mulmocast 1.1.11 → 1.2.1

package/assets/templates/akira_comic.json

@@ -1,7 +1,7 @@
 {
   "title": "Akira style",
   "description": "Template for Akira style comic presentation.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/ani.json

@@ -1,7 +1,7 @@
 {
   "title": "Presentation with Ani",
   "description": "Template for presentation with Ani.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. 言葉づかいは少しツンデレにして。Another AI will generate comic for each beat based on the image prompt of that beat. You don't need to specify the style of the image, just describe the scene. Mention the reference in one of beats, if it exists. Use the JSON below as a template. Create appropriate amount of beats, and make sure the beats are coherent and flow well.",
+  "systemPrompt": "言葉づかいは思いっきりツンデレにして。Another AI will generate comic for each beat based on the image prompt of that beat. You don't need to specify the style of the image, just describe the scene. Mention the reference in one of beats, if it exists. Use the JSON below as a template. Create appropriate amount of beats, and make sure the beats are coherent and flow well.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/business.json

@@ -1,6 +1,6 @@
 {
   "title": "Business presentation",
   "description": "Template for business presentation.",
-  "systemPrompt": "Generate a script for a business presentation of the given topic. Use textSlides, markdown, mermaid, or chart to show slides. Extract image links in the article (from <img> tag) to reuse them in the presentation. Mention the reference in one of beats, if it exists. Use the JSON below as a template. chartData is the data for Chart.js",
+  "systemPrompt": "Use textSlides, markdown, mermaid, or chart to show slides. Extract image links in the article (from <img> tag) to reuse them in the presentation. Mention the reference in one of beats, if it exists. Use the JSON below as a template. chartData is the data for Chart.js",
   "scriptName": "business.json"
 }

package/assets/templates/characters.json

@@ -11,6 +11,6 @@
       "height": 1024
     }
   },
-  "systemPrompt": "Generate a script for a the given story with multiple characters. Generate image prompts for each character, and make references to them in the beats. Use the JSON below as a template.",
+  "systemPrompt": "Use multiple characters. Generate image prompts for each character, and make references to them in the beats. Use the JSON below as a template.",
   "scriptName": "image_refs.json"
 }

package/assets/templates/children_book.json

@@ -1,7 +1,7 @@
 {
   "title": "Children Book",
   "description": "Template for children book.",
-  "systemPrompt": "Please generate a script for a children book on the topic provided by the user. Each page (=beat) must haven an image prompt appropriate for the text.",
+  "systemPrompt": "This script is for a children book. Each page (=beat) must haven an image prompt appropriate for the text.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/coding.json

@@ -1,6 +1,6 @@
 {
   "title": "Coding presentation",
   "description": "Template for software and coding presentation.",
-  "systemPrompt": "Generate a script for a technical presentation of the given topic. Use markdown with a code block to show some code on a slide. Avoid long coding examples, which may not fit in a single slide. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Use markdown with a code block to show some code on a slide. Avoid long coding examples, which may not fit in a single slide. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "scriptName": "coding.json"
 }

package/assets/templates/comic_strips.json

@@ -1,7 +1,7 @@
 {
   "title": "American Comic Strips",
   "description": "Template for Dilbert-style comic strips.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/drslump_comic.json

@@ -1,7 +1,7 @@
 {
   "title": "Dr. Slump Style",
   "description": "Template for Dr. Slump style comic presentation.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/ghibli_comic.json

@@ -1,7 +1,7 @@
 {
   "title": "Ghibli comic style",
   "description": "Template for Ghibli-style comic presentation.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/ghibli_image_only.json

@@ -1,7 +1,7 @@
 {
   "title": "Ghibli comic image-only",
   "description": "Template for Ghibli-style image-only comic presentation.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate an image for each beat based on the text description of that beat. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate an image for each beat based on the text description of that beat. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/ghibli_shorts.json

@@ -1,7 +1,7 @@
 {
   "title": "Ghibli style for YouTube Shorts",
   "description": "Template for Ghibli-style comic presentation.",
-  "systemPrompt": "Generate a Japanese script for a Youtube shorts of the given topic. Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "This script is for YouTube shorts. Another AI will generate comic strips for each beat based on the text description of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/ghost_comic.json

@@ -1,7 +1,7 @@
 {
   "title": "Ghost in the shell style",
   "description": "Template for Ghost in the shell style comic presentation.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/html.json

