ type MeasuredFrame = {
1879
+ width?: number;
1880
+ height?: number;
1881
+ } | undefined;
1882
+ /**
1883
+ * First frame-size violation for the videos a request sends, as a user-facing message, from
1884
+ * their measured pixel sizes (`references.videoPixels`, `sourceVideoPixels`). An unmeasured
1885
+ * video passes: the provider still has the last word.
1886
+ */
1887
+ declare function videoSizeViolation(model: CatalogModel, input: {
1888
+ referenceVideos?: MeasuredFrame[];
1889
+ sourceVideo?: MeasuredFrame;
1890
+ }): string | undefined;
1861
1891
  declare function resolveImageOptions(model: CatalogModel, request: ImageRequest, onUnsupported?: OnUnsupported): ResolveResult<ImageRequest>;
1862
1892
  declare function resolveVideoOptions(model: CatalogModel, request: VideoRequest, onUnsupported?: OnUnsupported): ResolveResult<VideoRequest>;
1863
1893
  declare function resolveAudioOptions(model: CatalogModel, request: AudioRequest, onUnsupported?: OnUnsupported): ResolveResult<AudioRequest>;
@@ -2263,4 +2293,4 @@ declare function describeThrown(cause: unknown): ThrownErrorInfo;
2263
2293
  /** Add the AI SDK `GeneratedFile` accessors to adapter image bytes. base64 is computed on first read. */
2264
2294
  declare function toGeneratedImage(image: ImageBytes): GeneratedImage;
2265
2295
 
2266
- export { ADAPTER_IDS, ADAPTER_KEYS, ARRAY_INPUT_SPEC_KEYS, AUDIO_INPUT_SPEC_KEYS, type Adapter, type AdapterContext, type AdapterEndpoints, type AdapterId, type AdapterRegistry, type AiTextProviderOptions, type ArrayInputSpecKey, type AspectRatio, type AudioNodeMethods, type AudioOutput, type AudioRequest, type AudioResult, type CallContext, Catalog, type CatalogAuth, type CatalogCache, type CatalogCapabilities, type CatalogDisplay, type CatalogEnvelope, type CatalogModel, CatalogModelSchema, CatalogModelSchemaBase, type CatalogOrigin, CatalogSchema, type CatalogSource, type CatalogTextModel, DEFAULT_ADAPTERS, type DataOnlyRow, type DataOnlyVerdict, type DefaultObjectOutput, type DubbingOutcome, type DubbingRequest, type DubbingResult, type Exposure, ExposureSchema, type FalWebhookBody, type GeneratedImage, IMAGE_INPUT_SPEC_KEYS, INPUT_SPEC_SUPPORTS_KEYS, type ImageBytes, type ImageCallOptions, type ImageCallResult, type ImageQuality, type ImageRequest, type ImageResult, type ImageSafetySetting, type ImageSize, type InputSpec, type InputSpecDimensions, type InputSpecIssue, type InputSpecKey, InputSpecSchema, type JsonObject, JsonObjectSchema, type JsonPrimitive, type JsonValue, JsonValueSchema, type KeyTier, type LanguageModelHandle, type LanguageOption, type Lifecycle, type LipsyncRequest, type LipsyncResult, type LiveCatalogState, MODEL3D_IMAGE_SLOTS, MODEL3D_MODELS, MODELS, type MaskBox, type MaskPoint, type MediaJobCallContext, type MediaNodeDefaults, type MediaOutput, type MediaPart, type MediaRef, type Model3dCall, type Model3dConstraints, type Model3dEndpoints, type Model3dFamily, type Model3dInputRefusal, type Model3dPricedInputs, type Model3dPricedOptions, type Model3dRequest, type Model3dResult, type Model3dStatus, type Model3dSubmission, type Model3dTexture, type Model3dViews, type Model3dWired, type ModelConstraints, type ModelKit, type ModelKitConfig, ModelKitError, type ModelKitErrorCode, type ModelKitErrorInit, type