@mixio-pro/kalaasetu-mcp 1.1.2 → 1.1.4

package/src/tools/hunyuan-avatar.ts

@@ -1,160 +0,0 @@
-import { z } from "zod";
-import { callFalModel } from "../utils/fal.utils";
-
-/**
- * Calculate number of frames based on audio duration at 25 FPS
- * Adds 1 second buffer to ensure complete audio coverage
- */
-function calculateFramesFromAudioDuration(
-  audioDurationSeconds: number
-): number {
-  const totalDuration = audioDurationSeconds + 1; // Add 1 second buffer
-  const frames = Math.round(totalDuration * 25); // 25 FPS
-
-  // Clamp to valid range (129-401 frames)
-  return Math.max(129, Math.min(401, frames));
-}
-
-/**
- * FAL AI Hunyuan Avatar - High-Fidelity Audio-Driven Human Animation
- */
-export const hunyuanAvatar = {
-  name: "hunyuan_avatar",
-  description:
-    "Generate high-fidelity audio-driven human animation videos using FAL AI Hunyuan Avatar. Creates realistic talking avatar animations from an image and audio file.",
-  parameters: z.object({
-    image_url: z
-      .string()
-      .describe("Public URL of the reference image for the avatar."),
-    audio_url: z
-      .string()
-      .describe("Public URL of the audio file to drive the animation."),
-    audio_duration_seconds: z
-      .number()
-      .optional()
-      .describe(
-        "Duration of the audio in seconds. If provided, will automatically calculate optimal frames (audio duration + 1 second buffer at 25 FPS)."
-      ),
-    text: z
-      .string()
-      .optional()
-      .describe(
-        "Text prompt describing the scene. Default: 'A cat is singing.'"
-      ),
-    num_frames: z
-      .number()
-      .optional()
-      .describe(
-        "Number of video frames to generate at 25 FPS. Range: 129 to 401. If not provided and audio_duration_seconds is given, will be calculated automatically. Default: 129"
-      ),
-    num_inference_steps: z
-      .number()
-      .optional()
-      .describe(
-        "Number of inference steps for sampling. Higher values give better quality but take longer. Range: 30 to 50. Default: 30"
-      ),
-    turbo_mode: z
-      .boolean()
-      .optional()
-      .describe(
-        "If true, the video will be generated faster with no noticeable degradation in visual quality. Default: true"
-      ),
-    seed: z.number().optional().describe("Random seed for generation."),
-    fal_key: z
-      .string()
-      .optional()
-      .describe(
-        "FAL API key. If not provided, will use FAL_KEY environment variable."
-      ),
-  }),
-  execute: async (args: {
-    image_url: string;
-    audio_url: string;
-    audio_duration_seconds?: number;
-    text?: string;
-    num_frames?: number;
-    num_inference_steps?: number;
-    turbo_mode?: boolean;
-    seed?: number;
-    fal_key?: string;
-  }) => {
-    // Calculate frames from audio duration if provided and num_frames not specified
-    let calculatedFrames = args.num_frames;
-    if (
-      args.audio_duration_seconds !== undefined &&
-      args.num_frames === undefined
-    ) {
-      calculatedFrames = calculateFramesFromAudioDuration(
-        args.audio_duration_seconds
-      );
-    }
-
-    // Validate num_frames range if provided
-    if (
-      calculatedFrames !== undefined &&
-      (calculatedFrames < 129 || calculatedFrames > 401)
-    ) {
-      throw new Error("num_frames must be between 129 and 401");
-    }
-
-    // Validate num_inference_steps range if provided
-    if (
-      args.num_inference_steps !== undefined &&
-      (args.num_inference_steps < 30 || args.num_inference_steps > 50)
-    ) {
-      throw new Error("num_inference_steps must be between 30 and 50");
-    }
-
-    // Build input payload
-    const input: any = {
-      image_url: args.image_url,
-      audio_url: args.audio_url,
-    };
-
-    // Add optional parameters if provided
-    if (args.text !== undefined) {
-      input.text = args.text;
-    }
-    if (calculatedFrames !== undefined) {
-      input.num_frames = calculatedFrames;
-    }
-    if (args.num_inference_steps !== undefined) {
-      input.num_inference_steps = args.num_inference_steps;
-    }
-    if (args.turbo_mode !== undefined) {
-      input.turbo_mode = args.turbo_mode;
-    }
-    if (args.seed !== undefined) {
-      input.seed = args.seed;
-    }
-
-    const result = await callFalModel("fal-ai/hunyuan-avatar", input, {
-      falKey: args.fal_key,
-    });
-
-    // Extract video data from the response
-    const videoData = result.data?.video;
-
-    if (!videoData || !videoData.url) {
-      throw new Error(
-        `No video data in completed response: ${JSON.stringify(result.data)}`
-      );
-    }
-
-    const videoUrl = videoData.url;
-    const fileName = videoData.file_name || "hunyuan_avatar.mp4";
-
-    return JSON.stringify({
-      videos: [
-        {
-          url: videoUrl,
-          filename: fileName,
-          mimeType: "video/mp4",
-          filesize: videoData.file_size,
-        },
-      ],
-      message: "Hunyuan Avatar video generated successfully",
-      requestId: result.requestId,
-    });
-  },
-};