@@ -1,6 +1,17 @@
 {
   "title": "Business presentation in HTML",
   "description": "Template for business presentation in HTML.",
-  "systemPrompt": "Generate a script for a business presentation of the given topic. Another LLM will generate actual slides from the prompt and data for each beat. Adding optional data would help it to generate more compelling slide. Mention the reference in one of beats, if it exists. The valid type of reference is 'article', 'paper', 'image', 'video', 'audio'. Use the JSON below as a template.",
+  "systemPrompt": "Another LLM will generate actual slides from the prompt and data for each beat. Adding optional data would help it to generate more compelling slide. Mention the reference in one of beats, if it exists. The valid type of reference is 'article', 'paper', 'image', 'video', 'audio'. Use the JSON below as a template.",
+  "presentationStyle": {
+    "$mulmocast": {
+      "version": "1.1",
+      "credit": "closing"
+    },
+    "lang": "en",
+    "canvasSize": {
+      "width": 1536,
+      "height": 1024
+    }
+  },
   "scriptName": "html.json"
 }

package/assets/templates/onepiece_comic.json

@@ -1,7 +1,7 @@
 {
   "title": "One Piece style",
   "description": "Template for One Piece style comic presentation.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate images for each beat based on the image prompt of that beat. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/portrait_movie.json

@@ -1,7 +1,7 @@
 {
   "title": "Photo realistic movie (portrait)",
   "description": "Template for photo realistic movie in portrait mode.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/realistic_movie.json

@@ -1,7 +1,7 @@
 {
   "title": "Photo realistic movie template",
   "description": "Template for photo realistic movie.",
-  "systemPrompt": "Generate a script for a presentation of the given topic. Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
+  "systemPrompt": "Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English. Mention the reference in one of beats, if it exists. Use the JSON below as a template.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/sensei_and_taro.json

@@ -1,7 +1,7 @@
 {
   "title": "Student and Teacher",
   "description": "Interactive discussion between a student and teacher",
-  "systemPrompt": "この件について、内容全てを高校生にも分かるように、太郎くん(Student)と先生(Teacher)の会話、という形の台本をArtifactとして作って。ただし要点はしっかりと押さえて。以下に別のトピックに関するサンプルを貼り付けます。このJSONフォーマットに従って。",
+  "systemPrompt": "全てを高校生にも分かるように、太郎くん(Student)と先生(Teacher)の会話、という形の台本にして。ただし要点はしっかりと押さえて。以下に別のトピックに関するサンプルを貼り付けます。このJSONフォーマットに従って。",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1",

package/assets/templates/shorts.json

@@ -1,7 +1,7 @@
 {
   "title": "Short movie template",
   "description": "Template for Youtube shorts.",
-  "systemPrompt": "Generate a script for a Youtube shorts of the given topic. The first beat should be a hook, which describes the topic. Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English.",
+  "systemPrompt": "This script is for YouTube shorts. The first beat should be a hook, which describes the topic. Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1"

package/assets/templates/trailer.json

@@ -1,7 +1,7 @@
 {
   "title": "Movie Trailer template",
   "description": "Template for A Movie Trailer.",
-  "systemPrompt": "Generate a script for a movie trailer of the given story. Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English.",
+  "systemPrompt": "This script is for a movie trailer. Another AI will generate images for each beat based on the image prompt of that beat. Movie prompts must be written in English.",
   "presentationStyle": {
     "$mulmocast": {
       "version": "1.1"

package/lib/actions/image_agents.d.ts

@@ -8,7 +8,7 @@ export declare const imagePreprocessAgent: (namedInputs: {
     imagePath: string;
     htmlPrompt: string | undefined;
     htmlPath: string;
-    htmlImageSystemPrompt: string[];
+    htmlImageSystemPrompt: string;
 } | {
     imagePath: string | undefined;
     referenceImageForMovie: string | undefined;

package/lib/actions/image_references.js

@@ -4,7 +4,7 @@ import { getReferenceImagePath } from "../utils/file.js";
 import { getExtention } from "../utils/utils.js";
 import { graphOption } from "./images.js";
 import { MulmoPresentationStyleMethods, MulmoStudioContextMethods } from "../methods/index.js";
-import { imageGoogleAgent, imageOpenaiAgent, mediaMockAgent } from "../agents/index.js";
+import { imageOpenaiAgent, mediaMockAgent, imageGenAIAgent } from "../agents/index.js";
 // public api
 // Application may call this function directly to generate reference image.
 export const generateReferenceImage = async (inputs) => {
@@ -39,7 +39,7 @@ export const generateReferenceImage = async (inputs) => {
         },
     };
     const options = await graphOption(context);
-    const graph = new GraphAI(image_graph_data, { imageGoogleAgent, imageOpenaiAgent, mediaMockAgent }, options);
+    const graph = new GraphAI(image_graph_data, { imageGenAIAgent, imageOpenaiAgent, mediaMockAgent }, options);
     await graph.run();
     return imagePath;
 };

