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

> ## Agent Instructions
> Follow the content contract in AGENTS.md.
> Treat text as the canonical explanation; video and screenshots enhance it.
> Do not publish unverified product behavior or duplicate an existing canonical article.

# Create and manage images in the Image workspace

> Generate image directions for each funnel stage, expand the strongest ideas, and organize the assets you want to use.

## What this article shows you

Use the Image workspace to turn a creative direction into a batch of images, expand the strongest result, and organize the assets you want to use next.

The most useful loop is:

1. Generate five distinct directions.
2. Choose one or two that are worth developing.
3. Create five to 10 more variations of each promising direction.
4. Edit, save, and download the best options.

## Practical use cases

<CardGroup cols={2}>
  <Card title="Top-of-funnel hooks" icon="megaphone">
    Create scroll-stopping concepts that introduce the problem, audience, or
    curiosity angle to cold traffic.
  </Card>

  <Card title="Middle-of-funnel education" icon="book-open-check">
    Turn benefits, mechanisms, comparisons, or proof into images that help an
    interested prospect understand the offer.
  </Card>

  <Card title="Bottom-of-funnel action" icon="mouse-pointer-click">
    Create offer, urgency, guarantee, or call-to-action images for people who
    already understand the product.
  </Card>

  <Card title="Native-style creative" icon="newspaper">
    Produce images that feel natural in the feed, such as screenshots,
    founder-style posts, notes, reviews, or editorial layouts.
  </Card>

  <Card title="Email and lifecycle images" icon="mail">
    Create product, offer, announcement, and supporting visuals for email and
    retention campaigns.
  </Card>

  <Card title="Scale a promising direction" icon="repeat-2">
    Generate five options, choose a direction, then create five to 10 more
    variations. Repeat until you have the 10–20 assets needed for the test.
  </Card>
</CardGroup>

## Watch the walkthrough

<Frame caption="Owen walks through the complete Image workspace workflow, from finished results and remixing through strategy, generation, editing, downloads, and animation. Duration: 11:30.">
  <video className="w-full h-auto rounded-xl" controls preload="metadata" poster="/images/tutorials/covers/use-the-image-workspace.png" playsInline alt="Owen demonstrates the complete Image workspace workflow" aria-label="Narrated Image workspace walkthrough presented by Owen">
    <source src="https://stefanbrain.com/api/video/adaptive/tutorials/use-the-image-workspace/use-the-image-workspace.mp4" type="video/mp4" />

    <track kind="captions" src="/captions/tutorials/use-the-image-workspace.vtt" srcLang="en" label="English" default />

    <a href="https://stefanbrain.com/api/video/adaptive/tutorials/use-the-image-workspace/use-the-image-workspace.mp4">
      Watch the Image workspace walkthrough
    </a>
  </video>
</Frame>

English captions: [download the VTT file](/captions/tutorials/use-the-image-workspace.vtt).

### Video chapters

| Time | Chapter |
| - | - |
| **00:00** | Review finished results and the variation loop |
| **00:49** | Remix a strong creative direction |
| **01:38** | Edit an image and add reference assets |
| **02:21** | Review the Meta copy fields |
| **02:33** | Build the first image batch |
| **03:33** | Set output, context, and research controls |
| **05:14** | Build a full-funnel creative strategy in Chat |
| **07:02** | Turn the strategy into images and reusable prompts |
| **08:41** | Save, download, inspect, and reuse results |
| **09:19** | Animate a static image |
| **09:47** | Review newly generated results |
| **10:18** | Prompting checklist and recap |

## How it works and when to use it

Select **Images** in the sidebar to open the workspace.

* **Ads** is for ad-oriented images.
* **Anything** is for general images or direct edits to an attached image.
* **Product** adds stable product context.
* **Project** adds the direction for the current campaign or workstream.
* **Add assets** attaches product shots, inspiration, or an image you want to edit.
* **Aspect** and **Variations** control the requested output set.

You can begin in Chat when you need ideas first. Ask StefanBrain to propose image directions or write generation prompts, then bring the chosen direction into the Image workspace.

## Before you start

Have these ready:

* The audience and funnel stage.
* The job the image needs to do.
* The Product or Project that contains the useful context.
* Any product shots or references that should guide the result.
* The placement or aspect ratio, if it is already known.

