diff --git a/app/api/enhance/pro/route.ts b/app/api/enhance/pro/route.ts index 2cbc28596..c34986a1f 100644 --- a/app/api/enhance/pro/route.ts +++ b/app/api/enhance/pro/route.ts @@ -328,7 +328,6 @@ async function createEnhanceProJob(params: { width: params.width, height: params.height, image_url: params.sourceImageUrl, - batchSize: params.requestBody.batchSize, seed: params.requestBody.seed, prompt: String(params.requestBody.prompt || "").trim(), output_format: params.requestBody.output_format || params.config.defaultOutputFormat, @@ -1065,9 +1064,8 @@ export async function POST(request: NextRequest) { config.maxOutputEdge, ); - const batchSize = config.allowBatchSize - ? clampInteger(body.batchSize, 1, config.maxBatchSize, 1) - : 1; + // Krea removed batch generation from the enhance endpoint. + const batchSize = 1; const upscalingActivated = body.upscaling_activated === true; const imageScalingFactor = clampNumber( body.image_scaling_factor, diff --git a/lib/ai/krea-models.ts b/lib/ai/krea-models.ts index 4b4cc10d3..1eb559900 100644 --- a/lib/ai/krea-models.ts +++ b/lib/ai/krea-models.ts @@ -27,23 +27,33 @@ const REFERENCE_INPUT_FIELDS = new Set([ "imageUrl", "imageUrls", "image_url", + "image_urls", "styleImages", + "style_images", "imageStyleRefs", + "image_style_references", "characterReferenceImages", + "character_reference_images", "backgroundImage", "referenceImages", + "reference_images", "startImage", "endImage", "init_image_url", "references", "subjects", "styleReferences", + "style_references", "elementReferences", + "element_references", "videoReference", + "video_reference", "startVideo", "init_video", "referenceVideos", + "reference_videos", "referenceAudios", + "reference_audios", ]); function hasReferenceInput(fields: string[]) { @@ -68,12 +78,15 @@ function createDefaultParams( function createImageConfig(fields: string[]) { return { - maxImages: fields.includes("batchSize") ? 4 : 1, + // Krea removed batch generation; one request now produces one output. + maxImages: 1, supportsSize: fields.includes("width") && fields.includes("height"), supports1K: true, supports2K: true, supports4K: fields.includes("resolution") || fields.includes("width"), - supportsPromptExtend: fields.includes("skipPromptExpansion"), + supportsPromptExtend: + fields.includes("skipPromptExpansion") || + fields.includes("skip_prompt_expansion"), supportsNegativePrompt: fields.includes("negative_prompt"), supportsWatermark: false, supportsSeed: fields.includes("seed"), @@ -233,62 +246,65 @@ function videoPreset( } export const KREA_MODEL_PRESETS: KreaModelPreset[] = [ - imagePreset(10, "/generate/image/bfl/flux-1-dev", "Flux", ["strength", "seed", "steps", "styleImages", "imageStyleRefs", "prompt", "width", "height", "guidance_scale_flux", "imageUrl", "relaxedModeAccess", "styles"], ["prompt"]), - imagePreset(20, "/generate/image/bfl/flux-1-kontext-dev", "Flux Kontext", ["strength", "seed", "steps", "styleImages", "imageStyleRefs", "prompt", "width", "height", "guidance_scale_flux", "imageUrl", "relaxedModeAccess", "styles"], ["prompt"]), - imagePreset(30, "/generate/image/google/nano-banana-pro", "Nano Banana Pro", ["batchSize", "prompt", "width", "height", "imageUrls", "aspectRatio", "styleImages", "resolution"], ["prompt"]), - imagePreset(40, "/generate/image/google/nano-banana-2", "Nano Banana 2", ["batchSize", "prompt", "width", "height", "imageUrls", "aspectRatio", "styleImages", "resolution"], ["prompt"]), - imagePreset(50, "/generate/image/google/nano-banana", "Nano Banana", ["batchSize", "prompt", "width", "height", "imageUrls", "styleImages", "aspectRatio"], ["prompt"]), - imagePreset(60, "/generate/image/bfl/flux-1.1-pro", "Flux 1.1 Pro", ["batchSize", "prompt", "width", "height", "seed"], ["prompt", "width", "height"]), - imagePreset(70, "/generate/image/bfl/flux-1.1-pro-ultra", "Flux 1.1 Pro Ultra", ["batchSize", "prompt", "width", "height", "seed", "raw"], ["prompt"]), - imagePreset(80, "/generate/image/ideogram/ideogram-2-turbo", "Ideogram 2.0A Turbo", ["batchSize", "prompt", "width", "height", "seed"], ["prompt"]), - imagePreset(90, "/generate/image/ideogram/ideogram-3", "Ideogram 