ideogram4

Generates structured JSON captions for Ideogram 4 text-to-image generation with legible on-image text.

2.0k|348|Updated Dec 9, 2025
One-click install
npx skills add https://github.com/digitalsamba/claude-code-video-toolkit --skill ideogram4
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ideogram4
Source: https://github.com/digitalsamba/claude-code-video-toolkit/tree/main/.claude/skills/ideogram4
Command: npx skills add https://github.com/digitalsamba/claude-code-video-toolkit --skill ideogram4

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ideogram 4's superior in-image text rendering and color/layout control are locked behind a structured JSON caption format, and plain-text prompts produce generic results that miss the model's strengths. This Skill teaches the AI to act as the caption expander, converting casual requests into the exact JSON schema the model was trained on.

Core Features & Use Cases

  • Structured JSON Caption Generation: Builds captions with the required top-level fields, strict key ordering, y-first bounding boxes (0-1000 normalized), and uppercase hex color palettes.
  • Legible On-Image Text: Places literal strings in dedicated text elements so title cards, thumbnails, quote cards, and CTAs render readable text instead of garbled glyphs.
  • Brand Color and Layout Control: Pins exact brand hex colors per element and positions elements with bounding boxes for layout-critical images.
  • Use Case: A user asks for a YouTube thumbnail with the words "SHIP FASTER" and a rocket; the Skill produces a complete JSON caption with the headline as a text element, brand palette, and bounding boxes, then posts it via the hosted v4 API using tools/ideogram4.py.

Quick Start

Ask the AI to create a title card image with the headline "AI ENGINEERING REVIEW" on a dark tech background using the ideogram4 skill.

Frequently Asked Questions about ideogram4

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate images with legible text using Ideogram 4?

Build a structured JSON caption instead of a plain-text prompt, placing the literal string in a text element's text field. Keep rendered strings short (1-3 word headlines plus a subline) and use bounding boxes to control placement.

Ideogram 4 vs FLUX.2 for image generation, which should I use?

Use Ideogram 4 when the image needs legible on-image text, exact brand hex colors, or controlled layout via bounding boxes. Use FLUX.2 for text-free atmospheric backgrounds or when you want simple natural-language prompts.

Does Ideogram 4 require an API key or self-hosted weights?

The toolkit uses Ideogram's hosted v4 API, which requires an IDEOGRAM_API_KEY from developer.ideogram.ai in the .env file. Paid API plans include a commercial license, while the self-hostable open weights are non-commercial.

Why is my Ideogram 4 text garbled or in the wrong position?

Garbled text usually means the string was buried in a description instead of a text element's text field, or the string is too long. Wrong placement is fixed by adding a bbox in y-first order [y_min, x_min, y_max, x_max] with 0-1000 normalized coordinates.

How do I set exact brand colors in Ideogram 4 images?

Add uppercase hex values like #1456F0 to the color_palette field, using per-element palettes to pin colors on specific items like headlines or logos. The global style_description palette supports up to 16 colors and each element up to 5.

What are the limitations of Ideogram 4 text rendering?

Ideogram 4 handles headlines, sublines, and short CTAs well but degrades on paragraphs of body copy. Keep each text element to a short string and split multi-line content into separate elements with their own bounding boxes.