package/src/tools/infinitalk.ts

@@ -1,156 +0,0 @@
-import { z } from "zod";
-import { callFalModel } from "../utils/fal.utils";
-
-/**
- * Calculate number of frames based on audio duration at 25 FPS
- * Adds 1 second buffer to ensure complete audio coverage
- */
-function calculateFramesFromAudioDuration(
-  audioDurationSeconds: number
-): number {
-  const totalDuration = audioDurationSeconds + 1; // Add 1 second buffer
-  const frames = Math.round(totalDuration * 25); // 25 FPS
-
-  // Clamp to valid range (41-721 frames)
-  return Math.max(41, Math.min(721, frames));
-}
-
-/**
- * FAL AI Infinitalk - Generate talking avatar video from image and audio
- */
-export const infinitalk = {
-  name: "infinitalk",
-  description:
-    "Generate a talking avatar video from an image and audio file using FAL AI Infinitalk. The avatar lip-syncs to the provided audio with natural facial expressions.",
-  parameters: z.object({
-    image_url: z
-      .string()
-      .describe(
-        "Public URL of the input image. If the input image does not match the chosen aspect ratio, it is resized and center cropped."
-      ),
-    audio_url: z
-      .string()
-      .describe("The Public URL of the audio file for lip-sync generation."),
-    audio_duration_seconds: z
-      .number()
-      .optional()
-      .describe(
-        "Duration of the audio in seconds. If provided, will automatically calculate optimal frames (audio duration + 1 second buffer at 25 FPS)."
-      ),
-    prompt: z
-      .string()
-      .describe(
-        "The text prompt to guide video generation (e.g., 'A woman with colorful hair talking on a podcast')"
-      ),
-    num_frames: z
-      .number()
-      .optional()
-      .describe(
-        "Number of frames to generate. Must be between 41 to 721. If not provided and audio_duration_seconds is given, will be calculated automatically. Default: 145"
-      ),
-    resolution: z
-      .enum(["480p", "720p"])
-      .optional()
-      .describe("Resolution of the video to generate. Default: '480p'"),
-    seed: z
-      .number()
-      .optional()
-      .describe(
-        "Random seed for reproducibility. If not provided, a random seed is chosen. Default: 42"
-      ),
-    acceleration: z
-      .enum(["none", "regular", "high"])
-      .optional()
-      .describe(
-        "The acceleration level to use for generation. Default: 'regular'"
-      ),
-    fal_key: z
-      .string()
-      .optional()
-      .describe(
-        "FAL API key. If not provided, will use FAL_KEY environment variable."
-      ),
-  }),
-  execute: async (args: {
-    image_url: string;
-    audio_url: string;
-    audio_duration_seconds?: number;
-    prompt: string;
-    num_frames?: number;
-    resolution?: "480p" | "720p";
-    seed?: number;
-    acceleration?: "none" | "regular" | "high";
-    fal_key?: string;
-  }) => {
-    // Calculate frames from audio duration if provided and num_frames not specified
-    let calculatedFrames = args.num_frames;
-    if (
-      args.audio_duration_seconds !== undefined &&
-      args.num_frames === undefined
-    ) {
-      calculatedFrames = calculateFramesFromAudioDuration(
-        args.audio_duration_seconds
-      );
-    }
-
-    // Validate num_frames range if provided
-    if (
-      calculatedFrames !== undefined &&
-      (calculatedFrames < 41 || calculatedFrames > 721)
-    ) {
-      throw new Error("num_frames must be between 41 and 721");
-    }
-
-    // Build input payload
-    const input: any = {
-      image_url: args.image_url,
-      audio_url: args.audio_url,
-      prompt: args.prompt,
-    };
-
-    // Add optional parameters if provided
-    if (calculatedFrames !== undefined) {
-      input.num_frames = calculatedFrames;
-    }
-
-    input.resolution = args.resolution || "480p";
-
-    if (args.seed !== undefined) {
-      input.seed = args.seed;
-    }
-    if (args.acceleration !== undefined) {
-      input.acceleration = args.acceleration;
-    }
-
-    const result = await callFalModel("fal-ai/infinitalk", input, {
-      falKey: args.fal_key,
-    });
-
-    // Extract video data from the response
-    const videoData = result.data?.video;
-    const seed = result.data?.seed;
-
-    if (!videoData || !videoData.url) {
-      throw new Error(
-        `No video data in completed response: ${JSON.stringify(result.data)}`
-      );
-    }
-
-    const videoUrl = videoData.url;
-    const fileName = videoData.file_name || "infinitalk.mp4";
-
-    return JSON.stringify({
-      videos: [
-        {
-          url: videoUrl,
-          filename: fileName,
-          mimeType: "video/mp4",
-          filesize: videoData.file_size,
-        },
-      ],
-      message: "Infinitalk video generated successfully",
-      seed: seed,
-      requestId: result.requestId,
-    });
-  },
-};