ModelType, type ModelkitGenerateObjectOptions, type ModelkitGenerateTextOptions, type ModelkitStreamObjectOptions, type ModelkitStreamTextOptions, type ObjectOutput, type ObjectResult, type OnUnsupported, type OutputFormat, type PricingKind, type PrimeCatalogResult, type PromptConventions, type ProviderErrorBody, type ProviderKeyName, type ProviderKeys, type ProviderOptionValue, type ProviderOptions, type QuarantineResult, type ReferenceLimits, type ReferenceSyntax, type RequiredInput, type RequiredKeyName, type ResolveResult, type Routing, SURFACES, type SpeechRequest, type SubtitleOptions, type SubtitleTextStyle, type SupportedOption, type Surface, SurfaceSchema, type TextMethods, type TextModelInstance, type TranscriptWord, type TranscriptionRequest, type TranscriptionResult, type UpscaleOptions, type UsageEvent, type UsageModality, type UsageUnit, VIDEO_INPUT_SPEC_KEYS, VIDEO_TOOL_MODELS, type VideoCallJob, type VideoCallOptions, type VideoJob, type VideoModel, type VideoRequest, type VideoResolution, VideoResolutionSchema, type VideoResult, type VideoSecondsLimits, type VideoToolCallContext, type VideoToolConstraints, type VideoToolKind, type VideoToolOutputFormat, type VideoToolOutputSpec, type VideoToolPlan, type VideoToolRequest, type VideoToolResult, type VideoToolSource, type VideoToolStatus, type VideoToolSubmission, type VoiceCandidate, type VoiceChangeRequest, type VoiceCloneRequest, type VoiceCloneResult, type VoiceCloneState, type VoiceDeleteRequest, type VoiceDesignRequest, type VoiceDesignResult, type VoiceGender, type VoiceOption, type VoiceSample, type Warning, adapterServesType, aiSdkTextAdapter, arkAdapter, aspectRatioToSize, classifyHttpError, compareCatalogOrder, createDefaultModelKit, createModelKit, dataOnlyCapable, describe, describeThrown, elevenLabsAdapter, generateImage as experimental_generateImage, falAdapter, falWebhookBodySchema, fileCatalogCache, fishAdapter, generateAudio, generateImage, generateObject, generateText, generateVideo, googleGenaiAdapter, googleImageAdapter, hasAwsCredentials, higgsfieldAdapter, isArrayInputSpecKey, isInputSpecKey, isListed, isListedOn, isModelKitError, keysFromEnv, languageModel, liveCatalog, liveState, lumaAdapter, lumaReframeTooTall, mergeTextProviderOptions, missingTierKeys, model3dCall, model3dConstraints, model3dInputRefusal, model3dPricingSku, model3dResolution, model3dSeedFits, model3dSkus, model3dTexture, nearestAspectRatio, normalizeExposure, openaiImageAdapter, passthroughTextModel, planVideoTool, pollUntil, prepareMediaPart, primeCatalog, prodiaAdapter, readErrorBody, replicateAdapter, resetLiveCatalog, resolutionTier, resolveAudioOptions, resolveDubbingOptions, resolveImageOptions, resolveKeys, resolveLipsyncOptions, resolveSpeechOptions, resolveTranscriptionOptions, resolveVideoOptions, resolveVoiceChangeOptions, resolveVoiceCloneOptions, resolveVoiceDesignOptions, shouldRetry, sleep, sortCatalogRows, streamObject, streamText, takesTargetAspectRatio, textUsageEvent, toGeneratedImage, validateInputSpec, videoLengthViolation, videoToolConstraints, withRetry, xaiVideoAdapter };
2296