3.0", ["batchSize", "prompt", "width", "height", "seed", "styleImages", "characterReferenceImages"], ["prompt"]), - imagePreset(100, "/generate/image/google/imagen-3", "Imagen 3", ["batchSize", "prompt", "width", "height", "seed"], ["prompt"]), - imagePreset(110, "/generate/image/google/imagen-4", "Imagen 4", ["batchSize", "prompt", "width", "height", "seed"], ["prompt"]), - imagePreset(120, "/generate/image/google/imagen-4-fast", "Imagen 4 Fast", ["batchSize", "prompt", "width", "height", "seed"], ["prompt"]), - imagePreset(130, "/generate/image/google/imagen-4-ultra", "Imagen 4 Ultra", ["batchSize", "prompt", "width", "height", "seed"], ["prompt"]), - imagePreset(140, "/generate/image/runway/gen-4-image", "Runway Gen-4", ["batchSize", "prompt", "width", "height", "seed", "referenceImages"], ["prompt", "referenceImages"]), - imagePreset(150, "/generate/image/openai/gpt-image", "ChatGPT Image", ["batchSize", "prompt", "width", "height", "quality", "backgroundImage", "imageUrls", "styleImages", "styles"], ["prompt"]), - imagePreset(160, "/generate/image/bytedance/seedream-3", "Seedream 3", ["batchSize", "prompt", "width", "height", "seed", "model"], ["prompt", "model"]), - imagePreset(170, "/generate/image/bytedance/seedream-4", "Seedream 4", ["batchSize", "prompt", "seed", "width", "height", "styleImages"], ["prompt", "width", "height"]), - imagePreset(180, "/generate/image/bytedance/seedream-5-lite", "Seedream 5 Lite", ["batchSize", "prompt", "seed", "width", "height", "styleImages"], ["prompt", "width", "height"]), - imagePreset(190, "/generate/image/qwen/2512", "Qwen 2512", ["prompt", "negative_prompt", "width", "height", "num_inference_steps", "cfg_scale", "seed", "batchSize", "skipPromptExpansion", "styles"], ["prompt"]), - imagePreset(200, "/generate/image/z-image/z-image", "Z Image", ["prompt", "width", "height", "batchSize", "seed", "imageUrl", "denoising_strength", "styleImages", "skipPromptExpansion", "styles"], ["prompt", "width", "height"]), - imagePreset(210, "/generate/image/bytedance/seededit", "SeedEdit", ["batchSize", "prompt", "seed", "imageUrl"], ["prompt", "imageUrl"]), - enhancePreset(220, "/generate/enhance/topaz/generative-enhance", "Topaz Generative", ["batchSize", "width", "height", "seed", "prompt", "output_format", "subject_detection", "face_enhancement", "face_enhancement_creativity", "face_enhancement_strength", "crop_to_fill", "image_url", "model", "upscaling_activated", "image_scaling_factor", "creativity", "texture", "sharpen", "denoise", "detail"], ["width", "height", "image_url"]), - enhancePreset(230, "/generate/enhance/topaz/standard-enhance", "Topaz", ["batchSize", "width", "height", "seed", "prompt", "output_format", "subject_detection", "face_enhancement", "face_enhancement_creativity", "face_enhancement_strength", "crop_to_fill", "image_url", "model", "upscaling_activated", "image_scaling_factor", "sharpen", "denoise", "fix_compression", "strength"], ["width", "height", "image_url", "model"]), - enhancePreset(240, "/generate/enhance/topaz/bloom-enhance", "Topaz Bloom", ["batchSize", "width", "height", "seed", "prompt", "output_format", "crop_to_fill", "image_url", "model", "creativity", "face_preservation", "color_preservation", "upscaling_activated", "image_scaling_factor"], ["width", "height", "image_url", "model"]), - videoPreset(250, "/generate/video/kling/kling-1", "Kling 1.0", ["startImage", "prompt", "aspectRatio", "duration", "cameraControl", "endImage", "mode"], ["prompt"]), - videoPreset(260, "/generate/video/kling/kling-1.5", "Kling 1.5", ["startImage", "prompt", "aspectRatio", "duration", "cameraControl", "endImage"], ["prompt"]), - videoPreset(270, "/generate/video/kling/kling-1.6", "Kling 1.6", ["startImage", "prompt", "aspectRatio", "duration", "cameraControl", "endImage", "mode"], ["prompt"]), - videoPreset(280, "/generate/video/kling/kling-2", "Kling 2.0", ["startImage", "prompt", "aspectRatio", "endImage", "mode"], ["prompt"]), - videoPreset(290, "/generate/video/kling/kling-2.1", "Kling 2.1", ["startImage", "prompt", "aspectRatio", "duration", "endImage", "mode"], ["prompt"]), - videoPreset(300, "/generate/video/kling/kling-2.5", "Kling 2.5", ["startImage", "prompt", "aspectRatio", "duration", "endImage"], ["prompt"]), - videoPreset(310, "/generate/video/kling/kling-2.6", "Kling 