Skip to content

Create your first AI Shoot

With AI Shoot, you can produce high-quality on-model imagery at scale for your product description pages. This guide walks you through the key steps to get started.

Like in a traditional photoshoot, you need specific ingredients to run an AI shoot:

  • Products (SKUs)
  • Models
  • Specifications

The first step is to add products (shown as SKUs in the UI) to the system. Each SKU you add can be reused in any number of Shoots. From the SKUs tab, upload your product images in bulk and AutoRetouch groups them into SKUs for you, ready to review before you create them.

Learn more: Create and manage SKUs

A Shoot defines what you want to shoot and how. On the Shoot page, you need to configure:

  • Name - A descriptive name for your shoot
  • Number of Generations - The number of generations to produce per view (billable)

Shoot page configuration

Then:

  • Models - The AI models you want to use
  • SKUs - The products you want to shoot
  • Views - The angles you want to produce, including their framing and post-processing workflows
  • Post-processing - Optional, but lets you customize the background color (the color you get is deterministic), margin and export format

Shoot views and post-processing configuration

Once configured, create the Shoot to proceed to Look creation.

Looks are combinations of a model, SKUs, and styling options. Each Look will produce all the views configured at the Shoot level.

Look creation interface

You can create as many Looks as needed, but we recommend starting with groups of 5 or fewer as you get familiar with the system.

When you select a Hero SKU by clicking the star icon, the Look name will automatically match that SKU. You can override this with a custom value. Your final results will be named using this value combined with the view index.

Hero SKU selection

Configure styling

It’s possible to configure styling options. Available options differ depending on the SKU category.

  • Tucking: specify a tucking style
  • Shoe break: specify how trousers fall on shoes
  • Open/Close: specify if the SKU should be opened or closed (e.g., shirt, jackets)

If you don’t specify styling options (which is the default behavior), the styling will be handled automatically during the AI generation process.

Once you’ve created your Looks, the system automatically starts generating your images. Generation typically takes 1-4 minutes per Look. You will land on the results page, which can be viewed by view (one row for each view produced) or by Look (one row per look, with all its views).

Results ready for selection

To enter the selection flow, click “Select AI Result” for an image; the system will take you through the selection for each image of the Look it belongs to.

If a result isn’t exactly how you want it, you can use the Edit feature, which allows you to state in natural language what needs to be fixed. This feature is accessible in the selection flow, and can only be used on results that haven’t been selected yet.

Editing AI results

Edits are relevant for small corrections, and your instructions should remain laser focused to produce the best results. With this feature you can for instance:

  • Lengthen a garment
  • Adjust hands position
  • Change the model’s mood
  • Style a garment

Every Edit is made on top of the original AI Result (you can’t edit an edit). Once an edit is applied, a version selector appears on the result, letting you switch between the original and the edited version.

When you are happy with one of your results, you can select it using the “Select AI Result” button, under the image you want to select. You must select at least one image per Look for the next step to be available (post-processing, or result download).

Selecting AI results

It’s possible to regenerate all results if you haven’t been able to achieve what you wanted, by clicking the “Regenerate” button. We’d advise trying an edit first, and if that doesn’t work, modifying the Look or using different styling options, as regenerating without changing anything tends to produce results similar to what you already have.

If you have edited a result using the Edit feature, make sure to choose the version you want via the version selector before clicking the “Select AI Result” button, as the result that will be selected is the one currently active.

Selecting the right version

Once an image is selected, it will go through the post-processing workflow configured at the Shoot level (if one exists). This sets your image to your target specifications. This process can take several minutes depending on the workflow used and system load.

Once processing is complete, download your results from either the “by view” or “by look” page.

Download results

Looking to automate your AI Shoot workflow? See the AI Shoot API documentation to manage SKUs and retrieve results programmatically.