package/lib/actions/images.js

@@ -1,26 +1,25 @@
 import dotenv from "dotenv";
 import fs from "fs";
 import { GraphAI, GraphAILogger, TaskManager } from "graphai";
-import { GoogleAuth } from "google-auth-library";
 import * as vanilla from "@graphai/vanilla";
 import { openAIAgent } from "@graphai/openai_agent";
 import { anthropicAgent } from "@graphai/anthropic_agent";
 import { fileWriteAgent } from "@graphai/vanilla_node_agents";
-import { imageGoogleAgent, imageOpenaiAgent, movieGoogleAgent, movieReplicateAgent, mediaMockAgent, soundEffectReplicateAgent, lipSyncReplicateAgent, } from "../agents/index.js";
+import { imageGenAIAgent, imageOpenaiAgent, movieGenAIAgent, movieReplicateAgent, mediaMockAgent, soundEffectReplicateAgent, lipSyncReplicateAgent, } from "../agents/index.js";
 import { MulmoPresentationStyleMethods, MulmoStudioContextMethods } from "../methods/index.js";
 import { getOutputStudioFilePath, mkdir } from "../utils/file.js";
 import { fileCacheAgentFilter } from "../utils/filters.js";
-import { userAssert, settings2GraphAIConfig } from "../utils/utils.js";
+import { settings2GraphAIConfig } from "../utils/utils.js";
 import { extractImageFromMovie, ffmpegGetMediaDuration } from "../utils/ffmpeg_utils.js";
 import { getImageRefs } from "./image_references.js";
 import { imagePreprocessAgent, imagePluginAgent, htmlImageGeneratorAgent } from "./image_agents.js";
 const vanillaAgents = vanilla.default ?? vanilla;
 const imageAgents = {
-    imageGoogleAgent,
+    imageGenAIAgent,
     imageOpenaiAgent,
 };
 const movieAgents = {
-    movieGoogleAgent,
+    movieGenAIAgent,
     movieReplicateAgent,
     mediaMockAgent,
 };
@@ -151,6 +150,7 @@ const beat_graph_data = {
                 onComplete: [":imageGenerator", ":imagePlugin"], // to wait for imageGenerator to finish
                 prompt: ":beat.moviePrompt",
                 imagePath: ":preprocessor.referenceImageForMovie",
+                movieFile: ":preprocessor.movieFile", // for google genai agent
                 cache: {
                     force: [":context.force", ":forceMovie"],
                     file: ":preprocessor.movieFile",
@@ -331,20 +331,6 @@ const graph_data = {
         },
     },
 };
-const googleAuth = async () => {
-    try {
-        const auth = new GoogleAuth({
-            scopes: ["https://www.googleapis.com/auth/cloud-platform"],
-        });
-        const client = await auth.getClient();
-        const accessToken = await client.getAccessToken();
-        return accessToken.token;
-    }
-    catch (error) {
-        GraphAILogger.info("install gcloud and run 'gcloud auth application-default login'");
-        throw error;
-    }
-};
 export const graphOption = async (context, settings) => {
     const options = {
         agentFilters: [
@@ -356,17 +342,7 @@ export const graphOption = async (context, settings) => {
         ],
         taskManager: new TaskManager(MulmoPresentationStyleMethods.getConcurrency(context.presentationStyle)),
     };
-    const provider = MulmoPresentationStyleMethods.getText2ImageProvider(context.presentationStyle.imageParams?.provider);
-    const config = settings2GraphAIConfig(settings, process.env);
-    // We need to get google's auth token only if the google is the text2image provider.
-    if (provider === "google" || context.presentationStyle.movieParams?.provider === "google") {
-        userAssert(!!config.movieGoogleAgent || !!config.imageGoogleAgent, "GOOGLE_PROJECT_ID is not set");
-        GraphAILogger.log("google was specified as text2image engine");
-        const token = await googleAuth();
-        config["imageGoogleAgent"].token = token;
-        config["movieGoogleAgent"].token = token;
-    }
-    options.config = config;
+    options.config = settings2GraphAIConfig(settings, process.env);
     return options;
 };
 const prepareGenerateImages = async (context) => {

