SmartyPress AI Engine Documentation Article

Editing and Regenerating Images

Edit AI images with correction prompts and create attachments from the Media Library.

Editing And Regenerating Images is the focus of this documentation page. Edit AI images with correction prompts and create attachments from the Media Library. Follow the sections below to regenerate images and manage AI attachment metadata.

Edit existing AI-generated images with a correction prompt, or create new images directly from the Media Library.

Regenerating an Image

If you want to modify an AI-generated image:

  1. In the Image Generation page, find your previously generated image in the table.
  2. Click the Regenerate button in the Actions column.
  3. In the Regenerate Image modal, enter a correction prompt describing what to change (e.g., “Make the background blue instead of green”).
  4. Click Regenerate.
  5. The updated image replaces the original file on the same Media Library attachment (same URL; browsers may need a cache refresh to show the new pixels).
Regenerate Image modal in SmartyPress AI with a correction prompt textarea
Describe only the change you want; the original composition and style stay the same.
How it works: The plugin uses the provider’s image edit API when the selected model supports editing. The original image file, original generation prompt, and your correction prompt are sent together, producing an edited version that keeps the overall style and composition unless you ask to change them.
Note: If the selected image model does not support editing, regeneration fails with a message asking you to choose an image edit model in Image Generation settings.

“Create Attachment with AI” Button

On the Media Library page (wp-admin/upload.php), you will see a “Create attachment with AI” button. Click it to open the Image Generation screen and create a new image that is immediately added to your library.

WordPress Media Library page with the Create attachment with AI button next to Add Media File
The button opens the Image Generation screen and adds the result straight to this library.

AI Metadata on Attachments

Every AI-generated image stores metadata:

  • smartypress_ai_prompt — the original generation prompt (appended with edit notes after regeneration)
  • smartypress_ai_model — which model was used
  • smartypress_ai_resolution — image resolution
  • smartypress_ai_generated — flag marking it as AI-generated

Only attachments marked with smartypress_ai_generated can be regenerated from the Image Generation page. The stored prompt provides context for the edit request.

Editing And Regenerating Images Workflow Checklist

The editing and regenerating images workflow should be treated as a repeatable process, not a one-time setting. Start by confirming that the SmartyPress AI Engine module related to this page is enabled, then check the default provider, model, language, and post type settings before you run the feature on production content. This keeps generated output predictable and makes it easier to compare results between OpenAI, DeepSeek, OpenRouter, Yandex GPT, and GigaChat.

Use this checklist whenever you document, test, or train another editor on Editing and Regenerating Images. It helps keep the setup aligned with WordPress permissions, Rank Math SEO fields, and the editorial workflow used by your site. If a result looks different from what you expected, repeat the same checklist with a smaller prompt and a lower temperature before changing global settings.

  1. Confirm that the required SmartyPress AI module is enabled in the Components screen.
  2. Open Settings and verify that the selected provider has a valid API key and a tested connection.
  3. Choose the model that fits the task: fast models for drafts, premium models for final marketing copy, and web-search models for research.
  4. Review the output language, temperature, max tokens, and prompt variables before generating content.
  5. Save the result as a draft first, then review headings, links, facts, and SEO fields before publishing.

Recommended Settings

For most editing and regenerating images tasks, start with a balanced model such as GPT-4o Mini, GPT-4o, DeepSeek Chat, or a carefully selected OpenRouter model. Keep temperature around 0.5 to 0.8 for documentation and tutorials, because these pages need clarity more than experimentation. Increase temperature only when you need more creative headlines, introductions, or alternative wording.

Use a clear default language and keep prompt variables consistent across articles. For documentation pages, a neutral or professional writing tone usually works best. If the generated result is too short, increase the number of sections or paragraphs per section rather than asking the model to “write more” after the fact. Structured settings create more stable output and make Editing and Regenerating Images easier to reproduce later.

Scroll to Top