You need a favicon. You need a hero image. You need an OG social card. You need an apple-touch-icon. Each one has different dimensions, different requirements, and different platform quirks. And AI image generators love to render text into every image unless you explicitly stop them.
The Image Prompt Gen tool handles all of this. Pick what you need, enter your brand details, choose Midjourney or Ideogram, and get a prompt you can paste directly. It also gives you the ImageMagick commands to convert the output into every size you need.
What it generates
For Midjourney users: a formatted /imagine prompt: with the correct --ar ratio, --style raw for brand assets, --v 7 for the latest model, and --no text, words, letters to suppress text rendering.
For Ideogram web users: a clean prompt with recommended Style Type, Aspect Ratio, and Magic Prompt settings listed beside it.
For Ideogram API users: the prompt text plus a ready-to-run curl command for the v3 endpoint, a Node.js fetch() snippet, and documentation for the v2 fallback endpoint. All you add is your API key.
Why text suppression matters
AI image models default to rendering words into images. For brand assets, you almost never want that. A favicon with tiny unreadable text looks worse than a clean icon. A hero image with AI-hallucinated captions looks amateur.
The tool appends a long negative phrase to every prompt: "no text, no words, no letters, no typography, no labels, no watermark." For Midjourney it also sets the --no parameter. This blocks text rendering roughly 80-90% of the time. When text leaks through, regenerate rather than accepting the image.
ImageMagick post-processing
Every prompt output includes a collapsible section with the exact ImageMagick commands for the image type you selected. For a favicon set, that means generating favicon.ico (multi-size 16+32+48), apple-touch-icon.png (180x180), PWA icons (192 and 512), and converting to WebP. For a hero image, it is a single command:
magick source.png -quality 82 -resize 1600x900^ -gravity center -extent 1600x900 hero.webp
This is the same command I use for every blog hero image on jwatte.com. The ^ flag means "fill" not "fit," so the image crops to exact dimensions instead of letterboxing.
When to use which platform
Ideogram is better when you need flat editorial illustrations, clean geometric compositions, or reliable text suppression. It also has a real API, which matters if you are generating images programmatically.
Midjourney is better for photorealistic brand photos, complex stylized compositions, or images where artistic quality matters more than strict control. It has no public API, so you work through the Discord bot or the web UI at midjourney.com.
For favicons specifically, Ideogram at 1:1 aspect ratio with DESIGN style type produces the cleanest results. For hero images, either platform works well at 16:9.
Related reading
- How to Generate Favicons and Brand Images with Midjourney and Ideogram
- ImageMagick for Favicon Sets and Brand Image Processing
- Your Site Has No Favicon and It's Costing You Trust
Fact-check notes and sources
- Ideogram API v3 endpoint: developer.ideogram.ai/api-reference
- Midjourney parameters: docs.midjourney.com
- ImageMagick resize modes: imagemagick.org/script/command-line-processing.php
This post is informational, not design advice. Mentions of Midjourney, Ideogram, and ImageMagick are nominative fair use. No affiliation is implied.