package/lib/actions/movie.js

@@ -80,7 +80,7 @@ const getOutputOption = (audioId, videoId) => {
 const addCaptions = (ffmpegContext, concatVideoId, context, caption) => {
     const beatsWithCaptions = context.studio.beats.filter(({ captionFile }) => captionFile);
     if (caption && beatsWithCaptions.length > 0) {
-        const introPadding = context.presentationStyle.audioParams.introPadding;
+        const introPadding = MulmoStudioContextMethods.getIntroPadding(context);
         return beatsWithCaptions.reduce((acc, beat, index) => {
             const { startAt, duration, captionFile } = beat;
             if (startAt !== undefined && duration !== undefined && captionFile !== undefined) {
@@ -168,7 +168,7 @@ const createVideo = async (audioArtifactFilePath, outputVideoPath, context) => {
         const extraPadding = (() => {
             // We need to consider only intro and outro padding because the other paddings were already added to the beat.duration
             if (index === 0) {
-                return context.presentationStyle.audioParams.introPadding;
+                return MulmoStudioContextMethods.getIntroPadding(context);
             }
             else if (index === context.studio.beats.length - 1) {
                 return context.presentationStyle.audioParams.outroPadding;

package/lib/agents/add_bgm_agent.js

@@ -1,6 +1,7 @@
 import fs from "fs";
 import { GraphAILogger } from "graphai";
 import { FfmpegContextAddInput, FfmpegContextInit, FfmpegContextGenerateOutput, ffmpegGetMediaDuration } from "../utils/ffmpeg_utils.js";
+import { MulmoStudioContextMethods } from "../methods/mulmo_studio_context.js";
 const addBGMAgent = async ({ namedInputs, params, }) => {
     const { voiceFile, outputFile, context } = namedInputs;
     const { musicFile } = params;
@@ -11,7 +12,7 @@ const addBGMAgent = async ({ namedInputs, params, }) => {
         throw new Error(`AddBGMAgent musicFile not exist: ${musicFile}`);
     }
     const { duration: speechDuration } = await ffmpegGetMediaDuration(voiceFile);
-    const introPadding = context.presentationStyle.audioParams.introPadding;
+    const introPadding = MulmoStudioContextMethods.getIntroPadding(context);
     const outroPadding = context.presentationStyle.audioParams.outroPadding;
     const totalDuration = speechDuration + introPadding + outroPadding;
     GraphAILogger.log("totalDucation:", speechDuration, totalDuration);

package/lib/agents/image_genai_agent.d.ts

@@ -0,0 +1,5 @@
+import type { AgentFunction, AgentFunctionInfo } from "graphai";
+import type { AgentBufferResult, ImageAgentInputs, ImageAgentParams, GenAIImageAgentConfig } from "../types/agent.js";
+export declare const imageGenAIAgent: AgentFunction<ImageAgentParams, AgentBufferResult, ImageAgentInputs, GenAIImageAgentConfig>;
+declare const imageGenAIAgentInfo: AgentFunctionInfo;
+export default imageGenAIAgentInfo;

