langchain-emaki

LangChain tools for emaki (emaki.ai), the Japanese-language AI video generator for Seedance 2.5 and Seedance 2.0. Give an agent Seedance 2.5 text to video, image to video and AI images on one emaki API key.

pip install langchain-emaki
export EMAKI_API_KEY=sk-...   # create one at https://emaki.ai/settings/apikeys

from langchain_emaki import EmakiImageGenerator, EmakiVideoGenerator

video = EmakiVideoGenerator()   # Seedance 2.5 by default
image = EmakiImageGenerator()   # nano_banana_2_lite by default

video.invoke({"prompt": "a paper boat drifting down a rain gutter, low angle, morning light", "duration": 5})
image.invoke({"prompt": "a paper-cut layered mountain range at dusk", "aspect_ratio": "16:9"})

Hand both to any tool-calling agent:

from langchain.agents import create_agent

agent = create_agent("openai:gpt-5", tools=[image, video])
agent.invoke({"messages": [("user", "Make a 16:9 still of a lighthouse in fog, then a 5-second clip that starts from it")]})

Tools

Pass a public https URL as image_url and the clip starts from that frame. A generated image works as that frame, which is how the agent above chains the two tools.

Set model= on either tool for another engine: seedance_2, seedance_2_fast, seedance_2_mini, minimax_h3, veo_3_1 for video (Seedance 2.0 is the previous generation); nano_banana_2, seedream_5_lite, gpt_image_2 for images. Each run costs credits on your emaki account, the same price as on the website.

Refusals come back to the agent as text rather than an exception: a missing or wrong key, an empty balance, a model that needs a purchase, another run still generating, or a prompt the filter refused. The agent can reword, wait or switch model instead of crashing the run. Video takes minutes; the default timeout is 15 minutes.

Finished clips can be sharpened with the AI video upscaler on the site.

日本語での使い方は シーダンス へ。

Built on the emaki client. MIT licensed. Made by emaki.