Settings, chrome and shortcuts

The plugin logo at the far left. It is a label only and marks the editor, hovering shows the tooltip WunderPaint.

A small word before the file name that reads Creating: for a brand-new document or Editing: for an existing one, so you always know whether you are making a fresh image or changing an existing one.

The bold name of the open document. It updates when you rename or save under a new name.

A small badge shown only when the open document is a dynamic template. It reminds you that saving updates the template itself rather than creating a normal image.

Tip. Hover it to read: saving updates the template.

The pixel size of the canvas, shown faintly after the name (for example · 1080×1080). A quick reminder of the document dimensions.

A small dot that appears next to the size when the document has changes you have not saved yet. It disappears once you save.

Steps one change backwards. Greyed out when there is nothing left to undo.

Re-applies the last change you undid. Greyed out when there is nothing to redo.

Exports and downloads the image immediately using the settings from your last export (format, quality and scale), so you do not have to reopen the export dialog. Formats that need extra choices (GIF, APNG, WebM, PDF and extension formats) reopen the export dialog instead.

Tip. PSD exports run through the real layer writer, not a flattened PNG.

Press and hold to temporarily show the saved original underneath your edits, so you can see the before and after. Release to return to your edits. Only appears when the document was opened from an existing image.

Switches the editor between dark and light appearance. Your choice is remembered for next time.

Leaves the editor. The label reads Back to Post when you arrived from a post, Back to Media Library otherwise, and Cancel inside a nested Smart Object edit. If you have unsaved changes it first asks whether to discard them.

Shows your design inside a real post so you can judge it in context, and fills any dynamic fields with that post's content. Click again to leave the preview.

Saves your work as a fresh attachment in the media library without touching the original. Hidden in element-bound embeds (where Apply already saves a copy).

The primary button for a document opened as a dynamic template. It saves your changes back into the template where it lives, so every use of that template picks up the new design.

In an edit launched from a host (a page builder or the media library) this cancels back to the host without saving. It sits right next to the primary save action.

The primary button when a page builder or insert flow opened the editor. It saves a bound copy back to that element (so shared originals are never silently changed) and names the host, for example Apply to Elementor.

The main save button. It writes the image to the media library and, with versioning on, keeps the previous file as a restorable version. From the media library grid this is the primary action.

The primary button when you are editing the contents of a Smart Object in a nested editor. It bakes your edit back into the Smart Object in the parent document.

One open document. Click it to switch to that document. The tab shows the document name, and its tooltip adds the size. The active tab connects seamlessly into the panel below.

A small dot on a tab that marks a document with unsaved changes.

Closes the document. If the tab has unsaved changes the first click arms it and the tooltip changes to Click again to discard and close; a second click confirms. Moving focus away disarms it.

Shows which page you are on and how many pages the design has in total.

A small preview of each page. Click one to switch to it; the current page shows its number instead of a thumbnail.

Removes a page from the design. It asks to confirm first, and is only available when more than one page exists.

Adds a new empty page right after the current one, keeping the same canvas size, and switches to it.

Copies the current page (canvas and all its layers) into a new page right after it, and switches to the copy.

Hides the pages strip. The pages stay saved with the design, and you can show the strip again from the View menu.

The title of the bar, followed by a note that reads how many frames there are and reminds you that every visible top-level layer is one frame.

When ticked, faint versions of the neighbouring frames show through the current one, so you can line up motion between frames.

Plays the frames in order at their set delays, showing the result over the canvas. The button becomes Stop while playing. Disabled until there are two or more frames.

Opens the export dialog to save the animation as a GIF, APNG or WebM. Disabled until there are two or more frames.

Hides the timeline bar. Your layers and frame delays are unaffected.

A preview of one frame. Click it to select that frame's layer on the canvas; the active frame is highlighted.

The position of the frame in the sequence, starting at 1.

How long this frame stays on screen, in milliseconds, before the next one. Type a value or use the stepper.

Moves this frame one place earlier in the sequence. Disabled for the first frame.

Moves this frame one place later in the sequence. Disabled for the last frame.

