ImageAI

• 

  "Self-portrait". This is actually scaled-down by two. Link to full-scale below.

Inspiration

A post from Google about Nano Banana drop. They dropped one day, I dropped first version the next.

What it does

• Use LLMs, Image and video generators of your choice
  • Can work completely local, with only free downloads
  • Or use the best image generators
• For online services authenticate with
  • API keys
  • Cloud for Google and Anthropic
• Use reference images for image and video
  • Up to max allowed for each model
• Use AI to generate or enhance image prompts
• Ask about Files
• Ask about image prompt
• Build prompts using a built-in database of artists, styles, etc.
• Built-in examples and templates
• Full history
• Complete help

How I built it

Primarily with Claude Code. I used ChatGPT and Gemini a few times, early on.

Challenges I ran into

Too many to list here. Most are documented. These stand out:

• The video tab held special challenges to get the prompt generator to work with lyrics and music, or only text, depending on what you use.
• Getting all AI to play nicely was sometimes tricky.
• Recently, the font generator had special challenges recognizing and aligning glyphs.

Accomplishments that I'm proud of

The Font generator, AI video storyboard generation, complete with lip-sync and video prompt generators of your choice.

What I learned

Everything in the project is a learning project for me.

What's next for ImageAI

Enhance video generation.

Full-scale image on Flickr

Built With

• chatgpt
• claude
• claudecode
• gemini
• jetbrains