Generate video in Python

Completed

Tip

See the Text and images tab for more details!

To build applications that generate videos programmatically, you can use the OpenAI Python SDK with your Sora 2 deployment in Microsoft Foundry. Video generation is an asynchronous process—you submit a job, poll for status, and download the result when it's ready.

Generate a video

Video generation follows a three-step pattern: create the job, poll for completion, and download the result.

import time

# Create the video generation job
video = client.videos.create(
    model="sora-2",
    prompt="A robot walks through a rainy city street at dusk, neon signs reflecting in puddles",
    size="1280x720",
    seconds="4",
)

print(f"Video creation started. ID: {video.id}")

# Poll for completion
while video.status not in ["completed", "failed", "cancelled"]:
    print(f"Status: {video.status}. Waiting...")
    time.sleep(20)
    video = client.videos.retrieve(video.id)

# Download when complete
if video.status == "completed":
    content = client.videos.download_content(video.id, variant="video")
    content.write_to_file("output.mp4")
    print("Video saved to output.mp4")

Generate video from a reference image

To use an image as a starting frame, pass it to the input_reference parameter. The image resolution must match the target video size:

video = client.videos.create(
    model="sora-2",
    prompt="The camera slowly pans across the landscape as clouds drift overhead",
    size="1280x720",
    seconds="4",
    input_reference=open("landscape.png", "rb"),
)

Note

Reference images containing human faces are currently rejected. Use images of landscapes, objects, or animated characters instead.

Remix an existing video

To modify an existing video while preserving its structure, use the remix method with the original video's ID:

video = client.videos.remix(
    video_id="video_abc123",
    prompt="Change the color palette to warm sunset tones",
)

Handle job states

Video jobs can return these status values:

Status	Description
queued	Job is waiting to be processed
in_progress	Video is being generated
completed	Video is ready for download
failed	Generation failed (check error details)
cancelled	Job was canceled

When a job fails, check video.error for details about what went wrong.

Key considerations

• Rate limits: You can run up to two video creation jobs simultaneously
• Job expiration: Completed videos are available for download for 24 hours
• Resolution requirements: Reference images must match the target video resolution exactly
• Content filtering: Prompts are subject to content moderation; harmful content won't generate

In the exercise that follows, you'll apply these techniques to build a complete video generation application.