Skip to content

README.md

Latest commit

 

History

History
96 lines (65 loc) · 2.71 KB

README.md

File metadata and controls

96 lines (65 loc) · 2.71 KB
Raw

edream-sdk

copyright (c) 2025 e-dream, inc

the project and application has generally been renamed from e-dream to infinidream. this module, however, has not yet been renamed, but it works fine with the current servers.

local installation

If you need to setup locally to test or run code, follow BUILD.md documentation.

simple example

Get an API key from your profile on the server, and store it in a .env file. Then connect as follows:

import os
from dotenv import load_dotenv
from edream_sdk.client import create_edream_client

load_dotenv()

edream_client = create_edream_client(
    backend_url=os.getenv("BACKEND_URL", "https://api-alpha.infinidream.ai/api/v1"),
    api_key=os.getenv("API_KEY")
)
user = edream_client.get_logged_user()
print(user)

tests and more examples

There's a file to test api client:

python tests/run.py

this file has an example of every API call.

playlists and keyframes

Add many items to a playlist in one atomic request. The whole batch succeeds or nothing does, and the backend rejects it outright if any item is already on the playlist, so filter first:

from edream_sdk.types.playlist_types import PlaylistItemType

playlist = edream_client.get_playlist(playlist_uuid)
already = {i["dreamItem"]["uuid"] for i in playlist["items"] if i.get("dreamItem")}

added = edream_client.add_items_to_playlist(playlist_uuid, [
    {"type": PlaylistItemType.DREAM, "uuid": uuid}
    for uuid in dream_uuids if uuid not in already
])

At most 500 items per call. For a single item add_item_to_playlist is still the simpler choice.

Keyframes can be looked up by name rather than scanned for:

keyframe = edream_client.find_keyframe_by_name("00248=22588")

# or page through matches yourself -- search is a substring match
page = edream_client.get_keyframes(search="00248=", take=100)
print(page["count"], "matches")

AI generation

Use the Quick Start.

Set DREAM_UUID (a processed video dream) and STILL_UUID (a processed image dream) in your .env. See .env.example.

python tests/gen.py --algo ltx-i2v     # single algorithm
python tests/gen.py --all              # smoke test every endpoint
python tests/gen.py --all --timeout 7200

Available algorithms: animatediff, deforum, wan-t2v, wan-i2v, wan-i2v-lora, ltx-i2v, qwen-image, z-image-turbo, uprez, nvidia-uprez

For batch/playlist workflows see engines.

REST API

This Python SDK just wraps the servers' REST API. Its documentation is served by swagger on staging.