How FloImg Works

Image manipulation is fundamentally a series of steps. FloImg models this core abstraction and exposes it through interfaces that fit different contexts.

Try It in Studio Read the SDK Docs

The Pipeline Model

Every image workflow is a pipeline: generate → transform → save

Generate

Create from scratch

AI images (DALL-E, Imagen, Flux)
Charts & diagrams
QR codes
Screenshots

→

Transform

Modify deterministically

Resize & crop
Add text & captions
Adjust colors
Convert formats

→

Save

Store anywhere

Local filesystem
Amazon S3
Cloudflare R2
Tigris

Steps are composable. Chain as many transforms as you need. Each step takes an image and returns an image.

Why This Matters

Deterministic Transforms

AI editing is probabilistic—DALL-E generates new images, inpainting runs diffusion. FloImg transforms are mathematical: adjust hue by 30°, resize to exactly 1200×630, add caption at position (10, 10). The image stays intact except for exactly what you asked.

One Abstraction

The same pipeline concept works everywhere. No context-switching between "the visual editor's way" and "the API's way." Learn it once, use it in every interface.

Unified API

Stop wrangling OpenAI SDK, Sharp, AWS SDK, remove.bg, format converters. FloImg consolidates the patchwork of tools into one consistent interface.

Six Interfaces, One Pipeline

The same workflow definition is portable across all interfaces

SDK

Programmatic

Embed in any JavaScript/TypeScript application. Full type safety.

SDK Docs →

CLI

Terminal

Terminal workflows and CI/CD pipelines. No code required.

CLI Docs →

Studio

Visual

Drag-and-drop workflow builder. Perfect for prototyping.

Open Studio →

YAML

Declarative

Version-controlled workflow definitions. GitOps-friendly.

YAML Format →

MCP

AI Agents

Model Context Protocol for LLMs. Let Claude build workflows.

MCP Docs →

API

HTTP

REST API for any language. Execute workflows from any backend.

API Overview →

Same Workflow, Any Interface

Generate an AI image, resize it, and save to S3

import { floimg } from '@teamflojo/floimg';
import { openai } from '@teamflojo/floimg-openai';

await floimg()
  .generate(openai('dall-e-3'), {
    prompt: 'A mountain landscape at sunset',
  })
  .resize({ width: 1200, height: 630 })
  .save('s3://my-bucket/hero.png');

name: hero-image
steps:
  - type: generate
    generator: openai
    params:
      model: dall-e-3
      prompt: A mountain landscape at sunset

  - type: transform
    operation: resize
    params:
      width: 1200
      height: 630

  - type: save
    destination: s3://my-bucket/hero.png

# Run a YAML workflow
floimg run hero-workflow.yaml

# Or use the full API
floimg generate \
  --generator openai \
  --params '{"model":"dall-e-3","prompt":"A mountain landscape"}' \
  | floimg transform --op resize --params '{"width":1200,"height":630}' \
  | floimg save s3://my-bucket/hero.png

curl -X POST https://api.floimg.com/v1/workflows/execute \
  -H "Authorization: Bearer $FLOIMG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "steps": [
      {
        "type": "generate",
        "generator": "openai",
        "params": {
          "model": "dall-e-3",
          "prompt": "A mountain landscape at sunset"
        }
      },
      {
        "type": "transform",
        "operation": "resize",
        "params": { "width": 1200, "height": 630 }
      },
      {
        "type": "save",
        "destination": "s3://my-bucket/hero.png"
      }
    ]
  }'

Export from Studio to YAML. Embed YAML workflows in your SDK code. Run SDK pipelines from CLI. The concepts are identical—only the syntax changes.

Use It Your Way

Prototype → Production

Build workflows visually in Studio. Export to YAML. Embed in your app with the SDK. Same pipeline, different interface.

CI/CD Automation

Generate OG images on deploy. Resize assets in your build pipeline. Version control your YAML workflows alongside code.

AI-Powered Workflows

Let Claude or Cursor build image workflows via MCP. The LLM understands the pipeline model and creates working workflows.

Quick Tasks

Generate a QR code. Resize a batch of images. Convert to WebP. The CLI handles one-off tasks without writing code.

Start Building

Pick the interface that fits your workflow.

Try It in Studio Read the Docs