san8888/emaki repository overview

⁠emaki CLI image

Official SDK for emaki (emaki.ai), the Japanese-language AI video generator for Seedance 2.5 and Seedance 2.0. Generate Seedance 2.5⁠ and Seedance 2.0⁠ clips, and images, from the command line (a Docker image of the emaki CLI) with the API key of your emaki⁠ account.

docker run --rm -e EMAKI_API_KEY san8888/emaki "a paper lantern drifting down a night river, slow dolly-in"
# prints the URL of the finished MP4

Each run submits, waits for the result and prints the URL of the finished file. Every run costs credits on your emaki account, the same price as on the website.

⁠Install

docker pull san8888/emaki

A static Go binary on an empty base image, for linux/amd64 and linux/arm64. Source: github.com/hanshs474/emaki-docker⁠. Exit codes: 0 ok, 1 failure, 2 usage, 3 key problem, 4 out of credits or the model needs a purchase.

⁠API key

Sign up on emaki.ai⁠, create a key at emaki.ai/settings/apikeys⁠, then pass it or export it:

export EMAKI_API_KEY=sk-...

A key runs on your account exactly like the site does: same models, same credits, and every run shows up in your library on emaki.ai.

⁠Video

The default model is seedance_2_5 (5 seconds, 480p, 16:9, with sound). Flags map to the request parameters; anything without a flag goes in -params as JSON:

# Seedance 2.5: 8 seconds, 720p, vertical
docker run --rm -e EMAKI_API_KEY san8888/emaki -duration 8 -resolution 720p -aspect 9:16 "a street drummer in the rain, handheld"

# Seedance 2.0
docker run --rm -e EMAKI_API_KEY san8888/emaki -model seedance_2 -duration 5 "a koi pond seen from above, petals falling"

# any other parameter as JSON
docker run --rm -e EMAKI_API_KEY san8888/emaki -params '{"generate_audio":false}' "a quiet library at dawn"

⁠Image to video

Pass image, a public https URL, and it becomes the first frame; the prompt then describes the motion. This is the same image to video⁠ flow as on the site:

docker run --rm -e EMAKI_API_KEY san8888/emaki -first-frame https://example.com/statue.jpg "the camera slowly orbits the statue"

On Seedance 2.5, last_frame_image pins the closing frame as well.

⁠Images

The default image model is nano_banana_2_lite, the cheapest run on the site:

docker run --rm -e EMAKI_API_KEY san8888/emaki -image "a ramen stall in the rain, neon reflections"
docker run --rm -e EMAKI_API_KEY san8888/emaki -image -model nano_banana_2 -aspect 16:9 "a ukiyo-e style wave over a modern city"

Image models: nano_banana_2_lite, nano_banana_2 (resolution 1K / 2K / 4K), seedream_5_lite (quality basic / high / ultra), gpt_image_2 (resolution 1K / 2K / 4K). All take aspect_ratio.

⁠Submit now, collect later

docker run --rm -e EMAKI_API_KEY san8888/emaki -no-wait "a fox trotting through fresh snow"   # prints the run id
docker run --rm -e EMAKI_API_KEY san8888/emaki -status <id>                                  # JSON: status, videoUrl, failureCode
docker run --rm -e EMAKI_API_KEY san8888/emaki -credits                                      # free to call

A run waits up to 15 minutes by default, polling every 5 seconds; change that with -timeout 30m.

⁠Errors worth catching

Errors print to stderr prefixed with their kind:

• auth: no key, or the key is wrong or deleted (HTTP 401)
• quota: not enough credits left for this run (HTTP 402)
• plan: this model unlocks after the account has bought credits (HTTP 403)
• busy: another generation on this account is still in progress (HTTP 409)
• rejected: the content filter refused the prompt, or the account is cooling down after a refusal; reword rather than retry
• validation: a parameter is missing or out of range; the message says which
• failed: the run was accepted and then failed; the failure code says why
• timeout: the wait ran out; the run may still finish in your emaki library

Anything else (a network failure, a 5xx) comes back as other with the service's message.

⁠More on emaki

Clips you make here can be sharpened afterwards with the AI video upscaler⁠ on the site.

日本語での使い方は シーダンス⁠ へ。

MIT licensed. Made by emaki⁠.