+ export { ADAPTER_IDS, ADAPTER_KEYS, ARRAY_INPUT_SPEC_KEYS, AUDIO_INPUT_SPEC_KEYS, type Adapter, type AdapterContext, type AdapterEndpoints, type AdapterId, type AdapterRegistry, type AiTextProviderOptions, type ArrayInputSpecKey, type AspectRatio, type AudioNodeMethods, type AudioOutput, type AudioRequest, type AudioResult, type CallContext, Catalog, type CatalogAuth, type CatalogCache, type CatalogCapabilities, type CatalogDisplay, type CatalogEnvelope, type CatalogModel, CatalogModelSchema, CatalogModelSchemaBase, type CatalogOrigin, CatalogSchema, type CatalogSource, type CatalogTextModel, DEFAULT_ADAPTERS, type DataOnlyRow, type DataOnlyVerdict, type DefaultObjectOutput, type DubbingOutcome, type DubbingRequest, type DubbingResult, type Exposure, ExposureSchema, type FalWebhookBody, type GeneratedImage, IMAGE_INPUT_SPEC_KEYS, INPUT_SPEC_SUPPORTS_KEYS, type ImageBytes, type ImageCallOptions, type ImageCallResult, type ImageQuality, type ImageRequest, type ImageResult, type ImageSafetySetting, type ImageSize, type InputSpec, type InputSpecDimensions, type InputSpecIssue, type InputSpecKey, InputSpecSchema, type JsonObject, JsonObjectSchema, type JsonPrimitive, type JsonValue, JsonValueSchema, type KeyTier, type LanguageModelHandle, type LanguageOption, type Lifecycle, type LipsyncRequest, type LipsyncResult, type LiveCatalogState, MODEL3D_IMAGE_SLOTS, MODEL3D_MODELS, MODELS, type MaskBox, type MaskPoint, type MediaJobCallContext, type MediaNodeDefaults, type MediaOutput, type MediaPart, type MediaRef, type Model3dCall, type Model3dConstraints, type Model3dEndpoints, type Model3dFamily, type Model3dInputRefusal, type Model3dPricedInputs, type Model3dPricedOptions, type Model3dRequest, type Model3dResult, type Model3dStatus, type Model3dSubmission, type Model3dTexture, type Model3dViews, type Model3dWired, type ModelConstraints, type ModelKit, type ModelKitConfig, ModelKitError, type ModelKitErrorCode, type ModelKitErrorInit, type ModelType, type ModelkitGenerateObjectOptions, type ModelkitGenerateTextOptions, type ModelkitStreamObjectOptions, type ModelkitStreamTextOptions, type ObjectOutput, type ObjectResult, type OnUnsupported, type OutputFormat, type PricingKind, type PrimeCatalogResult, type PromptConventions, type ProviderErrorBody, type ProviderKeyName, type ProviderKeys, type ProviderOptionValue, type ProviderOptions, type QuarantineResult, type ReferenceLimits, type ReferenceSyntax, type RequiredInput, type RequiredKeyName, type ResolveResult, type Routing, SURFACES, type SpeechRequest, type SubtitleOptions, type SubtitleTextStyle, type SupportedOption, type Surface, SurfaceSchema, type TextMethods, type TextModelInstance, type TranscriptWord, type TranscriptionRequest, type TranscriptionResult, type UpscaleOptions, type UsageEvent, type UsageModality, type UsageUnit, VIDEO_INPUT_SPEC_KEYS, VIDEO_TOOL_MODELS, type VideoCallJob, type VideoCallOptions, type VideoJob, type VideoModel, type VideoPixelLimits, type VideoRequest, type VideoResolution, VideoResolutionSchema, type VideoResult, type VideoSecondsLimits, type VideoToolCallContext, type VideoToolConstraints, type VideoToolKind, type VideoToolOutputFormat, type VideoToolOutputSpec, type VideoToolPlan, type VideoToolRequest, type VideoToolResult, type VideoToolSource, type VideoToolStatus, type VideoToolSubmission, type VoiceCandidate, type VoiceChangeRequest, type VoiceCloneRequest, type VoiceCloneResult, type VoiceCloneState, type VoiceDeleteRequest, type