package/lib/agents/image_genai_agent.js

@@ -0,0 +1,52 @@
+import { GraphAILogger } from "graphai";
+import { getAspectRatio } from "./movie_google_agent.js";
+import { provider2ImageAgent } from "../utils/provider2agent.js";
+import { GoogleGenAI, PersonGeneration } from "@google/genai";
+export const imageGenAIAgent = async ({ namedInputs, params, config, }) => {
+    const { prompt } = namedInputs;
+    const aspectRatio = getAspectRatio(params.canvasSize);
+    const model = params.model ?? provider2ImageAgent["google"].defaultModel;
+    const apiKey = config?.apiKey;
+    if (!apiKey) {
+        throw new Error("API key is required for Google GenAI agent");
+    }
+    try {
+        const ai = new GoogleGenAI({ apiKey });
+        const response = await ai.models.generateImages({
+            model,
+            prompt,
+            config: {
+                numberOfImages: 1, // default is 4!
+                aspectRatio,
+                personGeneration: PersonGeneration.ALLOW_ALL,
+                // safetyFilterLevel: SafetyFilterLevel.BLOCK_ONLY_HIGH,
+            },
+        });
+        if (!response.generatedImages || response.generatedImages.length === 0) {
+            throw new Error("ERROR: generateImage returned no generated images");
+        }
+        const image = response.generatedImages[0].image;
+        if (image && image.imageBytes) {
+            return { buffer: Buffer.from(image.imageBytes, "base64") };
+        }
+        throw new Error("ERROR: generateImage returned no image bytes");
+    }
+    catch (error) {
+        GraphAILogger.info("Failed to generate image:", error);
+        throw error;
+    }
+};
+const imageGenAIAgentInfo = {
+    name: "imageGenAIAgent",
+    agent: imageGenAIAgent,
+    mock: imageGenAIAgent,
+    samples: [],
+    description: "Google Image agent",
+    category: ["image"],
+    author: "Receptron Team",
+    repository: "https://github.com/receptron/mulmocast-cli/",
+    // source: "https://github.com/receptron/mulmocast-cli/blob/main/src/agents/image_google_agent.ts",
+    license: "MIT",
+    environmentVariables: [],
+};
+export default imageGenAIAgentInfo;

package/lib/agents/image_openai_agent.js

@@ -39,6 +39,7 @@ export const imageOpenaiAgent = async ({ namedInputs, params, config, }) => {
         prompt,
         n: 1,
         size,
+        background: "opaque",
     };
     if (model === "gpt-image-1") {
         imageOptions.moderation = moderation || "auto";

package/lib/agents/index.d.ts

@@ -1,9 +1,9 @@
 import addBGMAgent from "./add_bgm_agent.js";
 import combineAudioFilesAgent from "./combine_audio_files_agent.js";
-import imageGoogleAgent from "./image_google_agent.js";
+import imageGenAIAgent from "./image_genai_agent.js";
 import imageOpenaiAgent from "./image_openai_agent.js";
 import tavilySearchAgent from "./tavily_agent.js";
-import movieGoogleAgent from "./movie_google_agent.js";
+import movieGenAIAgent from "./movie_genai_agent.js";
 import movieReplicateAgent from "./movie_replicate_agent.js";
 import mediaMockAgent from "./media_mock_agent.js";
 import ttsElevenlabsAgent from "./tts_elevenlabs_agent.js";
@@ -17,4 +17,4 @@ import { browserlessAgent } from "@graphai/browserless_agent";
 import { textInputAgent } from "@graphai/input_agents";
 import { openAIAgent } from "@graphai/openai_agent";
 import { fileWriteAgent } from "@graphai/vanilla_node_agents";
-export { openAIAgent, fileWriteAgent, browserlessAgent, textInputAgent, addBGMAgent, combineAudioFilesAgent, imageGoogleAgent, imageOpenaiAgent, tavilySearchAgent, movieGoogleAgent, movieReplicateAgent, mediaMockAgent, ttsElevenlabsAgent, ttsNijivoiceAgent, ttsOpenaiAgent, ttsGoogleAgent, validateSchemaAgent, soundEffectReplicateAgent, lipSyncReplicateAgent, };
+export { openAIAgent, fileWriteAgent, browserlessAgent, textInputAgent, addBGMAgent, combineAudioFilesAgent, imageGenAIAgent, imageOpenaiAgent, tavilySearchAgent, movieGenAIAgent, movieReplicateAgent, mediaMockAgent, ttsElevenlabsAgent, ttsNijivoiceAgent, ttsOpenaiAgent, ttsGoogleAgent, validateSchemaAgent, soundEffectReplicateAgent, lipSyncReplicateAgent, };