A hint shown when there are fewer than two frames: add at least two visible top-level layers to animate.

A badge that shows you are in the simplified Easy Mode shell.

Steps one change backwards, the same undo as the full editor. Greyed out when there is nothing to undo.

Re-applies the last undone change. Greyed out when there is nothing to redo.

A one-line reminder: click an element to edit it, and everything important appears on the right.

Leaves Easy Mode and returns to the full pro editor with all tools and panels. Documents are fully compatible either way.

The default arrow. Use it to move and arrange the elements on your canvas.

Adds a text box in the middle of the canvas. Double-click it to type your words.

Tip. New text is centered and uses your foreground color.

Opens the shapes in the bottom tray. Tip: drop a photo onto a shape and the photo fills the shape.

Opens free stock photos in the bottom tray. Drag one onto the canvas, or onto a shape to fill it.

Opens ready-styled text combinations in the tray. One click inserts a beautifully styled headline set.

Opens emoji you can drop in as design elements.

Lets you paint freehand with the brush. You can then turn your sketch into a real picture with the AI tools.

Trims your design down to a new size.

A swatch that sets the canvas background color. Shown when nothing is selected.

A dropdown of common canvas sizes (the first six presets). Pick one to resize the canvas.

Opens the library at your saved designs and templates.

Picks the font family for the selected text from the curated template font list. Fonts load on demand.

Sets the font size of the selected text in pixels.

A swatch that sets the text color.

Toggles bold on the selected text (weight 700 when on, 400 when off).

Toggles italic on the selected text.

A swatch that sets the fill color of the selected shape.

A slider that rounds the corners of a rectangle shape. Only shown for plain rectangles.

Cuts the background out of the selected image using the local AI model, leaving the subject on transparency.

A note shown when an image is clipped into a shape: double-click the image to reposition it inside its shape.

A slider that sets how see-through the selected element is.

A style picker that adds a look to your description, for example a photo or illustration style, so the generated image matches the mood you want.

A text box where you describe the image you want, for example a cozy autumn coffee flatlay.

Generates a new image from your description using the first configured AI provider.

Takes a rough drawing made with the Draw tool plus your description and turns it into a finished picture.

Chips that launch installed extension generators (studios) right from Easy Mode. Only shown when extensions are installed.

A simple list of the elements on the canvas, front to back. Click one to select it, and drag to change what lies in front. Groups can be unfolded one level.

Locks an element so it cannot be moved or deleted, or unlocks it again.

Removes the selected element. Only shown for the active, unlocked element.

The real preview cards from the pro Effects panel appear here for the selected element: one-click filters, quick effects, and for text also styles, effects and warp. Click a rendered thumbnail to apply it.

When you click Back to Media Library, Back to Post or Cancel while the document still has unsaved changes, a prompt titled Unsaved changes asks Discard unsaved changes?. Click Discard to leave and lose the changes, or dismiss it to keep editing.

Tip. Shown only while the unsaved-changes dot is lit.

Closing a document tab that has unsaved changes needs two clicks on its close button. The first click arms the button (its tooltip changes to Click again to discard and close); the second click discards the changes and closes the tab. Clicking elsewhere disarms it.

Tip. Clean tabs close on a single click.

A small chevron on a group row in the Easy Mode Elements list. Click it to unfold the group and show its child elements indented one level below; click again to fold them away. Only whole groups reorder as a unit.

Tip. Easy Mode shows only one level of nesting to keep the list short.

A hint under the Easy Mode Elements list. Drag a top-level element up or down to change its stacking order; an element higher in the list sits in front of the ones below it.

Everything, nothing hidden. The default.

Fewer tools, bigger icons: the everyday set at a comfortable size.

Just the basics, for a screen with little room or a job with few steps.

Turns on the picker overlay. Click any panel, section or tool in the editor to hide it; click Done when finished.

The label the overlay shows on whatever is under the pointer, so you can see what a click would remove.

Saves the current arrangement under a name of your own.

Updates a saved workspace to match what is on screen right now.

Switches to a saved arrangement.