2.6", ["startImage", "prompt", "aspectRatio", "duration", "endImage", "generateAudio"], ["prompt"]), - videoPreset(320, "/generate/video/kling/kling-o1", "Kling o1", ["startImage", "prompt", "aspectRatio", "duration", "endImage", "styleReferences", "elementReferences", "videoReference"], ["prompt"]), - videoPreset(330, "/generate/video/minimax/hailuo", "Hailuo", ["startImage", "prompt", "expandPrompt", "cameraControl", "model", "subjects"], ["prompt"]), - videoPreset(340, "/generate/video/minimax/hailuo-02", "Hailuo 02", ["startImage", "prompt", "expandPrompt", "endImage", "resolution", "duration"], ["prompt"]), - videoPreset(350, "/generate/video/minimax/hailuo-2.3", "Hailuo 2.3", ["startImage", "prompt", "expandPrompt", "endImage", "resolution", "duration"], ["prompt"]), - videoPreset(360, "/generate/video/minimax/hailuo-2.3-fast", "Hailuo 2.3 Fast", ["startImage", "prompt", "expandPrompt", "endImage", "resolution", "duration"], ["prompt"]), - videoPreset(370, "/generate/video/alibaba/wan-2.1", "Wan 2.1", ["startImage", "prompt", "width", "height", "endImage", "aspectRatio", "frame_num", "styles"], ["prompt", "width", "height"]), - videoPreset(380, "/generate/video/alibaba/wan-2.2", "Wan 2.2", ["startImage", "guide_scale", "prompt", "width", "height", "aspectRatio", "frame_num", "styles"], ["prompt", "width", "height"]), - videoPreset(390, "/generate/video/alibaba/wan-2.5", "Wan 2.5", ["startImage", "prompt", "aspectRatio", "resolution", "enablePromptExpansion", "duration", "seed", "audio_url", "generateAudio"], ["prompt"]), - videoPreset(400, "/generate/video/google/veo-2", "Veo 2", ["startImage", "prompt", "aspectRatio", "duration", "image_url"], ["prompt"]), - videoPreset(410, "/generate/video/google/veo-3", "Veo 3", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution"], ["prompt"]), - videoPreset(420, "/generate/video/google/veo-3-fast", "Veo 3 Fast", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution"], ["prompt"]), - videoPreset(430, "/generate/video/google/veo-3.1", "Veo 3.1", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution", "endImage", "referenceImages"], ["prompt"]), - videoPreset(440, "/generate/video/google/veo-3.1-fast", "Veo 3.1 Fast", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution", "endImage", "referenceImages"], ["prompt"]), - videoPreset(450, "/generate/video/google/veo-3.1-lite", "Veo 3.1 Lite", ["startImage", "prompt", "aspectRatio", "duration", "resolution", "endImage"], ["prompt"]), - videoPreset(460, "/generate/video/bytedance/seedance-1.0-lite", "Seedance Lite", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "duration", "resolution"], ["prompt"]), - videoPreset(470, "/generate/video/bytedance/seedance-1.0-pro", "Seedance Pro", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "duration", "resolution"], ["prompt"]), - videoPreset(480, "/generate/video/bytedance/seedance-1.0-pro-fast", "Seedance Pro Fast", ["startImage", "prompt", "aspectRatio", "duration", "resolution"], ["prompt"]), - videoPreset(490, "/generate/video/bytedance/seedance-2", "Seedance 2.0", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "referenceVideos", "referenceAudios", "duration", "generateAudio", "resolution", "seed"], ["prompt"]), - videoPreset(500, "/generate/video/bytedance/seedance-2-fast", "Seedance 2.0 Fast", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "referenceVideos", "referenceAudios", "duration", "generateAudio", "resolution", "seed"], ["prompt"]), - videoPreset(510, "/generate/video/openai/sora-2", "Sora 2", ["startImage", "prompt", "duration", "aspectRatio"], ["prompt"]), - videoPreset(520, "/generate/video/luma/ray-2", "Ray 2", ["startImage", "prompt", "aspectRatio", "loop", "width", "height"], ["prompt", "width", "height"]), - videoPreset(530, "/generate/video/runway/gen-3", "Runway Gen-3", ["startImage", "prompt", "aspectRatio", "duration", "seed"], ["prompt"]), - videoPreset(540, "/generate/video/runway/gen-4-video", "Runway Gen-4", ["startImage", "prompt", "aspectRatio", "duration", "seed"], ["prompt"]), - videoPreset(550, "/generate/video/runway/gen-4.5", "Runway Gen-4.5", ["startImage", "prompt", "aspectRatio", "duration", "seed"], ["prompt"]), - videoPreset(560, "/generate/video/runway/aleph", "Runway