Image Generator

Generate AI images from text or reference photos. Product shots, character art, posters with readable text, photoreal portraits, high resolution in seconds.

0.2

Model

Prompt(optional)

Reference Images

Upload product photos

Up to 5 images, different angles work best

Browse library

Reference Videos

Upload reference videos

Up to 2 short clips

Browse library

Aspect Ratio

Quality

Still needed: Prompt, Reference Images

Runs on

API

Examples

Nano Banana Pro

Pyra by lantern light (Nano Banana Pro)

Nano Banana Pro. The same prompt on the premium tier: crisper detail in the frill and scales, richer light falloff from the lantern. Prompt only, no reference image.

Reproduce

Best for

•Niche/specific subjects with no stock footage (historical recreations, sci-fi concepts, fantasy scenes)
•Conceptual visuals (abstract ideas, future technology, impossible scenes)
•Consistent visual style across multiple images (same lighting, color palette)
•Product visualizations and mockups
•Posters with readable text across Latin and non-Latin scripts

Tips

✓Be extremely specific: 'Photorealistic futuristic electric vehicle on a mountain road, dramatic low angle, golden hour, cinematic lighting' not 'car on road'
✓If the output drifts toward illustration when you want a photo, add 'photorealistic' explicitly
✓Style anchors that work: 'cinematic', 'dramatic lighting', 'volumetric fog', 'shallow depth of field'
✓Use reference images to keep the subject, pose, or product shape consistent across multiple generations

Recipes using this pipeline

One prompt → a vertical AI dance reel

Generate a labeled dance-move grid with GPT Image 2, hand it to Seedance 2 Pro as a reference so the choreography actually follows the moves, then stitch it into a music-synced vertical reel.

6 pipelines · CLI

Articles about this pipeline

Image Generator API

Call this pipeline from your own code. One request dispatches a run; the model is an input field, not a separate endpoint.

6 inputs

0.2 to 6.4 credits

TypeScript Python Go CLI

bun add @pipe2-ai/sdk

import { createClient } from "@pipe2-ai/sdk";

const client = createClient(process.env.PIPE2_TOKEN!);

const { run_pipeline } = await client.RunPipeline({
  pipeline_slug: "image-generator",
  input: {
    prompt: "A vintage brass telescope on a wooden desk, warm afternoon sunlight",
    model: "gemini-3-1-flash-image",
  },
});

pip install pipe2

import asyncio
import os

from pipe2 import Pipe2Client

client = Pipe2Client(token=os.environ["PIPE2_TOKEN"])

run = asyncio.run(client.run_pipeline(
    pipeline_slug="image-generator",
    input={
        "prompt": "A vintage brass telescope on a wooden desk, warm afternoon sunlight",
        "model": "gemini-3-1-flash-image",
    },
))

go get github.com/pipe2-ai/sdk-go

client := pipe2.NewClient(os.Getenv("PIPE2_TOKEN"))

input := json.RawMessage(`{
  "prompt": "A vintage brass telescope on a wooden desk, warm afternoon sunlight",
  "model": "gemini-3-1-flash-image"
}`)

run, err := pipe2.RunPipeline(context.Background(), client, "image-generator", input)
if err != nil {
    log.Fatal(err)
}

brew install pipe2-ai/tap/pipe2

pipe2 pipelines run \
  --pipeline image-generator \
  --input-json '{"prompt": "A vintage brass telescope on a wooden desk, warm afternoon sunlight", "model": "gemini-3-1-flash-image"}' \
  --wait

Get an API token →

Parameters

prompt one of: Text description of the image. The AI enhances this with composition, lighting, and style.

string
reference_images one of: Optional: 1-5 reference image URLs to guide the output style and composition.

array
aspect_ratio: Output aspect ratio.

1:116:99:164:33:4 default 1:1

Show all 6 inputs

model: Image engine to use. Leave unset and Pipe2 routes to the best fit for your prompt.

autogemini-3-1-flash-imagegemini-3-pro-imagegpt-image-2 default auto
quality: Engine-aware. Auto picks the balanced tier for the chosen engine — GPT Image 2 medium, Gemini 3.1 Flash 2K. Fast renders smaller and cheaper (GPT Image 2 low, Gemini 3.1 Flash 1K); hd takes the top tier, which on GPT Image 2 costs many times auto. Gemini 3 Pro changes size at one price.

autofasthd default auto
reference_videos: Video references (first ~8s of each is used), only the Gemini 3.1 Flash model supports these

array

Use it from an AI agent

Point any MCP client at pipe2 and your agent gets these tools. It finds this pipeline, reads the same schema above, then runs it.

list_pipelines
get_pipeline_schema
run_pipeline
get_pipeline_run_status
request_upload

https://mcp.pipe2.ai/mcp

{
  "mcpServers": {
    "pipe2ai": {
      "url": "https://mcp.pipe2.ai/mcp",
      "headers": { "Authorization": "Bearer YOUR_TOKEN" }
    }
  }
}

Full API reference MCP setup CLI reference

Frequently Asked Questions

What can the image generator make?

Product photos, character art, posters with readable text in any script, photoreal portraits, editorial imagery, brand-anchored variations from a reference photo, and a lot in between.

Do I need both a prompt and reference images?

No, either works on its own. A prompt creates from text; references create variations; both together give the most control.

How many reference images can I use?

Up to 5. More references help anchor subject identity, pose, or product shape across variants.

What image quality can I expect?

Production-grade across all three models. The premium model renders the cleanest typography and the highest-fidelity photoreal portraits; the balanced model is the best pick when you bring reference images.

Can it render text inside the image?

Yes, including non-Latin scripts (Japanese, Korean, Chinese, Hindi, Bengali, Arabic). Quote the literal text in your prompt for cleanest results.

How many credits does it cost?

0.2 to 6.4 credits, depending on the model you pick (or that Auto picks for you).

Explore more pipelines

See all →

from 9.5

Video Generator →

Generate AI videos from a text prompt, image, or reference clip. Cinematic shots, product reveals, lifestyle ads, with synchronized audio and natural motion.

Veo 3.1 Seedance 2.0