Removes a saved arrangement. The built-in presets cannot be deleted.

Large, Medium or Small, independent of which tools are visible.

Appears only when a view-only color proof is turned on (View, then Proof Colors). It names the proof so you remember the canvas is showing a simulated output, not your real colors.

Shows the canvas size and, when clicked, cycles the unit between pixels, centimetres and inches. Print sizes use the document DPI (72 by default).

Tip. Hover it to see the current DPI.

Reports the document color mode and bit depth, one of RGB · 8-bit, RGB · 16-bit or CMYK · 8-bit.

A rough estimate of how much memory the raw canvas uses, calculated from width times height times four bytes. A quick sense of how heavy the document is.

Names the tool you currently have active, so you can confirm what a click will do.

Zooms the view out one step (divides the current zoom by 1.25).

Shows the current zoom as a percentage and lets you type an exact value. Press Enter or click away to apply; an invalid entry snaps back to the current value.

Zooms the view in one step (multiplies the current zoom by 1.25).

Scales the view so the whole canvas fits inside the visible area.

The eye button next to the zoom control (also View, Reveal Off-Canvas) makes everything outside the document frame visible: layers parked on the pasteboard show up around the canvas instead of hiding behind the dark surround. A handy scratch space - park variants and material next to the frame, drag them in when needed. Export always crops to the frame, so nothing parked ever ships.

Highlights the tool rail: many tools with single-key shortcuts (V, B, T and so on), and the options bar above the canvas that adapts to the active tool.

Highlights the canvas: hold Space to pan, use Cmd/Ctrl with plus or minus to zoom, and note that everything lands on layers so your original is untouched until you save.

Highlights the right-hand panels: layers, properties, adjustments with filters and effects, history, AI actions and recordable Actions.

Highlights the Photoshop-style menus: Image, Layer, Select and Filter act on the document; View holds proofing, guides and the post preview.

Highlights the Assets menu: the Asset Library, the Insert menu, your Brand Kits (colors, fonts, logo, company profile) and design sharing.

Highlights the tray under the canvas: text lockups, shapes, icons, photos, brand-kit logos, dynamic and starter templates, all click or drag onto the canvas.

Highlights the Automation menu: Generate Content writes draft posts with images, Featured Images fills posts with branded thumbnails, and the Image Processor batch-edits the media library.

Highlights the primary save button: Save to Media Library keeps restorable versions, Save as New Image creates a fresh attachment, and editable work lives in Designs.

Shows your place in the tour, for example 3 / 8.

Ends the tour immediately. It will not show again automatically.

Advances to the next tour step.

The final button; closes the tour and marks it done so it does not reappear.

Tip. Restart the tour any time from Help, then Show Tour.

A tile showing how many installed extensions are enabled out of the total.

A tile showing how many optional local AI models are installed out of the total.

A tile showing how many media library images are known to the search index.

A tile counting the starter templates available.

A tile counting the ready-made text combinations.

A tile counting the shape and element library items.

A tile counting the filters plus the effects, including any added by extensions.

A tile counting the font families the editor knows about: the bundled ones, your custom uploads, and the whole Google catalogue when that option is on. It counts the catalogue, not what is downloaded and ready to draw with.

A tile counting your configured Brand Kits.

A tile showing how many stock image providers are configured out of the total.

A section that reports the image search index: whether the model is installed, live indexing progress, and how many images are indexed. An Index now button appears when images are still pending.

Lists each local model with an installed or not-installed dot: Smart Select, Depth Blur, Alt Text, Alt Text Translation, Layout and Image Search. A final row names the inference device and whether it is GPU-accelerated.

Shows the running plugin version (and warns if a newer version is installed on the server or if REST is unreachable), which AI providers are configured, and a count of any extension issues.

An admin-only stacked bar over the whole disk with every number listed below it: Versions & project files, Local AI models, Extensions, Fonts & editor runtime, Media library (other files), Other server data, and Free disk space. Color only aids scanning.

A list of events recorded during this editor session, colored by level (errors and warnings stand out). Shows the time, source and message of each entry.