Aleph", ["startImage", "startVideo", "startVideoTrim", "endVideoTrim", "prompt", "init_video", "init_image_url", "references", "ratio", "seed", "contentModeration"], ["prompt", "init_video"]), + imagePreset(10, "/generate/image/krea/krea-2/medium", "Krea 2 Medium", ["prompt", "seed", "styles", "image_style_references", "aspect_ratio", "resolution", "creativity", "moodboards"], ["prompt", "aspect_ratio", "resolution"]), + imagePreset(20, "/generate/image/krea/krea-2/large", "Krea 2 Large", ["prompt", "seed", "styles", "image_style_references", "aspect_ratio", "resolution", "creativity", "moodboards"], ["prompt", "aspect_ratio", "resolution"]), + imagePreset(30, "/generate/image/bfl/flux-1-dev", "Flux", ["prompt", "seed", "steps", "width", "height", "image_url", "styles", "strength", "guidance_scale", "style_images", "image_style_references"], ["prompt"]), + imagePreset(40, "/generate/image/bfl/flux-1-kontext-dev", "Flux Kontext", ["prompt", "seed", "steps", "width", "height", "image_url", "strength", "guidance_scale", "style_images"], ["prompt"]), + imagePreset(50, "/generate/image/google/nano-banana-pro", "Nano Banana Pro", ["prompt", "width", "height", "image_urls", "aspect_ratio", "resolution", "style_images"], ["prompt"]), + imagePreset(60, "/generate/image/google/nano-banana-2", "Nano Banana 2", ["prompt", "width", "height", "image_urls", "aspect_ratio", "resolution", "style_images"], ["prompt"]), + imagePreset(70, "/generate/image/google/nano-banana", "Nano Banana", ["prompt", "width", "height", "image_urls", "aspect_ratio", "style_images"], ["prompt"]), + imagePreset(80, "/generate/image/bfl/flux-1.1-pro", "Flux 1.1 Pro", ["prompt", "width", "height", "seed"], ["prompt", "width", "height"]), + imagePreset(90, "/generate/image/bfl/flux-1.1-pro-ultra", "Flux 1.1 Pro Ultra", ["prompt", "width", "height", "seed", "raw"], ["prompt"]), + imagePreset(100, "/generate/image/ideogram/ideogram-2-turbo", "Ideogram 2.0A Turbo", ["prompt", "width", "height", "seed"], ["prompt"]), + imagePreset(110, "/generate/image/ideogram/ideogram-3", "Ideogram 3.0", ["prompt", "width", "height", "seed", "character_reference_images", "style_images"], ["prompt"]), + imagePreset(120, "/generate/image/google/imagen-3", "Imagen 3", ["prompt", "width", "height", "seed"], ["prompt"]), + imagePreset(130, "/generate/image/google/imagen-4", "Imagen 4", ["prompt", "width", "height", "seed"], ["prompt"]), + imagePreset(140, "/generate/image/google/imagen-4-fast", "Imagen 4 Fast", ["prompt", "width", "height", "seed"], ["prompt"]), + imagePreset(150, "/generate/image/google/imagen-4-ultra", "Imagen 4 Ultra", ["prompt", "width", "height", "seed"], ["prompt"]), + imagePreset(160, "/generate/image/runway/gen-4-image", "Runway Gen-4", ["prompt", "width", "height", "seed", "reference_images"], ["prompt", "reference_images"]), + imagePreset(170, "/generate/image/openai/gpt-image", "ChatGPT Image", ["prompt", "width", "height", "quality", "image_urls", "styles", "style_images"], ["prompt"]), + imagePreset(180, "/generate/image/openai/gpt-image-2", "ChatGPT Image 2", ["prompt", "quality", "image_urls", "width", "height", "aspect_ratio", "resolution", "style_images"], ["prompt"]), + imagePreset(190, "/generate/image/luma/uni-1", "Luma Uni 1", ["prompt", "width", "height", "mode", "style", "output_format", "web_search", "style_images"], ["prompt"]), + imagePreset(200, "/generate/image/bytedance/seedream-4", "Seedream 4", ["prompt", "width", "height", "seed", "style_images"], ["prompt", "width", "height"]), + imagePreset(210, "/generate/image/bytedance/seedream-5-lite", "Seedream 5 Lite", ["prompt", "width", "height", "seed", "style_images"], ["prompt", "width", "height"]), + imagePreset(220, "/generate/image/qwen/2512", "Qwen 2512", ["prompt", "negative_prompt", "width", "height", "num_inference_steps", "cfg_scale", "seed", "skip_prompt_expansion", "styles"], ["prompt"]), + imagePreset(230, "/generate/image/z-image/z-image", "Z Image", ["prompt", "seed", "denoising_strength", "styles", "skip_prompt_expansion", "aspect_ratio", "resolution", "image_url", "style_images"], ["prompt", "aspect_ratio", "resolution"]), + imagePreset(240, "/generate/image/bytedance/seededit", "SeedEdit", ["prompt", "seed", "image_url"], ["prompt", "image_url"]), + enhancePreset(250, "/generate/enhance/topaz/generative-enhance", "Topaz