VoiceDesignRequest, type VoiceDesignResult, type VoiceGender, type VoiceOption, type VoiceSample, type Warning, adapterServesType, aiSdkTextAdapter, arkAdapter, aspectRatioToSize, classifyHttpError, compareCatalogOrder, createDefaultModelKit, createModelKit, dataOnlyCapable, describe, describeThrown, elevenLabsAdapter, generateImage as experimental_generateImage, falAdapter, falWebhookBodySchema, fileCatalogCache, fishAdapter, generateAudio, generateImage, generateObject, generateText, generateVideo, googleGenaiAdapter, googleImageAdapter, hasAwsCredentials, higgsfieldAdapter, isArrayInputSpecKey, isInputSpecKey, isListed, isListedOn, isModelKitError, keysFromEnv, languageModel, liveCatalog, liveState, lumaAdapter, lumaReframeTooTall, mergeTextProviderOptions, missingTierKeys, model3dCall, model3dConstraints, model3dInputRefusal, model3dPricingSku, model3dResolution, model3dSeedFits, model3dSkus, model3dTexture, nearestAspectRatio, normalizeExposure, openaiImageAdapter, passthroughTextModel, planVideoTool, pollUntil, prepareMediaPart, primeCatalog, prodiaAdapter, readErrorBody, replicateAdapter, resetLiveCatalog, resolutionTier, resolveAudioOptions, resolveDubbingOptions, resolveImageOptions, resolveKeys, resolveLipsyncOptions, resolveSpeechOptions, resolveTranscriptionOptions, resolveVideoOptions, resolveVoiceChangeOptions, resolveVoiceCloneOptions, resolveVoiceDesignOptions, shouldRetry, sleep, sortCatalogRows, streamObject, streamText, takesTargetAspectRatio, textUsageEvent, toGeneratedImage, validateInputSpec, videoLengthViolation, videoSizeViolation, videoToolConstraints, withRetry, xaiVideoAdapter };
package/dist/index.js CHANGED
@@ -5093,6 +5093,7 @@ function klingMotionControl(tier) {
5093
5093
  videos: 1,
5094
5094
  audios: 0,
5095
5095
  videoSeconds: { min: 3, max: 30.05 },
5096
+ videoPixels: { minWidth: 340, minHeight: 340 },
5096
5097
  min: { images: 1, videos: 1 }
5097
5098
  },
5098
5099
  requires: ["references"],
@@ -5159,6 +5160,7 @@ var VIDEO_FAMILY_MODELS = [
5159
5160
  references: { ...NO_REFERENCES, images: 4 },
5160
5161
  requires: ["sourceVideo"],
5161
5162
  sourceVideoSeconds: { min: 3, max: 15.05 },
5163
+ sourceVideoPixels: { minWidth: 720, minHeight: 720 },
5162
5164
  // generateAudio keeps (true) or drops (false) the source soundtrack.
5163
5165
  supports: ["generateAudio"],
5164
5166
  prompt: KLING_PROMPT,
@@ -5199,7 +5201,8 @@ var VIDEO_FAMILY_MODELS = [
5199
5201
  ...NO_REFERENCES,
5200
5202
  images: 10,
5201
5203
  videos: 5,
5202
- videoSeconds: { min: 3, max: 30.05, total: 30 }
5204
+ videoSeconds: { min: 3, max: 30.05, total: 30 },
5205
+ videoPixels: { minWidth: 700, minHeight: 700 }
5203
5206
  },
5204
5207
  supports: ["lastFrame"],
5205
5208
  prompt: KLING_V4_PROMPT,
@@ -5233,6 +5236,7 @@ var VIDEO_FAMILY_MODELS = [
5233
5236
  references: { ...NO_REFERENCES, images: 10 },
5234
5237
  requires: ["sourceVideo"],
5235
5238
  sourceVideoSeconds: { min: 3, max: 30.05 },
5239
+ sourceVideoPixels: { minWidth: 700, minHeight: 700 },
5236
5240
  supports: [],
5237
5241
  prompt: KLING_V4_PROMPT,
5238
5242
  audio: KLING_V4_AUDIO
@@ -7777,6 +7781,10 @@ var VideoSecondsLimitsSchema = z3.looseObject({
7777
7781
  total: z3.number().optional(),
7778
7782
  editMin: z3.number().optional()
7779
7783