Downloads one JSON file to attach to a support request: status counts, device and graphics facts (browser, OS, display, WebGL hardware/software) and the debug log - never document content, images or API keys.

Empties the session debug log.

The client machine at a glance: browser, operating system, display size and pixel ratio, CPU cores and memory, plus the graphics stack - WebGL version, hardware vs software rendering and WebGPU availability. Software rendering explains slow or frozen 3D studios in one look.

Chooses whether a new guide is vertical or horizontal.

The distance of the new guide from the top-left, entered as a number. Press Enter to add it.

Whether the position is read as pixels or a percentage of the canvas.

Adds the guide at the entered orientation and position, clamped within the canvas.

Adds guides at the rule-of-thirds positions for balanced composition.

Adds guides at the golden-ratio positions.

Adds a margin frame 5% in from each edge.

Removes every guide. Only shown when guides exist.

Lists each existing guide with its orientation and pixel position. Edit the number to move a guide, or use the close icon to delete it.

Turns the on-canvas grid overlay on or off.

Sets the grid cell size in pixels.

A search box that filters the article list as you type. It is focused automatically when the Handbook opens.

Five article groups in the sidebar: The basics, Design, AI, Automation and Library & system. Every article belongs to exactly one.

The list of articles, grouped by category while the search is empty; typing filters to a flat result list. No matches. shows when the search finds nothing.

Interactive walkthroughs that run the spotlight tour over the real editor to teach a task. Click one to start it.

The selected article on the right: intro paragraphs, an optional numbered Step by step walkthrough, Good to know tip callouts and buttons that open the dialogs or guides the article describes.

A small question-mark button in the header of every dialog that opens the Handbook to the article for that dialog. The Handbook always stacks above the dialog it was opened from.

Pill buttons at the top of an article that close the Handbook and open the described dialog directly - for example Open Design Generator in the Design Generator article.

A numbered walkthrough inside an article: badge-numbered steps connected by a line, describing the task in order.

Highlighted tip callouts at the end of an article for non-obvious extras, like saving a generated design as a dynamic template.

Uploads and installs an extension package from a ZIP file. Only shown to users allowed to manage extensions.

Before the upload, a dialog explains that an extension runs with your permissions inside the editor (it can read and change your documents and act on your behalf, like a WordPress plugin) and asks you to confirm you trust the source.

Turns an extension on or off. Enabling can take effect live; disabling an already loaded extension takes effect after the next reload.

Deletes the extension and its files after a confirm prompt.

A small tag on an extension row that is currently turned off.

A tag marking an extension whose on-disk state differs from the running session, so a reload is needed to apply it.

Chips listing what an extension adds, such as Effects, Filter presets, Menu items, Panels, Panel sections, Tools, Export formats, Generators, Library sections, Template packs or AI tools.

A link to the extension's homepage, shown when the package declares one.

Any load or runtime issues an extension has reported appear in red on its row.

The dialog subtitle names the extension API version the editor supports, for compatibility.

The 3D studios (Text, Mockup, Effects, Animation, Gallery, Objects) and Living Photos render through WebGL2 and need hardware graphics. Every current desktop browser provides that; virtual machines, remote desktops and browsers older than Safari 16.4 often do not - the studio then explains itself instead of opening, and System Status → Device & graphics shows the exact graphics stack.

Appears in the Semantic image search row when some media library images are not yet indexed. Click it to embed the pending images so they become searchable by description. Until the search model is installed, the row instead tells you to download it under Settings, then Local AI Models.

Tip. Progress shows as Indexing… X / Y while it runs.

A row under Local AI models showing which engine runs the local models: WebGPU (GPU) when your browser supports GPU acceleration, otherwise WASM (CPU). A device is always available, so local features work either way, just faster on WebGPU.

A row under System showing the version of the editor currently loaded in your browser. If a newer version is installed on the server it shows X installed, reload the editor; if the server cannot be reached it shows REST unreachable!.

A row under System listing which cloud AI providers currently have a key configured (for example gemini, openai), or none when no keys are set.