Generative", ["prompt", "width", "height", "seed", "image_url", "model", "output_format", "subject_detection", "face_enhancement", "face_enhancement_creativity", "face_enhancement_strength", "crop_to_fill", "upscaling_activated", "image_scaling_factor", "creativity", "texture", "sharpen", "denoise", "detail"], ["width", "height", "image_url"]), + enhancePreset(260, "/generate/enhance/topaz/standard-enhance", "Topaz", ["prompt", "width", "height", "seed", "image_url", "model", "output_format", "subject_detection", "face_enhancement", "face_enhancement_creativity", "face_enhancement_strength", "crop_to_fill", "upscaling_activated", "image_scaling_factor", "sharpen", "denoise", "fix_compression", "strength"], ["width", "height", "image_url", "model"]), + enhancePreset(270, "/generate/enhance/topaz/bloom-enhance", "Topaz Bloom", ["prompt", "width", "height", "seed", "image_url", "output_format", "crop_to_fill", "creativity", "face_preservation", "color_preservation", "upscaling_activated", "image_scaling_factor"], ["width", "height", "image_url"]), + videoPreset(280, "/generate/video/kling/kling-1", "Kling 1.0", ["startImage", "prompt", "aspectRatio", "duration", "cameraControl", "endImage", "mode"], ["prompt"]), + videoPreset(290, "/generate/video/kling/kling-1.5", "Kling 1.5", ["startImage", "prompt", "aspectRatio", "duration", "cameraControl", "endImage"], ["prompt"]), + videoPreset(300, "/generate/video/kling/kling-1.6", "Kling 1.6", ["startImage", "prompt", "aspectRatio", "duration", "cameraControl", "endImage", "mode"], ["prompt"]), + videoPreset(310, "/generate/video/kling/kling-2", "Kling 2.0", ["startImage", "prompt", "aspectRatio", "endImage", "mode"], ["prompt"]), + videoPreset(320, "/generate/video/kling/kling-2.1", "Kling 2.1", ["startImage", "prompt", "aspectRatio", "duration", "endImage", "mode"], ["prompt"]), + videoPreset(330, "/generate/video/kling/kling-2.5", "Kling 2.5", ["startImage", "prompt", "aspectRatio", "duration", "endImage"], ["prompt"]), + videoPreset(340, "/generate/video/kling/kling-2.6", "Kling 2.6", ["startImage", "prompt", "aspectRatio", "duration", "endImage", "generateAudio"], ["prompt"]), + videoPreset(350, "/generate/video/kling/kling-o1", "Kling o1", ["startImage", "prompt", "aspectRatio", "duration", "endImage", "styleReferences", "elementReferences", "videoReference"], ["prompt"]), + videoPreset(360, "/generate/video/minimax/hailuo", "Hailuo", ["startImage", "prompt", "expandPrompt", "cameraControl", "model", "subjects"], ["prompt"]), + videoPreset(370, "/generate/video/minimax/hailuo-02", "Hailuo 02", ["startImage", "prompt", "expandPrompt", "endImage", "resolution", "duration"], ["prompt"]), + videoPreset(380, "/generate/video/minimax/hailuo-2.3", "Hailuo 2.3", ["startImage", "prompt", "expandPrompt", "endImage", "resolution", "duration"], ["prompt"]), + videoPreset(390, "/generate/video/minimax/hailuo-2.3-fast", "Hailuo 2.3 Fast", ["startImage", "prompt", "expandPrompt", "endImage", "resolution", "duration"], ["prompt"]), + videoPreset(400, "/generate/video/alibaba/wan-2.1", "Wan 2.1", ["startImage", "prompt", "width", "height", "endImage", "aspectRatio", "frame_num", "styles"], ["prompt", "width", "height"]), + videoPreset(410, "/generate/video/alibaba/wan-2.2", "Wan 2.2", ["startImage", "guide_scale", "prompt", "width", "height", "aspectRatio", "frame_num", "styles"], ["prompt", "width", "height"]), + videoPreset(420, "/generate/video/alibaba/wan-2.5", "Wan 2.5", ["startImage", "prompt", "aspectRatio", "resolution", "enablePromptExpansion", "duration", "seed", "audio_url", "generateAudio"], ["prompt"]), + videoPreset(430, "/generate/video/google/veo-2", "Veo 2", ["startImage", "prompt", "aspectRatio", "duration", "image_url"], ["prompt"]), + videoPreset(440, "/generate/video/google/veo-3", "Veo 3", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution"], ["prompt"]), + videoPreset(450, "/generate/video/google/veo-3-fast", "Veo 3 Fast", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution"], ["prompt"]), + videoPreset(460, "/generate/video/google/veo-3.1", "Veo 