> ## Documentation Index
> Fetch the complete documentation index at: https://developers.apob.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Generation workflow

> Plan a generation, submit it once, follow the existing task, and retrieve the finished media.

<Steps>
  <Step title="Choose the operation">
    Decide whether you are creating new media, editing an existing asset, or
    extending footage. Select a ready character model if the workflow requires
    one. An extension adds footage to its source; account for both when planning
    the final video.
  </Step>

  <Step title="Prepare inputs and settings">
    [Prepare your media](/guides/preparing-media), choose references, and check the
    available quality, aspect ratio, duration, and audio options. Verify the units
    shown by your client: a ratio is different from pixel dimensions, and time
    values can be expressed in seconds or milliseconds.
  </Step>

  <Step title="Review credits and submit once">
    Check [credits and billing](/credits-and-billing), including automatic reload.
    Submit the operation once and save its receipt and task identifier.
  </Step>

  <Step title="Follow the task">
    Check the existing operation until it reaches a terminal outcome. Space out
    reads and use a reasonable waiting limit in your client. If your client stops
    waiting, the server-side task may still be running.
  </Step>

  <Step title="Retrieve and inspect outputs">
    After successful completion, retrieve the final media and check its visual
    content, dimensions, duration, and audio. Download the outputs you need while
    their links remain valid.
  </Step>
</Steps>

## Write a clear creative brief

Describe the subject, action, setting, composition, and desired style. For video,
include camera movement and how the action develops over time. For an edit,
describe what should change and what should remain recognizable.

For example, a plain-language creative brief could be:

> A woman in a red coat walks through a snowy courtyard. The camera follows her
> at shoulder height. Soft morning light, realistic movement, and a calm mood.

Adapt the brief to your client's prompt format. Some workflows require structured
reference or voice notation rather than ordinary prose alone.

## Plan speech and sound

For ordinary image-to-video and extension workflows, place dialogue in the prompt
using the supported voice-section syntax, and select the intended voices.
Talking Avatar and Lip Sync have their own speech and audio input flows. Follow
the selected workflow's instructions rather than reusing the same speech inputs
across all operations.

## Handle an interruption

<Warning>
  Do not automatically repeat a generation after a timeout or an uncertain
  response. The task may already exist and a second submission can charge again.
</Warning>

If you have a task identifier, continue through its result flow. Otherwise,
reconcile the account's generation history before deciding whether to submit
again. See [tasks and results](/concepts/tasks-and-results) and
[troubleshooting](/troubleshooting).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.