Skip to main content
Text to Image

Recraft V3 (Red Panda)

Generate high-quality raster and vector images using Recraft V3 (Red Panda) with style control, negative prompts, and optional best-effort clean-copy-space guidance

View details
Try it in Ampere

Inputs

Loading input fields...
Execution Steps

Loading workflow structure...

Loading curated examples...

Overview

Recraft V3 (Red Panda) turns a written prompt into one to six generated images with style and substyle controls. Use it for campaign visuals, ad concepts, social images, logo or emblem directions, vector-style graphics, and text-heavy visuals when you need style-controlled text-to-image generation with negative prompts and fixed size presets.

Use cases

  • Generate campaign or social visuals in a specific visual style.
  • Explore logo, emblem, vector, realistic, or digital illustration directions from one brief.
  • Create image options where rendered words or labels are part of the concept.
  • Use negative prompts to reduce unwanted objects, moods, or visual details.

Input tips

  • Keep the prompt under 1,000 characters and include the subject, style goal, composition, and intended use.
  • Choose a style_option such as realistic image, digital illustration, vector illustration, or logo raster.
  • Use a style::substyle option when you want more specific visual control.
  • Set size to one of the supported fixed dimensions, such as 1024x1024 or 1820x1024.
  • Set num_images from 1-6 when you want several directions.
  • Add a negative_prompt for elements, styles, or details to avoid.
  • Set avoid_text to true when you need a best-effort clean background or copy space; add approved copy in a separate design pass when exact typography matters.

Expected output

The AI Tool returns generated image URLs, the style used, optional substyle used, and cost metadata. The Recraft output view renders a single image or gallery and shows the combined style, image count, and Recraft V3 (Red Panda) model label. Image records are URL-only and do not include dimension or file metadata.

Caveats

  • Prompt wording and style selection strongly affect the result.
  • Style and substyle choices can change the output's format and visual language.
  • Negative prompts reduce unwanted elements but do not guarantee they disappear.
  • avoid_text is a best-effort prompt mode; Recraft can still render incidental marks or small text, so review copy space before use.
  • Vector, logo, and rendered-text results should be reviewed for brand fit and readability.
  • Image records are URL-only; dimensions and file metadata are not returned.

Related AI Tools