OpenParable
Skip to content
← Back to skills

Pollinations Image Gen

Design
pollinations-image-gen

Pollinations Image Gen is an Agent Skill for design published in the xream/scripts repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Generate images using the Pollinations.ai API. Supports models such as flux, zimage, imagen-4, klein, klein-large, gptimage (free), and seedream, kontext, nanobanana, gptimage-large (paid). Use when the user asks to generate, create, or draw images from text prompts.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/xream/scripts/blob/HEAD/skills/pollinations-image-gen/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/pollinations-image-gen, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Pollinations Image Generation

Generate images from text descriptions using the Pollinations.ai API.

When to use

  • User asks to generate, create, draw, or render an image
  • User mentions Pollinations, or any of the supported model names

How to use

Run the generation script at {baseDir}/scripts/generate.mjs with node. Pass all parameters as --key value CLI flags.

Required parameters

ParameterDescription
--promptText description of the image to generate. You should craft a detailed, high-quality prompt based on the user's request.
--modelAI model to use (see model list below)

Optional parameters

ParameterDescriptionDefault
--outputOutput file path. Parent directories will be created automatically. If omitted, saves to current directory with a timestamped filenameauto
--widthImage width in pixels1024
--heightImage height in pixels1024
--seedRandom seed for reproducible results. Use -1 for random0
--enhanceSet to true to let AI improve the prompt for better resultsfalse
--negative_promptThings to avoid in the image"worst quality, blurry"
--safeSet to true to enable safety content filtersfalse
--qualityImage quality: low, medium, high, hd (gptimage only)medium
--imageReference image URL(s), comma/pipe separated for multiple-
--transparentSet to true for transparent background (gptimage only)false

Available models

Free models (always available):

  • flux — Fast, high-quality image generation (default)
  • zimage — High-quality image generation
  • imagen-4 — Google Imagen 4
  • klein — Klein image generation
  • klein-large — Klein large image generation
  • gptimage — GPT Image (supports quality and transparent options)

Paid models (requires POLLINATIONS_PAID=true):

  • seedream / seedream-pro — Seedream image generation
  • kontext — Kontext model
  • nanobanana / nanobanana-pro — NanoBanana image generation
  • gptimage-large — GPT Image large variant

Examples

Basic: just prompt and model

node {baseDir}/scripts/generate.mjs --prompt "a beautiful sunset over mountains" --model flux

Custom size (--width, --height)

node {baseDir}/scripts/generate.mjs --prompt "a panoramic futuristic cityscape at night" --model imagen-4 --width 1536 --height 768

Reproducible result (--seed)

node {baseDir}/scripts/generate.mjs --prompt "a red panda sitting on a branch" --model klein --seed 42

AI-enhanced prompt (--enhance)

node {baseDir}/scripts/generate.mjs --prompt "cat" --model zimage --enhance true

Negative prompt (--negative_prompt)

node {baseDir}/scripts/generate.mjs --prompt "a professional product photo of sneakers" --model flux --negative_prompt "blurry, low quality, watermark, text"

Safety filter (--safe)

node {baseDir}/scripts/generate.mjs --prompt "a fantasy battle scene" --model klein-large --safe true

Quality control (--quality, gptimage only)

node {baseDir}/scripts/generate.mjs --prompt "an oil painting of a library interior" --model gptimage --quality hd

Transparent background (--transparent, gptimage only)

node {baseDir}/scripts/generate.mjs --prompt "a cartoon robot mascot" --model gptimage --transparent true

Reference image (--image)

node {baseDir}/scripts/generate.mjs --prompt "make it look like a watercolor painting" --model kontext --image "https://example.com/photo.jpg"

Save to a specific path (--output)

node {baseDir}/scripts/generate.mjs --prompt "a watercolor landscape" --model flux --output /tmp/landscape.jpg

Combining multiple options

node {baseDir}/scripts/generate.mjs --prompt "a detailed fantasy map" --model gptimage --width 2048 --height 2048 --quality hd --seed 123 --negative_prompt "blurry, text" --safe true --output ./map.png

Important notes

  • If --output is specified, the file is saved to that path; otherwise a timestamped filename is used in the current directory.
  • Model availability depends on the POLLINATIONS_PAID env var. If POLLINATIONS_PAID is not true, only free models can be used.
  • Choose the model based on the user's requirements. Default to flux for images if no preference.
  • When crafting prompts, be detailed and descriptive for best results.
  • Always tell the user the output file path after generation completes.