A segment of the admin-only Storage bar. It measures disk used by saved image versions and the editable project sidecar files the editor keeps. The exact size is listed beside the colored swatch.

A Storage bar segment measuring the disk used by downloaded local AI model files (stored under wp-content/uploads/wpie-models).

A Storage bar segment measuring the disk used by installed editor extension packages.

A Storage bar segment measuring the disk used by uploaded custom fonts and the editor's own runtime files.

A Storage bar segment (neutral gray) measuring the rest of your WordPress media library, the files that are not editor buckets.

A Storage bar segment (neutral gray) measuring everything else on the disk outside the uploads folder. Shown only when total and free disk figures are available.

The empty part of the Storage bar, listed as free of total (for example 40 GB of 100 GB). Shown only when the server reports disk totals.

The model behind the Smart Select tool. It finds and cuts out the object you click on, so you can select a subject without tracing it by hand.

A depth-estimation model. It builds a depth map of a photo so features like background blur and depth-aware effects can tell foreground from background.

A captioning model that reads the rendered image and writes an alt-text description locally, no cloud key needed. Powers the Generate alt text action.

A translation model that renders the locally generated alt text into another language, so captions match your site's language.

A small text-embedding model used for layout and salience decisions, helping the editor place and rank elements sensibly.

The embedding model behind semantic image search. It turns your media library images into vectors so you can find them by describing what they show.

An Orientation option for a new guide. A vertical guide is a top-to-bottom line placed at an X position, useful for aligning columns and margins.

An Orientation option for a new guide. A horizontal guide is a left-to-right line placed at a Y position, useful for aligning rows and baselines.

A Unit option in the New guide row. The number you type is taken as an exact pixel position on the canvas.

A Unit option in the New guide row. The number is read as a percentage of the canvas width (vertical guide) or height (horizontal guide) and converted to a pixel position, so 50% centers the guide.

The version badge next to each extension's name (for example v1.2.0). It shows which release of that package is installed, handy when checking whether a reload is loading a newer build.

A byline in an extension row's meta line showing who made the package, for example By TBIT Design. Only shown when the manifest names an author.

Small chips in an extension row that name what the package adds to the editor, drawn from a fixed vocabulary. They tell you at a glance whether a package brings tools, effects, generators and so on.

The confirmation that appears when you click Remove on an extension. It asks Remove "name" and delete its files?; confirming deletes the package from disk. This is a destructive action.

Tip. Removal deletes the files; already-loaded code stays until you reload.

The empty-state message shown in the Extensions dialog when nothing is installed. Use Install from ZIP… to add your first package.

The input. Ask in your own words; the assistant looks up the matching sections and answers from them.

Sends the question.

Turns the answer into a guided tour: the editor spotlights each control in turn instead of naming it.

Starts over with an empty thread. The previous exchange is not kept.

The starting prompts on an empty thread, covering the things people ask first: adding and styling text, removing a background, dynamic templates, shape layers, actions and batch processing.

Zooms the view in one step.

Zooms the view out one step.

Scales the view so the whole canvas fits the visible area.

Sets the zoom to actual size, one screen pixel per image pixel.

Selects the whole canvas.

Clears the current selection.

Swaps the selected and unselected areas.

Opens the Asset Library.

Opens the command palette to search and run any command.

Opens the AI panel on the right. The plain letter A is free because select all uses Cmd/Ctrl + A.

Opens the in-editor Handbook.

Cancels the current draft, or cancels a crop, or clears the selection, in that order.

Commits the current draft, such as confirming a crop or a shape.

Holding Space temporarily switches to the Hand tool so you can drag the canvas; releasing returns to your previous tool.

Steps one change backwards.

Re-applies the last undone change (Cmd/Ctrl + Y also works).

Saves. A document bound to a design saves the design, one bound to a template updates the template, otherwise the export dialog opens to save to the library.

Duplicates the selected layer.

Copies the selected layer to the clipboard.

Cuts the selected layer to the clipboard.

Pastes the layer from the clipboard.

Moves the selected layer one step towards the front.

Moves the selected layer one step towards the back.

Deletes the selected layer, or clears the pixels inside an active selection.