  });
7784
+ var VideoPixelLimitsSchema = z3.looseObject({
7785
+ minWidth: z3.number().int().positive().optional(),
7786
+ minHeight: z3.number().int().positive().optional()
7787
+ });
7780
7788
  var ReferenceLimitsSchema = z3.looseObject({
7781
7789
  images: z3.number().optional(),
7782
7790
  imagesWithVideo: z3.number().optional(),
@@ -7785,6 +7793,7 @@ var ReferenceLimitsSchema = z3.looseObject({
7785
7793
  total: z3.number().optional(),
7786
7794
  audioOnly: z3.boolean().optional(),
7787
7795
  videoSeconds: VideoSecondsLimitsSchema.optional(),
7796
+ videoPixels: VideoPixelLimitsSchema.optional(),
7788
7797
  audioSeconds: z3.looseObject({
7789
7798
  min: z3.number().optional(),
7790
7799
  max: z3.number().optional(),
@@ -7874,6 +7883,7 @@ var ModelConstraintsSchema = z3.looseObject({
7874
7883
  }).optional(),
7875
7884
  languages: z3.array(LanguageOptionSchema).optional(),
7876
7885
  outputFormats: z3.array(z3.string().min(1)).min(1).optional(),
7886
+ sourceVideoPixels: VideoPixelLimitsSchema.optional(),
7877
7887
  sourceVideoSeconds: VideoSecondsLimitsSchema.omit({ total: true }).extend({ inReferenceTotal: z3.boolean().optional() }).optional(),
7878
7888
  // segmentSeconds/maxTotalSeconds stay required: a kit before 0.3.0 drops a row whose
7879
7889
  // extension lacks them, so every extension block keeps them.
@@ -17616,6 +17626,28 @@ function videoLengthViolation(model, input) {
17616
17626
  return `${model.label} takes a source video of at least ${String(sourceMin)} s; this one is ${seconds2(source2, sourceMin)} s.`;
17617
17627
  return void 0;
17618
17628
  }
17629
+ function undersizedSide(frame, limits) {
17630
+ if (!frame || !limits) return void 0;
17631
+ const { width, height } = frame;
17632
+ if (limits.minWidth !== void 0 && width !== void 0 && width > 0 && width < limits.minWidth)
17633
+ return { side: "wide", min: limits.minWidth, actual: width };
17634
+ if (limits.minHeight !== void 0 && height !== void 0 && height > 0 && height < limits.minHeight)
17635
+ return { side: "tall", min: limits.minHeight, actual: height };
17636
+ return void 0;
17637
+ }
17638
+ function videoSizeViolation(model, input) {
17639
+ const refLimits = model.constraints.references?.videoPixels;
17640
+ const refs = input.referenceVideos ?? [];
17641
+ for (let i = 0; i < refs.length; i++) {
17642
+ const under2 = undersizedSide(refs[i], refLimits);
17643
+ if (under2)
17644
+ return `${model.label} takes reference videos at least ${String(under2.min)} px ${under2.side}; video ${i + 1} is ${String(under2.actual)} px.`;
17645
+ }
17646
+ const under = undersizedSide(input.sourceVideo, model.constraints.sourceVideoPixels);
17647
+ if (under)
17648
+ return `${model.label} takes a source video at least ${String(under.min)} px ${under.side}; this one is ${String(under.actual)} px.`;
17649
+ return void 0;
17650
+ }
17619
17651
  function resolveImageOptions(model, request, onUnsupported = "error") {
17620
17652
  const warnings = [];
17621
17653
  const c = model.constraints;
@@ -19735,6 +19767,7 @@ export {
19735
19767
  toGeneratedImage,
19736
19768
  validateInputSpec,
19737
19769
  videoLengthViolation,
19770
+ videoSizeViolation,
19738
19771
  videoToolConstraints,
19739
19772
  withRetry,
19740
19773
  xaiVideoAdapter