image_to_video

Instructions

Implementation Reference

• src/ghibli.ts:83-121 (handler)

  Core handler function in GhibliClient that performs the image-to-video conversion by making a POST request to the external API (/api/video) and returns the task ID.

  async imageToVideo(
      sourceImage: string, 
      prompt: string = "in the style of ghibli", 
      aspectRatio: string = "9:16",
      negativePrompt: string = "bad prompt",
      apiKey: string
    ): Promise<string> {
      const payload = {
        prompt,
        task_type: "img2video-14b",
        negative_prompt: negativePrompt,
        aspect_ratio: aspectRatio,
        image: sourceImage
      };
  
      // 打印请求信息
      process.stderr.write(`\n[Request] POST ${this.baseUrl}/api/video\n`);
      process.stderr.write(`[Headers] ${JSON.stringify(this.getHeaders(apiKey), null, 2)}\n`);
      process.stderr.write(`[Payload] Image length: ${sourceImage.length}, Prompt: ${prompt}\n`);
  
      const response = await fetch(`${this.baseUrl}/api/video`, {
        method: 'POST',
        headers: this.getHeaders(apiKey),
        body: JSON.stringify(payload)
      });
  
      // 打印响应状态
      process.stderr.write(`[Response] Status: ${response.status} ${response.statusText}\n`);
  
      if (!response.ok) {
        const error = `API request failed: ${response.statusText}`;
        process.stderr.write(`[Error] ${error}\n`);
        throw new Error(error);
      }
  
      const result = await response.json();
      process.stderr.write(`[Response Data] ${JSON.stringify(result, null, 2)}\n`);
      return result.data?.task_id;
    }

• src/index.ts:48-79 (registration)

  Registers the 'image_to_video' tool in the MCP server's list of available tools, defining its name, description, and input schema.

  {
    name: "image_to_video",
    description: "Convert image to animated video",
    inputSchema: {
      type: "object",
      properties: {
        image: {
          type: "string",
          description: "Base64 encoded source image or image URL"
        },
        prompt: {
          type: "string",
          description: "Optional prompt for video generation"
        },
        aspect_ratio: {
          type: "string",
          description: "Aspect ratio of the output video (e.g. '9:16')",
          default: "9:16"
        },
        negative_prompt: {
          type: "string",
          description: "Negative prompt to guide generation",
          default: "bad prompt"
        },
        api_key: {
          type: "string",
          description: "API key for authentication"
        }
      },
      required: ["image", "api_key"]
    }
  },

• src/index.ts:123-155 (handler)

  MCP CallToolRequest handler case for 'image_to_video' that extracts arguments, validates inputs, delegates to GhibliClient.imageToVideo, and formats the response.

  case "image_to_video": {
    const image = String(request.params.arguments?.image);
    const prompt = String(request.params.arguments?.prompt || "in the style of ghibli");
    const aspectRatio = String(request.params.arguments?.aspect_ratio || "9:16");
    const negativePrompt = String(request.params.arguments?.negative_prompt || "bad prompt");
    const apiKey = String(request.params.arguments?.api_key);
  
    if (!image) {
      throw new Error("Source image cannot be empty");
    }
    if (!apiKey) {
      throw new Error("API key cannot be empty");
    }
  
    try {
      const result = await ghibliClient.imageToVideo(
        image, 
        prompt, 
        aspectRatio,
        negativePrompt,
        apiKey
      );
      return {
        content: [{
          type: "text",
          text: `Video generation taskid: ${result}`
        }]
      };
    } catch (error) {
      const errorMessage = error instanceof Error ? error.message : 'Unknown error';
      throw new Error(`Video generation failed: ${errorMessage}`);
    }
  }

• src/index.ts:51-77 (schema)

  Input schema defining the parameters for the 'image_to_video' tool, including types, descriptions, defaults, and required fields.

  inputSchema: {
    type: "object",
    properties: {
      image: {
        type: "string",
        description: "Base64 encoded source image or image URL"
      },
      prompt: {
        type: "string",
        description: "Optional prompt for video generation"
      },
      aspect_ratio: {
        type: "string",
        description: "Aspect ratio of the output video (e.g. '9:16')",
        default: "9:16"
      },
      negative_prompt: {
        type: "string",
        description: "Negative prompt to guide generation",
        default: "bad prompt"
      },
      api_key: {
        type: "string",
        description: "API key for authentication"
      }
    },
    required: ["image", "api_key"]