Moves the selected layers by one pixel with the arrow keys, or ten pixels with Shift held. Groups move with all their children; a locked member freezes its whole unit.

Sets the active layer's opacity from the number keys: 1 to 9 give 10% to 90%, and 0 gives 100%.

Shrinks the current paint tool's brush size (by 5 above size 10, otherwise by 1). Works with Brush, Pencil and Eraser.

Grows the current paint tool's brush size (by 5 from size 10 up, otherwise by 1). Works with Brush, Pencil and Eraser.

Opens the quick-insert palette to drop in an element fast.

Swaps the foreground and background colors.

Selects the Move tool for moving and arranging layers.

Selects the Select (object) tool.

Selects the rectangular marquee selection tool.

Selects the freehand lasso selection tool.

Selects the Magic Wand, which selects areas of similar color.

Selects the Crop tool.

Selects the Brush for soft painting.

Selects the Pencil for hard-edged painting.

Selects the Eraser.

Selects the Paint Bucket to fill areas with a color.

Selects the Gradient tool.

Selects the Clone Stamp to paint copied pixels from another spot.

Selects the Blur/Sharpen brush.

Selects the Text tool.

Selects the Shape tool.

Selects the Pen tool for drawing paths.

Selects the Eyedropper to sample a color.

Selects the Hand tool to pan the canvas.

Selects the Zoom tool.

Selects the AI Smart Select tool that finds objects for you.

Editor defaults: the default canvas size and the starting theme.

API keys, the default image and text providers, and per-model overrides and prices.

A monthly spend limit plus usage statistics and charts.

Download optional models that run locally in the browser.

Built-in fonts, optional Google Fonts and custom font uploads.

The Integrations tab: keys for the stock providers Pexels, Pixabay and Unsplash, and the Meshy key that the 3D Objects studio uses to build models from a prompt.

Whether and how many previous file versions to keep.

The minimum capability required to use the editor.

The uninstall data-removal option.

Export, restore and scheduled remote backups.

The size a brand-new document starts at. Pick a common preset, or a custom size is shown if one is stored.

Seeds the editor's initial appearance. Users can still toggle dark or light inside the editor.

The key for Google Gemini, used for text-to-image and prompt-based image editing. Get one at aistudio.google.com.

The key for OpenAI, used for generation, true mask inpainting (gpt-image-1) and variations (dall-e-2). Get one at platform.openai.com.

The key for Anthropic Claude. Not for image generation; it powers Generate alt text, captions and descriptions from the rendered image. Falls back to the first configured text provider.

The eye button next to a key field that reveals or masks the stored key. Keys defined in wp-config.php show as read-only.

Sends a test request to the provider and reports OK or the error inline, so you can confirm a key works.

A status badge on each provider card showing whether a working key is stored.

Which provider handles image generation and editing unless a dialog picks another model.

Which provider powers the Design Generator, generated library items and illustrations, text layouts, alt text and captions, and SEO texts. Falls back to the first configured text-capable provider.

Text fields to override which model each provider uses for each task. Empty fields use the built-in defaults shown as placeholders. Slots include Image model, Fast text model, Design model and (OpenAI) Variations model.

Per-model price fields that feed the Costs & Budget statistics. Image models take a price per image; text models take prices per 1M input and per 1M output tokens. Keep them in sync with your providers' current pricing.

A dollar cap for AI usage each month. AI calls are blocked once the month reaches this amount.

A progress bar under the limit showing how much of the monthly budget is used. Only shown when a limit above zero is set.

A stat card with the total estimated AI cost so far this month.

A stat card with today's estimated AI cost.

A stat card counting AI requests made this month.

A stat card with the remaining monthly budget, or an infinity symbol when no limit is set.

A stacked bar chart of daily AI cost over the last 30 days, colored by provider (Gemini, OpenAI, Anthropic).

A table breaking down this month's usage per model: task, requests, images, tokens in and out, and cost. Shows a note when there is no usage yet.

A table of available models with columns for Model (and its source repo), Size, Status and an action button. It loads the list from the server when the tab opens.

