Image tools
POM AI has four image tools. One creates images from a description; three work on attachments that already exist in your media library.
The four tools
| Tool | What it does | Where |
|---|---|---|
| AI image generator | Creates a new image from a prompt | Media → AI generate images |
| AI image enhancer | Produces a more polished version of an existing image | Media → AI improve image |
| AI image bg remover | Removes or replaces the background of an existing image | Media → AI remove image background |
| AI metadata for images | Writes alternative text, title and caption in bulk | Media → AI generate image ALT, caption and title |
Each is enabled separately under Settings → POM AI → AI Tools and requires the upload_files capability.
Read this before using the two editing tools
AI improve image and AI remove image background can update supported references after generating the new image.
When Yes, update references and keep the original is selected, supported current-site references are updated in saved batches. Both attachments remain in the Media Library. Users without site media-reference permission can create copies instead.
Choose the intended scope before your first run: Replace the original or create a new attachment and Replace an image everywhere it appears.
Guides
Creating images
Editing existing images
- Choose the source image
- Improve an existing image
- Remove an image background
- Check a background removal
- Replace the original or create a new attachment
- Replace an image everywhere it appears
Metadata
Before publishing
What every image tool has in common
A model. Generation and improvement offer Nano Banana 2 (Google Gemini), Nano Banana Pro (Google Gemini), ChatGPT Sunburst (OpenAI) and ChatGPT Flare (OpenAI). Nano Banana 2 at 1K is the default. OpenAI adds a quality choice and displays exact dimensions for each ratio. Check the credit estimate before confirming. Background removal uses its own settings.
An estimate path. Two submit buttons: one prices the work and asks you to confirm, one generates immediately. Use the first for anything batched.
Batching. The three tools that work on existing attachments process a selection. The metadata tool accepts up to 48 images in one batch. Every image in a batch is its own charge.
Media library output. Results are ordinary WordPress attachments. Nothing is stored elsewhere.
A processing summary. Batch tools list each image with its outcome. Read it; a run can complete with some images failed.
Where images go
Generated and newly created images are uploaded to the media library like any other attachment, so any plugin that hooks attachment creation applies to them too.
When you choose to replace an original, the new file takes over the original's title, alternative text, caption and description, so existing accessibility work is not lost.
A safe first run
Start with the generator. It creates something new and touches nothing that exists:
- Enable AI image generator.
- Open Media → AI generate images.
- Use a short prompt, Nano Banana 2 (Google Gemini) at 1K and the default aspect ratio.
- Estimate, confirm, look at the result.
Only then try the editing tools, and try them first with No, create a new media library entry selected so existing pages keep their current images while you review the result.