3.1", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution", "endImage", "referenceImages"], ["prompt"]), + videoPreset(470, "/generate/video/google/veo-3.1-fast", "Veo 3.1 Fast", ["startImage", "prompt", "aspectRatio", "duration", "generateAudio", "resolution", "endImage", "referenceImages"], ["prompt"]), + videoPreset(480, "/generate/video/google/veo-3.1-lite", "Veo 3.1 Lite", ["startImage", "prompt", "aspectRatio", "duration", "resolution", "endImage"], ["prompt"]), + videoPreset(490, "/generate/video/bytedance/seedance-1.0-lite", "Seedance Lite", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "duration", "resolution"], ["prompt"]), + videoPreset(500, "/generate/video/bytedance/seedance-1.0-pro", "Seedance Pro", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "duration", "resolution"], ["prompt"]), + videoPreset(510, "/generate/video/bytedance/seedance-1.0-pro-fast", "Seedance Pro Fast", ["startImage", "prompt", "aspectRatio", "duration", "resolution"], ["prompt"]), + videoPreset(520, "/generate/video/bytedance/seedance-2", "Seedance 2.0", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "referenceVideos", "referenceAudios", "duration", "generateAudio", "resolution", "seed"], ["prompt"]), + videoPreset(530, "/generate/video/bytedance/seedance-2-fast", "Seedance 2.0 Fast", ["startImage", "prompt", "aspectRatio", "endImage", "referenceImages", "referenceVideos", "referenceAudios", "duration", "generateAudio", "resolution", "seed"], ["prompt"]), + videoPreset(540, "/generate/video/openai/sora-2", "Sora 2", ["startImage", "prompt", "duration", "aspectRatio"], ["prompt"]), + videoPreset(550, "/generate/video/luma/ray-2", "Ray 2", ["startImage", "prompt", "aspectRatio", "loop", "width", "height"], ["prompt", "width", "height"]), + videoPreset(560, "/generate/video/runway/gen-3", "Runway Gen-3", ["startImage", "prompt", "aspectRatio", "duration", "seed"], ["prompt"]), + videoPreset(570, "/generate/video/runway/gen-4-video", "Runway Gen-4", ["startImage", "prompt", "aspectRatio", "duration", "seed"], ["prompt"]), + videoPreset(580, "/generate/video/runway/gen-4.5", "Runway Gen-4.5", ["startImage", "prompt", "aspectRatio", "duration", "seed"], ["prompt"]), + videoPreset(590, "/generate/video/runway/aleph", "Runway Aleph", ["startImage", "startVideo", "startVideoTrim", "endVideoTrim", "prompt", "init_video", "init_image_url", "references", "ratio", "seed", "contentModeration"], ["prompt", "init_video"]), ]; export function getKreaModelPresetById(modelId: string) { diff --git a/lib/ai/providers/krea.ts b/lib/ai/providers/krea.ts index 807b4942f..ba418726a 100644 --- a/lib/ai/providers/krea.ts +++ b/lib/ai/providers/krea.ts @@ -318,11 +318,6 @@ function normalizeSeed(seed?: number) { return clamp(Math.round(Number(seed)), 0, 4294967295); } -function normalizeBatchSize(value?: number) { - if (!Number.isFinite(Number(value))) return 1; - return clamp(Math.round(Number(value)), 1, 4); -} - function normalizeVideoDuration(duration: unknown, endpoint: string) { const value = Number.isFinite(Number(duration)) ? Math.round(Number(duration)) : 5; if (endpoint.includes("/bytedance/seedance-2")) { @@ -387,9 +382,11 @@ function buildTaggedReferences(imageUrls: string[], tagPrefix = "ref", max = 3) const KREA_SOURCE_IMAGE_FIELDS = new Set([ "imageUrls", + "image_urls", "imageUrl", "image_url", "referenceImages", + "reference_images", "init_image_url", "references", ]); @@ -401,6 +398,19 @@ function hasKreaSourceImageField(fields: Iterable) { return false; } +function hasAnyKreaField(fields: Set, aliases: string[]) { + return aliases.some((field) => fields.has(field)); +} + +function getPreferredKreaField(fields: Set, aliases: string[]) { + for (const field of aliases) { + if (fields.has(field)) { + return field; + } + } + return null; +} + function buildReferencePreservingPrompt(params: { endpoint: string; kind: KreaModelKind; @@ -449,19 +459,26 @@ function summarizeKreaReferenceFields(body: Record) { for (const field of [ "imageUrls", + "image_urls", "imageUrl", "image_url", "styleImages", + "style_images", "imageStyleRefs", + "image_style_references", "characterReferenceImages", + "character_reference_images", "referenceImages", + "reference_images", "startImage", "endImage", "init_image_url", "references", "subjects", "styleReferences", + "style_references", "elementReferences", + "element_references", ]) { const