Shows whether each model is already downloaded to your server.

Downloads the model to your server, or re-downloads it if already installed. The button shows Downloading while it works and reports how many files were fetched or skipped.

The editor ships ten font families in the plugin itself, served from your own site. The Font library card downloads the rest of the catalogue to your server on request, after which those families are self-hosted too. Either way no visitor data leaves your server, and this source is always active.

When enabled, adds Google Fonts to the font picker. Privacy note: this loads fonts from fonts.googleapis.com while people work, which sends their IP to Google. Exported images are unaffected because text is rendered to pixels. Leave it off for a strictly self-hosted setup.

Upload your house fonts to serve from your site and offer in the font picker like any other family. A family name is optional and taken from the filename if left blank.

The family field, file chooser and Upload button used to add a custom font. A status line reports progress or errors.

Removes an uploaded custom font from the list and your server.

Enables searching Pexels free stock photos. Get a key at pexels.com/api.

Enables searching Pixabay free stock photos and illustrations. Get a key at pixabay.com/api/docs.

Enables searching Unsplash high-quality free photos. Get an Access Key at unsplash.com/developers.

Runs a tiny live search with the stored key and reports success or the provider's error.

Clears the stored key when you save; the masked field alone cannot distinguish removal from leaving it untouched.

When on, saving over an image keeps the previous file as a restorable version (recoverable from History).

How many previous versions to keep per image before the oldest are pruned.

Sets who may use the editor by required capability. Opening a specific attachment also requires permission to edit that attachment, and configuring AI keys always requires manage_options.

When ticked, deleting the plugin also removes all its data: settings, saved versions and project sidecars. Leave it off to keep your data if you reinstall later.

When ticked, the export bundles your installed editor extensions.

When ticked, the export includes your API keys in plain text. Keep the file safe.

Tip. Remote backups never contain API keys.

When ticked, the export includes stored image versions. This can be large and is mainly useful for same-site restores.

Builds and downloads the backup ZIP with the chosen options.

Uploads a backup ZIP to restore. Restoring replaces your dynamic templates, designs, library assets, custom fonts and any extensions in the archive; settings are merged and kit logos and watermarks are re-imported. Stored API keys stay untouched unless the backup contains keys.

A required checkbox acknowledging that restoring replaces existing templates, designs and library assets, before the Restore button will submit.

Scheduled off-site backups to any S3-compatible storage (AWS S3, Hetzner, DigitalOcean Spaces, Backblaze B2, Wasabi, MinIO). Needs the PHP extensions curl and SimpleXML on your server, which most hosts have. Remote archives never contain API keys. This is part of WunderPaint Pro; the local export and restore above stay free.

A provider card. Google Gemini does text-to-image and prompt-based image editing (model gemini-2.5-flash-image). Get a key at aistudio.google.com. The card shows a Connected or Not configured badge.

A provider card. OpenAI handles generation and true mask inpainting (gpt-image-1) and image variations (dall-e-2). Get a key at platform.openai.com.

A provider card. Anthropic Claude is not an image generator; it powers Generate alt text, captions and descriptions written from the rendered image.

A note on the AI Providers tab: keys are stored obfuscated (reversible, not true encryption) in the database. Background removal and 2x/4x upscaling always work locally with no key. For real secrecy, define keys as constants in wp-config.php.

When a provider key is set as a PHP constant in wp-config.php, its field on the settings page becomes a disabled, masked box labelled Defined in wp-config.php. The constant takes priority and cannot be edited from the admin screen.

The model id used for image generation and editing (Gemini and OpenAI). Type a different id to override the default, or leave it blank to use the placeholder default.

The quick text model each provider uses for alt text, captions and SEO text. Available for Gemini, OpenAI and Anthropic.

The stronger text model that powers the Design Generator, generated library items, illustrations and text layouts. Available for Gemini, OpenAI and Anthropic.

OpenAI only: the model used to make variations of an existing image (default dall-e-2).

The price input for an image model, in dollars per generated image. Used to estimate costs in the Costs & Budget tab. Empty uses the seeded default shown as the placeholder.