package/lib/agents/index.js

@@ -1,9 +1,9 @@
 import addBGMAgent from "./add_bgm_agent.js";
 import combineAudioFilesAgent from "./combine_audio_files_agent.js";
-import imageGoogleAgent from "./image_google_agent.js";
+import imageGenAIAgent from "./image_genai_agent.js";
 import imageOpenaiAgent from "./image_openai_agent.js";
 import tavilySearchAgent from "./tavily_agent.js";
-import movieGoogleAgent from "./movie_google_agent.js";
+import movieGenAIAgent from "./movie_genai_agent.js";
 import movieReplicateAgent from "./movie_replicate_agent.js";
 import mediaMockAgent from "./media_mock_agent.js";
 import ttsElevenlabsAgent from "./tts_elevenlabs_agent.js";
@@ -18,4 +18,4 @@ import { textInputAgent } from "@graphai/input_agents";
 import { openAIAgent } from "@graphai/openai_agent";
 // import * as vanilla from "@graphai/vanilla";
 import { fileWriteAgent } from "@graphai/vanilla_node_agents";
-export { openAIAgent, fileWriteAgent, browserlessAgent, textInputAgent, addBGMAgent, combineAudioFilesAgent, imageGoogleAgent, imageOpenaiAgent, tavilySearchAgent, movieGoogleAgent, movieReplicateAgent, mediaMockAgent, ttsElevenlabsAgent, ttsNijivoiceAgent, ttsOpenaiAgent, ttsGoogleAgent, validateSchemaAgent, soundEffectReplicateAgent, lipSyncReplicateAgent, };
+export { openAIAgent, fileWriteAgent, browserlessAgent, textInputAgent, addBGMAgent, combineAudioFilesAgent, imageGenAIAgent, imageOpenaiAgent, tavilySearchAgent, movieGenAIAgent, movieReplicateAgent, mediaMockAgent, ttsElevenlabsAgent, ttsNijivoiceAgent, ttsOpenaiAgent, ttsGoogleAgent, validateSchemaAgent, soundEffectReplicateAgent, lipSyncReplicateAgent, };

package/lib/agents/movie_genai_agent.d.ts

@@ -0,0 +1,9 @@
+import type { AgentFunction, AgentFunctionInfo } from "graphai";
+import type { AgentBufferResult, GenAIImageAgentConfig, GoogleMovieAgentParams, MovieAgentInputs } from "../types/agent.js";
+export declare const getAspectRatio: (canvasSize: {
+    width: number;
+    height: number;
+}) => string;
+export declare const movieGenAIAgent: AgentFunction<GoogleMovieAgentParams, AgentBufferResult, MovieAgentInputs, GenAIImageAgentConfig>;
+declare const movieGenAIAgentInfo: AgentFunctionInfo;
+export default movieGenAIAgentInfo;

package/lib/agents/movie_genai_agent.js

@@ -0,0 +1,86 @@
+import { readFileSync } from "fs";
+import { GraphAILogger, sleep } from "graphai";
+import { GoogleGenAI, PersonGeneration } from "@google/genai";
+export const getAspectRatio = (canvasSize) => {
+    if (canvasSize.width > canvasSize.height) {
+        return "16:9";
+    }
+    else if (canvasSize.width < canvasSize.height) {
+        return "9:16";
+    }
+    else {
+        return "1:1";
+    }
+};
+export const movieGenAIAgent = async ({ namedInputs, params, config, }) => {
+    const { prompt, imagePath, movieFile } = namedInputs;
+    const aspectRatio = getAspectRatio(params.canvasSize);
+    const model = params.model ?? "veo-2.0-generate-001"; // "veo-3.0-generate-preview";
+    const duration = params.duration ?? 8;
+    const apiKey = config?.apiKey;
+    if (!apiKey) {
+        throw new Error("API key is required for Google GenAI agent");
+    }
+    try {
+        const ai = new GoogleGenAI({ apiKey });
+        const payload = {
+            model,
+            prompt,
+            config: {
+                durationSeconds: duration,
+                aspectRatio,
+                personGeneration: undefined,
+            },
+            image: undefined,
+        };
+        if (imagePath) {
+            const buffer = readFileSync(imagePath);
+            const imageBytes = buffer.toString("base64");
+            payload.image = {
+                imageBytes,
+                mimeType: "image/png",
+            };
+        }
+        else {
+            payload.config.personGeneration = PersonGeneration.ALLOW_ALL;
+        }
+        const operation = await ai.models.generateVideos(payload);
+        const response = { operation };
+        // Poll the operation status until the video is ready.
+        while (!response.operation.done) {
+            await sleep(5000);
+            response.operation = await ai.operations.getVideosOperation(response);
+        }
+        if (!response.operation.response?.generatedVideos) {
+            throw new Error(`No video: ${JSON.stringify(response.operation, null, 2)}`);
+        }
+        const video = response.operation.response.generatedVideos[0].video;
+        if (!video) {
+            throw new Error(`No video: ${JSON.stringify(response.operation, null, 2)}`);
+        }
+        await ai.files.download({
+            file: video,
+            downloadPath: movieFile,
+        });
+        await sleep(5000); // HACK: Without this, the file is not ready yet.
+        return { saved: movieFile };
+    }
+    catch (error) {
+        GraphAILogger.info("Failed to generate movie:", error.message);
+        throw error;
+    }
+};
+const movieGenAIAgentInfo = {
+    name: "movieGenAIAgent",
+    agent: movieGenAIAgent,
+    mock: movieGenAIAgent,
+    samples: [],
+    description: "Google Movie agent",
+    category: ["movie"],
+    author: "Receptron Team",
+    repository: "https://github.com/receptron/mulmocast-cli/",
+    // source: "https://github.com/receptron/mulmocast-cli/blob/main/src/agents/image_google_agent.ts",
+    license: "MIT",
+    environmentVariables: [],
+};
+export default movieGenAIAgentInfo;

