Recraft V3 Text to Image

Recraft V3 generates images from a text prompt. Write what you want, choose a size and style, and hit run.

The workflow also accepts an optional reference image. Feed one in and Recraft V3 uses it as a starting point, blending your prompt with the visual structure of the input.

Write your prompt, pick your size, choose a style. That's the setup.

How do you use Recraft V3 for text to image generation?

Write a prompt, choose an image size from the presets or set custom width and height, and pick a style. The style setting shapes the visual treatment. Recraft V3 handles the rest. Add a reference image if you want image-to-image output instead of pure text-to-image.

Prompt Describe what you want in the image. Recraft V3 responds well to specific, visual descriptions. Include subject, setting, lighting, mood, and composition details. Cover all of these: subject, clothing or surface details, setting, lighting conditions, composition, and treatment. That level of detail gives you consistent, intentional output.

Image size Choose a preset from the image_size dropdown (landscape_16_9, square_1_1, portrait_4_5, and others) or set a custom resolution using the width and height fields. Use presets for standard formats. Use custom dimensions when your output needs to hit a specific pixel size.

Style This is the biggest lever in the workflow. Options include realistic_image, digital_illustration, vector_illustration, icon, and others.

Want photorealistic output for product shots or editorial imagery? Use realistic_image. Want a clean illustrated look for brand or marketing content? Try digital_illustration or vector_illustration. Want minimal iconographic output? Use icon.

Style ID Optional. Leave this blank unless you have a custom Recraft style ID to apply. Custom styles let you lock in a specific visual treatment across multiple generations.

What is Recraft V3 good for?

Marketing visuals, editorial imagery, product photography, illustration, and icon creation. Recraft V3 handles photorealistic and illustrated styles, making it useful across a wide range of commercial and creative briefs.

Recraft V3 works well when you need consistent, style-controlled output at a specific format. The preset sizes cover common social and editorial aspect ratios. Custom dimensions handle anything else.

The image-to-image path is useful for iterating on existing visuals: feed in a reference, write a prompt describing what to change, and generate variations while keeping the structural foundation of the original.

If you need fine-grained control over poses, exact compositions, or detailed retouching, a workflow with ControlNet or inpainting nodes will give you more precision. Recraft V3 is for prompt-driven generation where style consistency and output quality matter more than pixel-level control.