--- title: AI Image Editing slug: editorial/ai-image-editing-in-purple-dam docTags: createdAt: 2026-06-10T00:00:00.000Z --- :::BlockQuote **ℹī¸ Note:** AI Image Editing requires [Purple DAM Plus](#). [Purple Prompts](https://docs.purplepublish.com/editorial/purple-prompts-faq#8-how-does-setting-up-purple-prompts-work) must also be activated on your hub. ::: ## Overview AI Image Editing adds AI-powered image creation and editing to Purple DAM. Editors can generate new images from text prompts, create variants or combinations of existing assets, and apply AI-based edits without leaving the Media Library. AI image generation and editing always uses **OpenAI** — there is no Anthropic counterpart for it. A Hub that runs its prompts on Anthropic therefore still needs the OpenAI API key for these features, see [AI Models and API Keys](docId\:zTwrSgVsiEYd2-Lu5-fNo). ## Generate Click **Generate** in the Media Library toolbar to open the generation dialog. ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/-LUFSE0QRv1N5cCE0-cI0_screenshot-generate-button.png) 1. Enter a text prompt describing the image you want to create. 2. Optionally adjust the **Model**, **Format**, and **Ratio**. The defaults are set in [Settings](docId:). 3. Click **Generate**. ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/uKjE7vieJuCIBdwEWToje_screenshot-generate-dialog.png) A progress indicator is shown while the image is being created. When done, the new image is saved directly to the Media Library and the dialog closes. ## Create Variant A variant is a new image generated using one existing image as a visual reference. 1. Select exactly **one image** in the Media Library. 2. Click **Variant** in the bulk-action bar. ::Image[]{src="https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/imT2ZJh1dqxayWM6aQkHS_screenshot-variant-button.png" size="54" width="790" height="752" position="center" darkWidth="790" darkHeight="752" showCaption="false"} 3. Enter a prompt describing the variant you want, then click **Create Variant**. The generated variant is saved as a new asset in the Media Library. ## Combine Combining merges two or more existing images into a new image, guided by a prompt. 1. Select **two or more images** in the Media Library. 2. Click **Combine** in the bulk-action bar. ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/mrN9aLOjTAOUcUh3K_ZHL_screenshot-combine-button.png) 3. Enter a prompt and click **Combine**. The resulting image is saved as a new asset in the Media Library. ## Edit AI editing transforms an existing image based on a text instruction and saves the result back to that asset or as a copy. 1. Click any image in the Media Library to open its **File Detail** view. 2. Click **Edit** in the header. ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/pMy_IOtnpy_y99jmERu0Z_screenshot-edit-button.png) 3. Enter an instruction, for example: *Remove the background* or *Change the sky to a sunset*. 4. Optionally adjust the **Model**, **Format**, and **Ratio**. ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/UhbFn7xOdr4oRFzpFJKyg_screenshot-edit-prompt.png) 1. Click **Apply**. Once the AI has processed the instruction, a preview of the edited image is shown. Choose how to proceed: ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/7V9ioIR57aRRHoNuNgaAi_screenshot-edit-dialog.png) - **Save Edits**: overwrites the existing asset with the edited version. - **Save As Copy**: saves the result as a new asset, leaving the original unchanged. - **Reset**: discards the preview so you can try a different instruction. - **Cancel Editing**: closes the dialog without saving anything. ### Undo, Redo, and Restore Original After saving edits with **Save Edits**, additional controls appear in the File Detail header: ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/fI4Ze6Icj6QAbVeMegUak_screenshot-undo-restore.png) - **Undo**: reverts to the version before the last save. - **Redo**: reapplies a previously undone edit. - **Restore Original**: reverts all AI edits and restores the image to its pre-edit state. :::hint{type="info"} Edit history is stored for **1 hour** and supports up to 50 steps. History is cleared when the page is reloaded. ::: ## AI Image Indicators Images that were created or edited by AI are marked throughout the Media Library. In the grid view, a spark icon appears on the tile. Hovering it shows a tooltip with the full label, for example: *Created with gpt-image-1* or *Edited with gpt-image-1*. ::Image[]{src="https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/fRK2PYW6f_APeeiY--QvI_image.png" size="60" isUploading="false" width="728" height="524" darkWidth="728" darkHeight="524" showCaption="false"} In the File Detail view, an **AI source** field (also marked with a spark icon) shows the same information. ## Settings Administrators can configure AI Image Editing defaults under **Media → AI Image Editing** in the Hub sidebar. ![](https://api.archbee.com/api/optimize/ygR6KtT9QI_R0u3uKTJsm/ldvY4cKhauVnjzuQZJ0Zx_screenshot-ai-image-settings.png) | Setting | Options | | ------------------------ | --------------------------------------------------------- | | **Model** | gpt-image-1, gpt-image-1-mini, gpt-image-1.5, gpt-image-2 | | **Default Format** | png, jpeg, webp | | **Default Aspect Ratio** | 1:1, 3:2, 2:3 | The selected model, format, and aspect ratio are pre-filled as defaults in all AI dialogs. Editors can override them per operation. ## Features Each feature can be enabled or disabled individually using toggle switches. Disabling a feature hides its button from the interface: **Generate** removes the button from the Media Library toolbar, **Variant** and **Combine** remove their buttons from the bulk-action bar, and **Edit** removes the button from the File Detail view.