diff --git a/src/agentscope_bricks/components/__init__.py b/src/agentscope_bricks/components/__init__.py index eb6e51c..894cb42 100644 --- a/src/agentscope_bricks/components/__init__.py +++ b/src/agentscope_bricks/components/__init__.py @@ -95,6 +95,9 @@ from agentscope_bricks.components.generations.image_text_interleave_generation_wan26 import ( # noqa WanImageInterleaveGeneration, ) +from agentscope_bricks.components.generations.async_video_to_video_wan26 import ( # noqa + VideoToVideoW26Submit, +) class McpServerMeta(BaseModel): @@ -150,7 +153,7 @@ class McpServerMeta(BaseModel): QwenImageGen, QwenImageEdit, QwenImageEditNew, - ], # ← 新增 QwenImageEditNew + ], ), "modelstudio_web_search": McpServerMeta( instructions="提供实时互联网搜索服务，提供准确及时的信息检索功能", @@ -173,6 +176,7 @@ class McpServerMeta(BaseModel): WanVideoFetch, ImageEditWan26, WanImageInterleaveGeneration, + VideoToVideoW26Submit, ], ), "modelstudio_z_image": McpServerMeta( diff --git a/src/agentscope_bricks/components/generations/async_video_to_video_wan26.py b/src/agentscope_bricks/components/generations/async_video_to_video_wan26.py new file mode 100644 index 0000000..4f965e5 --- /dev/null +++ b/src/agentscope_bricks/components/generations/async_video_to_video_wan26.py @@ -0,0 +1,151 @@ +# -*- coding: utf-8 -*- +import uuid +import json +from http import HTTPStatus +from typing import Any, Optional, Dict +import aiohttp +from mcp.server.fastmcp import Context +from pydantic import BaseModel, Field + +from agentscope_bricks.base.component import Component +from agentscope_bricks.utils.tracing_utils.wrapper import trace +from agentscope_bricks.utils.api_key_util import ApiNames, get_api_key +from agentscope_bricks.utils.tracing_utils import TracingUtil + + +DASHSCOPE_API_BASE = "https://dashscope.aliyuncs.com/api/v1" + + +class VideoToVideoGenerationWan26Input(BaseModel): + """ + Input model for Alibaba Cloud Wan 2.6 Reference Video to Video Generation. + """ + + prompt: str = Field( + ..., + description="文本提示词，描述期望生成的视频内容。" + "使用 character1, character2 模型仅通过此方式引用参考视频中的角色。", + ) + reference_video_urls: list[str] = Field( + ..., + description="参考视频 URL 列表（1~3个），每个视频应只包含一个角色。传入多个视频时，按照数组顺序定义视频角色的顺序。" + "第1个对应 character1，第2个对应 character2，依此类推。", + ) + negative_prompt: Optional[str] = Field( + default=None, + description="反向提示词，长度不超过500字符，超过部分会自动截断。", + ) + size: Optional[str] = Field( + default="1920*1080", + description="视频分辨率，可选分辨率：720P、1080P对应的所有分辨率。" + "默认为 1920*1080。", + ) + duration: Optional[int] = Field( + default=5, + description="视频时长，单位秒。可选值：5 或 10。默认为 5。", + ) + shot_type: Optional[str] = Field( + default="single", + description="镜头类型：'single'（单镜头）或 'multi'（多镜头）。默认为 'single'。" + "参数优先级： shot_type > prompt 。例如，若 shot_type 设置为'single'，" + "即使 prompt 中包含“生成多镜头视频”，模型仍会输出单镜头视频。", + ) + watermark: Optional[bool] = Field( + default=False, + description="是否添加右下角水印 'AI 生成'。默认为 False。", + ) + seed: Optional[int] = Field( + default=None, + description="随机种子，取值范围 [0, 2147483647]，用于提升可复现性。", + ) + ctx: Optional[Context] = Field( + default=None, + description="HTTP request context containing headers " + "for mcp only, don't generate it", + ) + + +class VideoToVideoGenerationWan26Output(BaseModel): + task_id: str = Field( + ..., + description="异步视频生成任务 ID，可用于后续查询任务状态和结果。", + ) + request_id: Optional[str] = Field( + default=None, + description="请求唯一 ID，用于日志追踪。", + ) + + +class VideoToVideoW26Submit( + Component[ + VideoToVideoGenerationWan26Input, + VideoToVideoGenerationWan26Output, + ], +): + name: str = "modelstudio_video_to_video_wan26_submit_task" + description: str = ( + "[版本: wan2.6] 通义万相参考生视频模型（wan2.6-r2v]）异步任务提交工具，" + "基于参考视频的角色形象生成新视频，返回 task_id 供后续查询。" + ) + + @trace(trace_type="AIGC", trace_name="video_to_video_wan26_submit_task") + async def arun( + self, + args: VideoToVideoGenerationWan26Input, + **kwargs: Any, + ) -> VideoToVideoGenerationWan26Output: + try: + api_key = get_api_key(ApiNames.dashscope_api_key, **kwargs) + except AssertionError: + raise ValueError("Please set valid DASHSCOPE_API_KEY!") + + input_data = { + "prompt": args.prompt, + "reference_video_urls": args.reference_video_urls, + } + if args.negative_prompt is not None: + input_data["negative_prompt"] = args.negative_prompt + parameters = { + "size": args.size, + "duration": args.duration, + "shot_type": args.shot_type, + "watermark": args.watermark, + } + if args.seed is not None: + parameters["seed"] = args.seed + + payload = { + "model": "wan2.6-r2v", + "input": input_data, + "parameters": parameters, + } + + headers = { + "Authorization": f"Bearer {api_key}", + "Content-Type": "application/json", + "X-DashScope-Async": "enable", + } + + create_url = f"{DASHSCOPE_API_BASE}/services/aigc/video-generation/video-synthesis" # noqa + + async with aiohttp.ClientSession() as session: + async with session.post( + create_url, + headers=headers, + json=payload, + ) as resp: + if resp.status != HTTPStatus.OK: + error_text = await resp.text() + raise RuntimeError( + f"Failed to create video task: {error_text}", + ) + + response_json = await resp.json() + task_id = response_json["output"]["task_id"] + + request_id = TracingUtil.get_request_id() or str(uuid.uuid4()) + + return VideoToVideoGenerationWan26Output( + task_id=task_id, + request_id=request_id, + )