## Follow along

<Steps>
  <Step title="Choose a direction">
    Start with the audience, funnel stage, and intended outcome. If you need help developing ideas, ask in Chat:

    ```text theme={"system"}
    Give me 10 image directions for [audience] at [top, middle, or bottom] of funnel. The goal is [outcome]. For each direction, explain the hook and write a prompt I can use in the Image workspace.
    ```

    Choose the directions that are meaningfully different, not small rewrites of the same concept.
  </Step>

  <Step title="Open Images and add useful context">
    Select **Images**, then choose **Ads** or **Anything**.

    Open **More** to select the relevant Product or Project. Use **Add assets** when a product shot, reference image, or existing asset should guide the output.

    Use the smallest relevant context set. More context is not automatically better.
  </Step>

  <Step title="Set the output controls">
    Choose **Auto** under **Aspect** when the format can follow the idea. Choose an exact ratio when the placement already defines it.

    Set **Variations** to five for the first pass. This gives you enough range to compare without committing a large batch to an unproven direction.

    <Frame caption="Set the image type, aspect ratios and first-pass variation count before you generate.">
      <img src="https://mintcdn.com/copyaccelerator/0b7kYwfWqo7VfDOQ/images/knowledge-base/images/image-output-controls.jpg?fit=max&auto=format&n=0b7kYwfWqo7VfDOQ&q=85&s=07a6b3a5c0be00755a4edb2cd5cd0f19" alt="Image workspace composer set to Ads, square and vertical aspect ratios, and five variations" width="890" height="124" data-path="images/knowledge-base/images/image-output-controls.jpg" />
    </Frame>
  </Step>

  <Step title="Generate the first five directions">
    Describe the outcome and the few details that matter:

    ```text theme={"system"}
    Create five native-style static ad concepts for [audience] at [funnel stage]. Use @Product and the attached product reference. The goal is [outcome]. Keep the product packaging accurate, and vary the hook, framing, and composition across the five images.
    ```

    Select **Generate** once. Pending cards appear while the run is in progress, followed by the completed images.
  </Step>

  <Step title="Expand the strongest direction">
    Open the results and choose one or two directions worth developing. Then use the result as the reference for the next batch:

    ```text theme={"system"}
    Create 5–10 more variations of this direction. Keep [the core hook, product treatment, or composition] and vary [the opening line, framing, setting, or layout].
    ```

    This is where variation becomes useful: preserve what made the direction interesting and change only the dimensions you want to test.
  </Step>

  <Step title="Edit, save, and download">
    Use **Edit** when one image is close but needs a targeted change. Say what should stay and what should change.

    Select the heart on images you want to keep. Open **Saved** to return to the shortlist, select the assets you need, then choose **Download** and **Images ZIP**.
  </Step>
</Steps>

## Review the batch

Before handing the images off, check:

* Does each image match the intended funnel stage and job?
* Are the product, visible text, and brand details accurate?
* Is there enough difference between variations to learn something from the test?
* Did the second batch preserve the part of the first direction you liked?
* Are the saved assets the ones you actually want the next person to use?

## Common mistakes and troubleshooting

<AccordionGroup>
  <Accordion title="The results all look like the same idea">
    Ask for distinct hooks, settings, formats, or visual mechanisms in the first batch. Save small variations for the second pass.
  </Accordion>

  <Accordion title="The results are too generic">
    Add the audience, funnel stage, desired outcome, and one relevant Product, Project, or reference asset. Remove context that does not support this job.
  </Accordion>

  <Accordion title="The product details are wrong">
    Attach an accurate product reference, select the correct Product, and name the details that must remain unchanged.
  </Accordion>

  <Accordion title="The variation count is higher than expected">
    Check the selected aspect ratios. The workspace keeps the variation count at least as high as the number of concrete ratios selected.
  </Accordion>

  <Accordion title="A run fails or a failed card remains">
    Check the inline message. Use **Try again** when it appears, or **Reuse prompt** to restore and revise the earlier request.
  </Accordion>
</AccordionGroup>

## Next steps

* [Browse Image workspace guidance](/knowledge-base/images)
* [Develop native ad concepts from a reference ad](/knowledge-base/images/create-native-ad-concepts-from-winning-ad)
* [Choose between Products, Projects, and Chats](/knowledge-base/get-started/products-projects-and-chats)
* [Find campaign angles before creating assets](/knowledge-base/creative-strategist-research/use-angle-finder)


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