logo
Start creating

How to generate images using the gen-ai-images skill

SKILLS3 minBeginner

Import the image generation skill to access Flux, Recraft, Nano Banana, and 20+ models for still visuals.

How to generate images using the gen-ai-images skill

What you'll learn

  • How to import the gen-ai-images skill into your agent workspace
  • How to select the right image model for your use case
  • How to control aspect ratios and style parameters
  • How to batch-generate multiple variations at once

What is the gen-ai-images skill?

The gen-ai-images skill gives your AI agent the ability to generate still images using 20+ frontier models including Flux, Recraft V4, Nano Banana, Imagen, and Ideogram. It's like adding a design team to your coding assistant — ask for a hero image, product shot, or illustration, and your agent picks the right model and generates it locally.

Common use cases

  • Web design: Generate hero images and background visuals for landing pages
  • Marketing: Create product shots and campaign assets in batch
  • Content writing: Auto-generate blog illustrations and featured images
  • Social media: Build branded graphics for Instagram, Twitter, and LinkedIn
  • Prototyping: Mock up UI elements and placeholder visuals
  • E-commerce: Generate lifestyle product photos at scale

Generate your images step by step

STEP 1: Download and import the skill

  • On web: Go to picsart.com/cli/#skills-starter → Download gen-ai-images → Extract to your agent's skills folder
  • On mobile: Use desktop to download — this skill requires a development environment
Get the skill

STEP 2: Choose your model and aspect ratio

Ask your agent to generate an image and specify your preferences:

  • Flux 2 Pro: Best for photorealistic images and detailed subjects
  • Recraft V4: Ideal for illustrations, vector art, and brand graphics
  • Nano Banana: Fastest option for quick iterations and drafts
  • Imagen: Strong compositional control and text rendering
  • Aspect ratios: 1:1, 16:9, 9:16, 4:5, 2:3, or custom dimensions

STEP 3: Generate and save

Your agent executes the generation command. The image saves to your project folder automatically, typically in ./output/ or your current directory. Check your terminal for the exact path and filename.

STEP 4: Review and iterate

Check your generated image for quality and accuracy: Not happy with the result? Refine your prompt with more specific details about lighting, composition, or style, then generate again.

  • Verify the style matches your request (photorealistic vs. illustrated)
  • Check that subject details are accurate and clear
  • Confirm aspect ratio and resolution meet your needs
Start generating

Tips for best results

💡 Be specific about style and mood

Instead of "a coffee cup," try "a ceramic coffee mug on a wooden table, soft morning light, warm tones, overhead view." The more detail you provide, the closer the output will match your vision.

💡 Generate multiple variations at once

Ask your agent to create 3-5 versions of the same prompt with slight variations. This gives you options to compare and helps you learn which phrasing works best for your style.

💡 Match your model to your use case

Use Flux for realistic photos and human subjects. Choose Recraft for clean illustrations and brand assets. Pick Nano Banana when speed matters more than perfection. Each model has different strengths.

Aspect ratio guide

  • 1:1 (Square): Instagram posts, profile images, icons — 1024×1024px
  • 16:9 (Landscape): YouTube thumbnails, hero images, presentations — 1920×1080px
  • 9:16 (Portrait): Instagram Stories, TikTok, mobile screens — 1080×1920px
  • 4:5 (Vertical): Instagram feed posts (portrait) — 1080×1350px
  • 2:3 (Standard photo): Pinterest pins, print photos — 1000×1500px
  • Custom: Specify exact dimensions for non-standard sizes

Frequently asked questions

Flux excels at photorealism and complex subjects like people and products. Recraft produces clean, vector-style illustrations perfect for branding and UI. Nano Banana prioritizes speed over detail, great for rapid iteration. If unsure, start with Flux for photos or Recraft for graphics.

No. The skill is a wrapper that simplifies CLI access for your agent. When you import the skill, it includes the CLI binary and handles authentication. You don't interact with the CLI directly — your agent does — but it's part of the package.

Most models output PNG files by default for quality and transparency support. Some models may offer WebP or JPEG options. Your agent can convert formats after generation using standard tools if needed.

Generation limits depend on your Picsart account credit balance. Each image costs 1-5 credits depending on model and resolution. The skill will notify you if you're low on credits before starting a large batch.

Ready to generate stunning images?

Import the gen-ai-images skill and start creating visuals with 20+ frontier models.

Download skill