Documentación de EvoX
  • Producto
  • Precios
  • Documentación
  • Mercado
Volver a EvoMap
Iniciar sesiónRegistrarse
Overview
Practice cases
Features
Workflows
Projects and chatsLocal workspaceLong-term memoryScheduled tasksLong-running workNotificationsSitesVisualizationsPets
Capabilities
File handlingImage inputAttach an appWeb searchBrowserUse everyday ChromeVoiceImage generationPluginsSelf-evolution
Reusable experience
GenesCapsulesEvolver skillsEvolution events.gepx evolution archives
Swarm
Using swarmContinuous swarmRecursive tasksRecovery and follow-upDevice collaboration
Reference
CommandsSlash commandsSettingsTroubleshooting
Configuration
Developers
Security
Administration
Overview
Practice cases
Features
Workflows
Projects and chatsLocal workspaceLong-term memoryScheduled tasksLong-running workNotificationsSitesVisualizationsPets
Capabilities
File handlingImage inputAttach an appWeb searchBrowserUse everyday ChromeVoiceImage generationPluginsSelf-evolution
Reusable experience
GenesCapsulesEvolver skillsEvolution events.gepx evolution archives
Swarm
Using swarmContinuous swarmRecursive tasksRecovery and follow-upDevice collaboration
Reference
CommandsSlash commandsSettingsTroubleshooting
Configuration
Developers
Security
Administration
Documentación de EvoX/Product/Image generation

Image generation

Create images from a description or use reference images to edit existing assets in EvoX.

Use image generation for article illustrations, presentation covers, banners, backgrounds, and artwork. Describe the purpose, subject, and style. To continue an existing design, attach the original or reference images and explain what each contributes.

Generate or edit an image

EvoX uses the image-generation plugin to call an image service. The plugin starts disabled. Choose a provider in its dedicated configuration and use credentials managed under Settings > AI services. Missing credentials, a failed test, or cancellation leaves it disabled.

Once enabled, describe the image in your task. For edits, provide the original and define the change. Review the provider, input images, possible cost, and output location before approving. The plugin does not silently switch providers after a failure.

Review the result

Images are saved to the task artifact directory, with the provider, model, dimensions, file type, size, and local path returned. Review whether the image fits its purpose before requesting changes.

Check subject details, text, edges, and transparency. Generated images are not evidence of real events, people, or actual product behavior.

Write an effective prompt

One to three clear sentences are often enough. Focus on the details that determine success:

  • Purpose: Where the image will appear and who will see it.
  • Subject: What to depict and what is happening.
  • Composition and style: Placement, setting, lighting, colors, and materials.
  • Size and space: Landscape or portrait, and space needed for a title.
  • Constraints: What must stay and what must not appear.

Instead of asking for a “more premium” look, specify muted colors, a simple background, and soft side lighting.

Create a landscape cover illustration for a team knowledge base. Show a person organizing project notes at a desk with a laptop and a few sticky notes. Use a clean editorial illustration style, muted colors, and soft daylight. Place the subject on the right and leave room for a title on the left. Do not add text, logos, or watermarks.

Refine an image

Change one aspect at a time to avoid unintentionally changing the subject, composition, and style together. State both what should change and what should remain unchanged.

Edits can use an original image, mask, or references, depending on the selected provider. Inputs must be ordinary local files in allowed locations and pass file-type and size checks.

Edit this cover image: replace only the sticky notes on the desk with an open notebook. Preserve the person, laptop, camera angle, lighting, colors, and empty space on the left. Do not add any other objects or text.

Use multiple reference images

Identify each reference in order and explain its role: subject, style, or layout. Avoid simply asking to “use these references.” Reference support depends on the configured provider.

Image 1 defines the product appearance to preserve. Image 2 is only a color and background reference. Keep the product shape, proportions, and camera angle from image 1, with the pale gray background and soft shadows from image 2. Center the product and leave the upper-right corner empty. Do not include other objects from image 2.

Add text to an image

Keep text short and quote the exact wording. Specify placement, color, and type style. Check every word afterward, especially names, numbers, and non-Latin characters.

Add only the title “TEAM OPEN DAY” at the top center of this event poster, in clear white bold sans-serif lettering on one line. Preserve the original subject and composition. Do not add a date, subtitle, or any other text.

Create infographics and complex layouts

Image generation can explore visual directions for posters, flowcharts, and timelines. Define reading order and information hierarchy, and keep each area's text short.

For dense copy, exact values, or layouts that need further editing, use the generated image as artwork and finish the layout in a design tool, presentation, or webpage.

Before you use an image

  • Confirm portrait rights for real people and appropriate permission for reference images and brand assets.
  • Describe original design directions rather than presenting existing work or brand treatments as your own.
  • Review image uploads and costs before remote calls; avoid sending sensitive material as a reference.
  • Outputs go to the task artifact directory. The plugin does not accept arbitrary output directories.

Related docs

  • Skills and plugins
  • Image input
  • File handling
AnteriorVoiceSiguientePlugins