The price input for a text model's input, in dollars per one million input tokens. Feeds the cost estimates.

The price input for a text model's output, in dollars per one million output tokens. Feeds the cost estimates.

The table column showing the model id used (with its provider named underneath).

The table column showing what the calls were for (the action, such as generate or caption).

The table column counting how many API calls were made to that model this month.

The table column counting how many images that model produced this month (a dash when none).

The table column totaling input tokens sent to that model this month (a dash when none).

The table column totaling output tokens received from that model this month (a dash when none).

The table column showing the estimated dollar cost for that model this month.

A column in the model download table giving the approximate download size for each model (for example ~150 MB), so you can judge the download before starting it.

Under each model's name the table shows its source repository (the Hugging Face repo id). Downloads run server-to-server: the runtime comes from jsDelivr and the models from Hugging Face.

A note on the tab: downloading stores each model once on your own server (wp-content/uploads/wpie-models). After that it is served from your own domain, so nothing is fetched from a third-party CDN at run time and the features work offline.

A note on the tab: the server-to-server downloads can take a while. If one times out, click Download again and it resumes where it left off rather than starting over.

The checkbox under Google Fonts (CDN). When on, the font picker offers extra Google Fonts loaded from fonts.googleapis.com while people work, which sends their IP to Google. Exported images are unaffected because text is rendered to pixels. Leave it off for a strictly self-hosted setup.

The custom-fonts uploader accepts woff2, woff, ttf or otf files. Uploaded fonts are served from your own site and appear in the font picker like any built-in family.

The optional text field next to the custom-font file picker. Type a family name to control how the font appears in the picker; leave it blank and the name is taken from the filename.

The standard WordPress submit button at the foot of the settings form. It writes every tab's fields at once. It is hidden on the Backup tab, whose export and restore forms submit on their own.

The Default canvas size dropdown on the General tab offers 1080 x 1080, 1920 x 1080 and 1200 x 630. If a custom size was stored, it appears as an extra option marked (custom). This sets the starting size for new documents.

The Theme dropdown on the General tab seeds the editor's initial look: Dark, Light or System (follow the operating system). Users can still toggle the theme inside the editor afterwards.

Besides Shift+Cmd+Z, plain Cmd+Y (or Ctrl+Y) also redoes the last undone step, matching the common Windows convention.

Zoom in accepts either the + or the = key while holding Cmd or Ctrl, so you do not need to press Shift to reach + on most keyboards.

Delete or Backspace normally removes the active layer. But when an area selection is active, the same keys instead clear the pixels inside that selection and leave the layer in place.

Arrow keys move the selected layer or group by 1 pixel. Hold Shift while pressing an arrow to move by 10 pixels instead. A locked member freezes its whole unit.

Shortcuts are remappable. Each command has a default combination; assigning a new one stores an override, and the remapped command stops firing on its old key. The editor detects when a combination is already taken by another command so you can resolve the clash.

In Easy Mode only everyday keys work: Delete, Backspace, Escape, Enter, the arrow keys, and a small set of Cmd/Ctrl combinations. Single-key tool switches and pro-only shortcuts stay quiet so beginners cannot stumble into hidden tools.

Stored obfuscated like every provider key; the eye toggles visibility, Remove clears it. Sign up at meshy.ai - trial credits are free, then pay-as-you-go.

Verifies the stored key against Meshy and shows your remaining credit balance.

Which Meshy generation model to use: Latest (recommended) always tracks the newest, or pin Meshy 7 / Meshy 6 / Meshy 5. Meshy 7 exists for image-to-3D only, so text-to-3D quietly falls back to the newest model that endpoint offers.

One friendly quality level for every generation: Standard (recommended), High (4K textures on Meshy 6 and newer, larger files), Ultra (High plus the extra refinement pass Meshy 7 runs on images - sharper geometry, slower, 5 credits more) or Low (a compact 10,000-polygon remesh, fast and small). From Meshy 6 on the mesh is left as generated except at Low, because remeshing grinds down the very geometry the newer models are better at. PBR material maps are always included.