value = body[field]; if (Array.isArray(value)) { @@ -841,6 +858,7 @@ function buildKreaBody(params: { const required = new Set(requiredFields); const body: Record = {}; const isVideo = kind === "video"; + const isImageLike = !isVideo; const videoMode = isVideo ? videoParams?.mode || @@ -864,7 +882,6 @@ function buildKreaBody(params: { const videoDimensions = resolveVideoDimensions(ratio, videoResolution); if (fields.has("prompt")) body.prompt = prompt; - if (fields.has("batchSize")) body.batchSize = normalizeBatchSize(imageParams?.imageCount); if ( fields.has("width") && @@ -878,13 +895,20 @@ function buildKreaBody(params: { ) { body.height = isVideo ? videoDimensions.height : imageDimensions.height; } - if (fields.has("aspectRatio") && shouldInferImageGeometryFromReference && inferredReferenceAspectRatio) { - body.aspectRatio = inferredReferenceAspectRatio; + const aspectRatioField = getPreferredKreaField( + fields, + isImageLike ? ["aspect_ratio", "aspectRatio"] : ["aspectRatio", "aspect_ratio"] + ); + if (aspectRatioField && shouldInferImageGeometryFromReference && inferredReferenceAspectRatio) { + body[aspectRatioField] = inferredReferenceAspectRatio; } else if ( - fields.has("aspectRatio") && - (isVideo || !shouldInferImageGeometryFromReference || required.has("aspectRatio")) + aspectRatioField && + (isVideo || + !shouldInferImageGeometryFromReference || + required.has("aspectRatio") || + required.has("aspect_ratio")) ) { - body.aspectRatio = ratio === "auto" ? "1:1" : ratio; + body[aspectRatioField] = ratio === "auto" ? "1:1" : ratio; } if (fields.has("ratio")) body.ratio = ratio === "auto" ? "16:9" : ratio; if (fields.has("resolution")) { @@ -896,8 +920,12 @@ function buildKreaBody(params: { if (fields.has("negative_prompt") && imageParams?.negativePrompt) { body.negative_prompt = imageParams.negativePrompt; } - if (fields.has("skipPromptExpansion") && typeof imageParams?.promptExtend === "boolean") { - body.skipPromptExpansion = !imageParams.promptExtend; + const skipPromptExpansionField = getPreferredKreaField(fields, [ + "skip_prompt_expansion", + "skipPromptExpansion", + ]); + if (skipPromptExpansionField && typeof imageParams?.promptExtend === "boolean") { + body[skipPromptExpansionField] = !imageParams.promptExtend; } if (fields.has("enablePromptExpansion")) { body.enablePromptExpansion = true; @@ -908,8 +936,12 @@ function buildKreaBody(params: { if (fields.has("duration")) { body.duration = normalizeVideoDuration(videoParams?.duration, endpoint); } - if (fields.has("generateAudio") && typeof videoParams?.generateAudio === "boolean") { - body.generateAudio = videoParams.generateAudio; + const generateAudioField = getPreferredKreaField(fields, [ + "generate_audio", + "generateAudio", + ]); + if (generateAudioField && typeof videoParams?.generateAudio === "boolean") { + body[generateAudioField] = videoParams.generateAudio; } if (fields.has("loop")) { body.loop = false; @@ -924,31 +956,65 @@ function buildKreaBody(params: { if (fields.has("output_format")) body.output_format = "jpg"; if (fields.has("backgroundImage")) body.backgroundImage = "auto"; - if (fields.has("imageUrls") && imageUrls.length > 0) { - body.imageUrls = imageUrls.slice(0, 8); - } - if (fields.has("imageUrl") && imageUrls[0]) { - body.imageUrl = imageUrls[0]; - } - if (fields.has("image_url") && imageUrls[0]) { - body.image_url = imageUrls[0]; - } - if (fields.has("styleImages") && imageUrls.length > 0 && !body.imageUrls && !body.imageUrl) { - body.styleImages = buildStyleImages(imageUrls.slice(0, 4)); - } - if (fields.has("imageStyleRefs") && imageUrls.length > 1) { - body.imageStyleRefs = buildStyleImages(imageUrls.slice(1, 5)); - } - if (fields.has("characterReferenceImages") && imageUrls.length > 0) { - body.characterReferenceImages = imageUrls.slice(0, 4); + const imageUrlsField = getPreferredKreaField( + fields, + isImageLike ? ["image_urls", "imageUrls"] : ["imageUrls", "image_urls"] + ); + if (imageUrlsField && imageUrls.length > 0) { + body[imageUrlsField] = imageUrls.slice(0, 8); } + const imageUrlField = getPreferredKreaField( + fields, + isImageLike ? ["image_url", "imageUrl"] : ["imageUrl", "image_url"] + ); + if (imageUrlField && imageUrls[0]) { + body[imageUrlField] = imageUrls[0]; + } + + const styleImagesField = getPreferredKreaField( + fields, + isImageLike ? ["style_images", "styleImages"] : ["styleImages", "style_images"] + ); if ( - fields.has("referenceImages") && + styleImagesField && + imageUrls.length > 0 && + !hasAnyKreaField(new Set(Object.keys(body)), [ + "image_urls", + "imageUrls", + "image_url", + "imageUrl", + ]) + ) { + body[styleImagesField] = buildStyleImages(imageUrls.slice(0, 4)); + } + + const imageStyleReferencesField = getPreferredKreaField(fields, [ + "image_style_references", + "imageStyleRefs", + ]); + if (imageStyleReferencesField && imageUrls.length > 1) { + body[imageStyleReferencesField] = buildStyleImages(imageUrls.slice(1, 5)); + } + + const characterReferenceImagesField = getPreferredKreaField(fields, [ + "character_reference_images", + "characterReferenceImages", + ]); + if (characterReferenceImagesField && imageUrls.length > 0) { + body[characterReferenceImagesField] = imageUrls.slice(0, 4); + } + + const referenceImagesField = getPreferredKreaField(fields, [ + "reference_images", + "referenceImages", + ]); + if ( + referenceImagesField && imageUrls.length > 0 && (!isVideo || videoMode === "reference-images") ) { - body.referenceImages = endpoint.includes("/image/runway/gen-4-image") + body[referenceImagesField] = endpoint.includes("/image/runway/gen-4-image") ? buildTaggedReferences(imageUrls, "ref", 3) : imageUrls.slice(0, endpoint.includes("/seedance-2") ? 9 : 4); } @@ -970,27 +1036,50 @@ function buildKreaBody(params: { if (fields.has("subjects") && imageUrls[0] && videoParams?.mode === "reference-images") { body.subjects = [{ type: "character", image: [imageUrls[0]] }]; } - if (fields.has("styleReferences") && imageUrls.length > 0) { - body.styleReferences = buildTaggedReferences(imageUrls, "style", 3); + const styleReferencesField = getPreferredKreaField(fields, [ + "style_references", + "styleReferences", + ]); + if (styleReferencesField && imageUrls.length > 0) { + body[styleReferencesField] = buildTaggedReferences(imageUrls, "style", 3); } - if (fields.has("elementReferences") && imageUrls.length > 0) { - body.elementReferences = [ + + const elementReferencesField = getPreferredKreaField(fields, [ + "element_references", + "elementReferences", + ]); + if (elementReferencesField && imageUrls.length > 0) { + body[elementReferencesField] = [ { tag: "element_1", images: imageUrls.slice(0, 3), }, ]; } - if (fields.has("videoReference") && videoUrls[0]) { - body.videoReference = { url: videoUrls[0], tag: "video_1" }; + + const videoReferenceField = getPreferredKreaField(fields, [ + "video_reference", + "videoReference", + ]); + if (videoReferenceField && videoUrls[0]) { + body[videoReferenceField] = { url: videoUrls[0], tag: "video_1" }; } if (fields.has("startVideo") && videoUrls[0]) body.startVideo = videoUrls[0]; if (fields.has("init_video") && videoUrls[0]) body.init_video = videoUrls[0]; - if (fields.has("referenceVideos") && videoUrls.length > 0) { - body.referenceVideos = videoUrls.slice(0, 3); + + const referenceVideosField = getPreferredKreaField(fields, [ + "reference_videos", + "referenceVideos", + ]); + if (referenceVideosField && videoUrls.length > 0) { + body[referenceVideosField] = videoUrls.slice(0, 3); } - if (required.has("referenceImages") && !body.referenceImages) { + if ( + (required.has("referenceImages") || required.has("reference_images")) && + !body.referenceImages && + !body.reference_images + ) { throw new Error("该 Krea 模型需要至少 1 张参考图"); } if ((required.has("imageUrl") || required.has("image_url")) && !body.imageUrl && !body.image_url) { @@ -1038,7 +1127,7 @@ export function createKrea({ apiKey, baseURL }: KreaConfig) { preset?.requestFields || (kind === "video" ? ["prompt", "aspectRatio", "duration", "startImage", "endImage", "referenceImages"] - : ["prompt", "width", "height", "imageUrls", "imageUrl", "batchSize"]); + : ["prompt", "width", "height", "image_urls", "image_url"]); const requiredFields = preset?.requiredFields || ["prompt"]; const requestedImageAspectRatio = normalizeText(options.imageParams?.aspectRatio).toLowerCase(); const referenceAspectRatio =