package/src/utils/fal.utils.ts

@@ -1,53 +0,0 @@
-import { fal } from "@fal-ai/client";
-
-export async function callFalModel(
-  modelName: string,
-  input: any,
-  options: { falKey?: string; logs?: boolean } = {}
-) {
-  const { falKey, logs = true } = options;
-  const key = falKey || process.env.FAL_KEY;
-  if (!key) {
-    throw new Error(
-      "FAL_KEY is required. Provide it via fal_key parameter or FAL_KEY environment variable."
-    );
-  }
-
-  fal.config({
-    credentials: key,
-  });
-
-  console.error(`[${modelName}] Submitting request to FAL AI...`);
-
-  try {
-    const result = await fal.subscribe(modelName, {
-      input,
-      logs,
-      onQueueUpdate: (update) => {
-        if (update.status === "IN_PROGRESS") {
-          console.error(`[${modelName}] Status: ${update.status}`);
-          if (logs && "logs" in update && update.logs) {
-            update.logs.forEach((log) => {
-              console.error(`[${modelName}] ${log.message}`);
-            });
-          }
-        } else if (update.status === "IN_QUEUE") {
-          console.error(
-            `[${modelName}] Status: ${update.status} - Waiting in queue...`
-          );
-        }
-      },
-    });
-
-    console.error(`[${modelName}] Generation completed successfully`);
-
-    return result;
-  } catch (error: any) {
-    console.error(`[${modelName}] Error:`, error);
-    throw new Error(
-      `FAL AI ${modelName} generation failed: ${
-        error.message || JSON.stringify(error)
-      }`
-    );
-  }
-}