package/lib/cli/commands/tool/prompt/handler.js

@@ -10,5 +10,5 @@ export const handler = async (argv) => {
     setGraphAILogger(verbose, {
         template,
     });
-    await dumpPromptFromTemplate({ templateName: template });
+    await dumpPromptFromTemplate({ templateName: template, suppressFirstStatement: false });
 };

package/lib/cli/helpers.js

@@ -37,9 +37,11 @@ export const getFileObject = (args) => {
             // We generate a new unique script file from clipboard text in the output directory
             const generatedFileName = generateTimestampedFileName("script");
             const clipboardText = clipboardy.readSync();
+            const json = JSON.parse(clipboardText);
+            const formattedText = JSON.stringify(json, null, 2);
             const resolvedFilePath = resolveDirPath(outDirPath, `${generatedFileName}.json`);
             mkdir(outDirPath);
-            fs.writeFileSync(resolvedFilePath, clipboardText, "utf8");
+            fs.writeFileSync(resolvedFilePath, formattedText, "utf8");
             return { fileOrUrl: resolvedFilePath, fileName: generatedFileName };
         }
         const resolvedFileOrUrl = file ?? "";

package/lib/methods/mulmo_beat.js

@@ -2,7 +2,7 @@ import { findImagePlugin } from "../utils/image_plugins/index.js";
 export const MulmoBeatMethods = {
     getHtmlPrompt(beat) {
         if (beat?.htmlPrompt?.data) {
-            return beat.htmlPrompt.prompt + "\n\n data\n" + JSON.stringify(beat.htmlPrompt.data, null, 2);
+            return beat.htmlPrompt.prompt + "\n\n[data]\n" + JSON.stringify(beat.htmlPrompt.data, null, 2);
         }
         return beat?.htmlPrompt?.prompt;
     },

package/lib/methods/mulmo_presentation_style.js

@@ -71,7 +71,7 @@ export const MulmoPresentationStyleMethods = {
         // Notice that we copy imageParams from presentationStyle and update
         // provider and model appropriately.
         const imageParams = { ...presentationStyle.imageParams, ...beat?.imageParams };
-        const provider = MulmoPresentationStyleMethods.getText2ImageProvider(imageParams?.provider);
+        const provider = MulmoPresentationStyleMethods.getText2ImageProvider(imageParams?.provider) ?? defaultProviders.text2image;
         const agentInfo = provider2ImageAgent[provider];
         // The default text2image model is gpt-image-1 from OpenAI, and to use it you must have an OpenAI account and have verified your identity. If this is not possible, please specify dall-e-3 as the model.
         const defaultImageParams = {

package/lib/methods/mulmo_studio_context.d.ts

@@ -12,4 +12,5 @@ export declare const MulmoStudioContextMethods: {
     setSessionState(context: MulmoStudioContext, sessionType: SessionType, value: boolean): void;
     setBeatSessionState(context: MulmoStudioContext, sessionType: BeatSessionType, index: number, value: boolean): void;
     needTranslate(context: MulmoStudioContext, includeCaption?: boolean): boolean | "" | undefined;
+    getIntroPadding(context: MulmoStudioContext): number;
 };

package/lib/methods/mulmo_studio_context.js

@@ -71,4 +71,11 @@ export const MulmoStudioContextMethods = {
         }
         return context.studio.script.lang !== context.lang;
     },
+    getIntroPadding(context) {
+        if (context.studio.script.beats[0].enableLipSync) {
+            // NOTE: We must set introPadding to 0 when enableLipSync is true. Otherwise, the lipsync will be out of sync.
+            return 0;
+        }
+        return context.presentationStyle.audioParams.introPadding;
+    },
 };