Skip to main content

Triggering renders from Python

available from v4.0.15

To trigger a Lambda render using Python, install the remotion-lambda package using pip, by executing pip install remotion-lambda from your python project. Use the same version as the remotion version you are using from NPM, e.g. pip install remotion-lambda==4.0.15 (see newest version).

You first need to complete the Lambda setup.

Limitations​

Before v4.0.315: Sending large input props (>200KB) was not supported.

Starting from v4.0.315, the Python SDK automatically handles large input props by uploading them to S3 when they exceed AWS Lambda payload limits. Props larger than 200KB for video renders and 5MB for still renders are automatically compressed using the same logic as the JavaScript SDK. The SDK will automatically find or create an appropriate S3 bucket following Remotion's naming conventions.

Rendering a video​

Below is a snippet showing how to initiate a render request and get its status.

render_media.py
python
from remotion_lambda import RenderMediaParams, Privacy, ValidStillImageFormats
from remotion_lambda import RemotionClient
import os
from dotenv import load_dotenv
load_dotenv()
# Load env variables
REMOTION_APP_REGION = os.getenv('REMOTION_APP_REGION')
if not REMOTION_APP_REGION:
    raise Exception("REMOTION_APP_REGION is not set")
REMOTION_APP_FUNCTION_NAME = os.getenv('REMOTION_APP_FUNCTION_NAME')
if not REMOTION_APP_FUNCTION_NAME:
    raise Exception("REMOTION_APP_FUNCTION_NAME is not set")
REMOTION_APP_SERVE_URL = os.getenv('REMOTION_APP_SERVE_URL')
if not REMOTION_APP_SERVE_URL:
    raise Exception("REMOTION_APP_SERVE_URL is not set")
# Construct client
client = RemotionClient(region=REMOTION_APP_REGION,
                        serve_url=REMOTION_APP_SERVE_URL,
                        function_name=REMOTION_APP_FUNCTION_NAME)
# Set render request
render_params = RenderMediaParams(
    composition="react-svg",
    privacy=Privacy.PUBLIC,
    image_format=ValidStillImageFormats.JPEG,
    input_props={
        'hi': 'there'
    },
)
render_response = client.render_media_on_lambda(render_params)
if render_response:
    # Execute render request
    print("Render ID:", render_response.render_id)
    print("Bucket name:", render_response.bucket_name)
    # Execute progress request
    progress_response = client.get_render_progress(
        render_id=render_response.render_id, bucket_name=render_response.bucket_name)
    while progress_response and not progress_response.done:
        print("Overall progress")
        print(str(progress_response.overallProgress * 100) + "%")
        progress_response = client.get_render_progress(
            render_id=render_response.render_id, bucket_name=render_response.bucket_name)
    print("Render done!", progress_response.outputFile)

Large input props available from v4.0.315​

The Python SDK automatically handles large input props by uploading them to S3 when they exceed AWS Lambda payload limits:

  • Video/audio renders: Props larger than ~194KB
  • Still renders: Props larger than ~4.9MB

When large props are detected, they are automatically uploaded to S3 and referenced by the Lambda function. The SDK will find an existing Remotion bucket or create a new one automatically, following the same logic as the JavaScript SDK. This process is transparent to the user.

large_props_example.py
python
# Large input props are handled automatically
large_props = {
    'bigData': ['x' * 1000] * 250,  # ~250KB of data
    'description': 'This will be automatically uploaded to S3'
}
render_params = RenderMediaParams(
    composition="my-composition",
    input_props=large_props,  # Automatically compressed
    # ... other parameters
)
# The render works the same way, compression is handled internally
response = client.render_media_on_lambda(render_params)

Rendering an image​

Below is a snippet showing how to initiate a still image render. Note that it does not require monitoring the render progress.

render_still.py
python
from remotion_lambda import RenderStillParams, Privacy, ValidStillImageFormats
from remotion_lambda import RemotionClient
import os
from dotenv import load_dotenv
load_dotenv()
# Load env variables
REMOTION_APP_REGION = os.getenv('REMOTION_APP_REGION')
if not REMOTION_APP_REGION:
    raise Exception("REMOTION_APP_REGION is not set")
REMOTION_APP_FUNCTION_NAME = os.getenv('REMOTION_APP_FUNCTION_NAME')
if not REMOTION_APP_FUNCTION_NAME:
    raise Exception("REMOTION_APP_FUNCTION_NAME is not set")
REMOTION_APP_SERVE_URL = os.getenv('REMOTION_APP_SERVE_URL')
if not REMOTION_APP_SERVE_URL:
    raise Exception("REMOTION_APP_SERVE_URL is not set")
# Construct client
client = RemotionClient(region=REMOTION_APP_REGION,
                        serve_url=REMOTION_APP_SERVE_URL,
                        function_name=REMOTION_APP_FUNCTION_NAME)
# Set render still request
render_params = RenderStillParams(
    composition="still-helloworld",
    privacy=Privacy.PUBLIC,
    image_format=ValidStillImageFormats.JPEG,
    input_props={
        'message': 'Hello from props!'
    },
)
render_response = client.render_still_on_lambda(render_params)
if render_response:
    # Execute render request
    print("Render ID:", render_response.render_id)
    print("Bucket name:", render_response.bucket_name)
    print("Render done! File at ", render_response.url)

Breaking changes​

4.0.82​

  • The data field is now input_props.
  • RenderParams has been renamed to RenderMediaParams.
  • RenderProgress has been renamed to RenderMediaProgress.
  • RenderResponse is now RenderMediaResponse, and its fields have been updated: renderId is now render_id, and bucketName has been changed to bucket_name.
  • The following types have been added: CostsInfo, Privacy, ValidStillImageFormats, LogLevel, OpenGlRenderer, ChromiumOptions, CustomCredentialsWithoutSensitiveData, CustomCredentials, OutNameInputObject, PlayInBrowser, Download, DeleteAfter. These can be used for the fields on RenderMediaParams and RenderStillParams.

See also​