jbilcke-hf/ai-comic-factory

1export type ProjectionMode = 'cartesian' | 'spherical'2 3export type CacheMode = "use" | "renew" | "ignore"4 5export interface RenderRequest {6  prompt: string7 8  // whether to use video segmentation9  // disabled (default)10  // firstframe: we only analyze the first frame11  // allframes: we analyze all the frames12  segmentation: 'disabled' | 'firstframe' | 'allframes'13 14  // segmentation will only be executed if we have a non-empty list of actionnables15  // actionnables are names of things like "chest", "key", "tree", "chair" etc16  actionnables: string[]17 18  nbFrames: number19  nbFPS: number20 21  nbSteps: number // min: 1, max: 5022 23  seed: number24 25  width: number // fixed at 1024 for now26  height: number // fixed at 512 for now27 28  // upscaling factor29  // 0: no upscaling30  // 1: no upscaling31  // 2: 2x larger32  // 3: 3x larger33  // 4x: 4x larger, up to 4096x4096 (warning: a PNG of this size can be 50 Mb!)34  upscalingFactor: number35 36  projection: ProjectionMode37 38  /**39   * Use turbo mode40   * 41   * At the time of writing this will use SSD-1B + LCM42   * https://huggingface.co/spaces/jbilcke-hf/fast-image-server43   */44  turbo: boolean45 46  cache: CacheMode47 48  wait: boolean // wait until the job is completed49 50  analyze: boolean // analyze the image to generate a caption (optional)51 52  identityImage: string // reference image for the main entity53}54 55export interface ImageSegment {56  id: number57  box: number[]58  color: number[]59  label: string60  score: number 61}62 63export type RenderedSceneStatus =64  | "pregenerated"65  | "pending"66  | "completed"67  | "error"68 69export interface RenderedScene {70  renderId: string71  status: RenderedSceneStatus72  assetUrl: string 73  alt: string74  error: string75  maskUrl: string76  segments: ImageSegment[]77}78 79export interface ImageAnalysisRequest {80  image: string // in base6481  prompt: string82}83 84export interface ImageAnalysisResponse {85  result: string86  error?: string87}88 89export type GeneratedPanel = {90  panel: number91  instructions: string92  speech: string93  caption: string94}95 96export type GeneratedPanels = GeneratedPanel[]97 98// LLMVendor = what the user configure in the UI (eg. a dropdown item called default server)99// LLMEngine = the actual engine to use (eg. hugging face)100export type LLMEngine =101  | "INFERENCE_API"102  | "INFERENCE_ENDPOINT"103  | "OPENAI"104  | "REPLICATE"105  | "GROQ"106  | "ANTHROPIC"107 108export type RenderingEngine =109  | "VIDEOCHAIN"110  | "OPENAI"111  | "REPLICATE"112  | "INFERENCE_API"113  | "INFERENCE_ENDPOINT"114 115export type RenderingModelVendor =116  | "SERVER"117  | "OPENAI"118  | "REPLICATE"119  | "HUGGINGFACE"120 121// LLMVendor = what the user configure in the UI (eg. a dropdown item called default server)122// LLMEngine = the actual engine to use (eg. hugging face)123export type LLMVendor =124  | "SERVER"125  | "OPENAI"126  | "GROQ"127  | "ANTHROPIC"128 129export type LLMVendorConfig = {130  vendor: LLMVendor131  apiKey: string132  modelId: string133}134 135export type LLMPredictionFunctionParams = {136  systemPrompt: string137  userPrompt: string138  nbMaxNewTokens: number139  llmVendorConfig: LLMVendorConfig140}141 142export type PostVisibility =143  | "featured" // featured by admins144  | "trending" // top trending / received more than 10 upvotes145  | "normal" // default visibility146 147export type Post = {148  postId: string149  appId: string150  prompt: string151  previewUrl: string152  assetUrl: string153  createdAt: string154  visibility: PostVisibility155  upvotes: number156  downvotes: number157}158 159export type CreatePostResponse = {160  success?: boolean161  error?: string162  post: Post163}164 165export type GetAppPostsResponse = {166  success?: boolean167  error?: string168  posts: Post[]169}170 171export type GetAppPostResponse = {172  success?: boolean173  error?: string174  post: Post175}176 177export type LayoutProps = {178  page: number179  nbPanels: number180}181 182// TODO: rename the *Model fields to better indicate if this is a LLM or RENDER mdoel183export type Settings = {184  renderingModelVendor: RenderingModelVendor185  renderingUseTurbo: boolean186  llmVendor: LLMVendor187  huggingFaceOAuth: string188  huggingfaceApiKey: string189  huggingfaceInferenceApiModel: string190  huggingfaceInferenceApiModelTrigger: string191  huggingfaceInferenceApiFileType: string192  replicateApiKey: string193  replicateApiModel: string194  replicateApiModelVersion: string195  replicateApiModelTrigger: string196  openaiApiKey: string197  openaiApiModel: string198  openaiApiLanguageModel: string199  groqApiKey: string200  groqApiLanguageModel: string201  anthropicApiKey: string202  anthropicApiLanguageModel: string203  hasGeneratedAtLeastOnce: boolean204  userDefinedMaxNumberOfPages: number205}206 207export type DynamicConfig = {208  maxNbPages: number209  nbPanelsPerPage: number210  nbTotalPanelsToGenerate: number211  oauthClientId: string212  oauthRedirectUrl: string213  oauthScopes: string214  enableHuggingFaceOAuth: boolean215  enableHuggingFaceOAuthWall: boolean216}217