RunfraInstall

Summary

Generate one or more candidate images from a single text prompt using specified AI models, and return the top pick and all accepted image URLs.

Inputs

• Prompt (required) — Text description (1–2000 characters) of the scene to generate as an image.
• Batch Size — Number of candidate images to generate (1–100). Larger batch sizes improve the selection of the top pick. Tier caps apply (anonymous=4, free=8, paid=100).
• Model — Choice of AI model to use for image generation (SDXL, FLUX Schnell, or RunFra Pro). Model selection affects credit cost.
• Width — Width in pixels for the output image (512–1024).
• Height — Height in pixels for the output image (512–1024).
• Additional Options — Optional settings including Negative Prompt (things to avoid; ignored for FLUX Schnell and RunFra Pro), Quality Mode (aggressiveness of quality filter: strict, soft, off), Notify On Complete (send email on job completion if user has verified email), and Polling Timeout in seconds (max time to wait for job completion before returning a running status).

Output shape

a single JSON object per input item containing the job ID, status, input mode, seed values, counts and scores related to quality, plus URLs of the top pick image and all accepted images; in case of failure, includes an error message; if timed out, returns job ID with status='running' indicating polling should continue externally.

The node polls the Runfra API for the job result up to the specified timeout. If the timeout is exceeded, it returns the job ID and status='running', allowing workflows to continue polling in subsequent steps. The output includes both a bestResultUrl (top pick) and an array of resultUrls for all accepted candidates. Errors during generation result in an errorMessage field. Negative Prompts are only effective for the SDXL model.

Links

• n8n community nodes installation guide