Generative AI Review

Category: Art

  • The Ghost in ChatGPT-4o’s Image Generator: What the Prompts Don’t Tell You

    The Ghost in ChatGPT-4o’s Image Generator: What the Prompts Don’t Tell You

    How I uncovered the hidden information that makes AI image replication possible—and why it matters for the future of generative AI

    [cross post from substack]

    Overview

    This is a summary of my substack post about the same issue. Here I get better image galleries to compare the images side by side.

    Side by Side

    Source Image
    WITH hidden Information
    WITHOUT hidden information

    ChatGPT will give you what it considers the prompt for recreating the image

    Here’s the exact text prompt you can use to regenerate the same image in a fresh chat:

    > A photograph taken on April 5, 2025, features a young black and tan mixed-breed puppy sitting on a vintage-style patterned area rug. The puppy has wavy, slightly curly fur with distinct tan markings on its eyebrows, muzzle, chest, and legs. It sits calmly, facing forward with large brown eyes and a shiny black nose. Behind the puppy is an old wooden chest with visible grain texture and a metal keyhole, adding a rustic charm to the scene. The bottom left corner of the image is signed with the date “2025-04-05” as evidence. The lighting is warm and natural.

    You can give the prompt a go yourself, you should get consistent results. I asked a few friends on LLM London WhatsApp to generate the images (thanks!) and they were very similar

    Here is one generated by Midjourney from the same disclosed prompt:

    From midjourney but by the same disclosed prompt
  • OpenAI 4-o Image Generation – First Look – Transforming Art Styles

    OpenAI 4-o Image Generation – First Look – Transforming Art Styles

    OpenAI just announced their latest image generation tool 4-o. I’ve taken two of the advertised features for a test drive

    • Character Consistency
      • Same subjects in between generations
    • Upload and Restyle
      • POH can supply an image or photo to begin the work

    Both ✅Character Consistency and ✅Upload and Restyle worked very well as you will see below

    Source Image

    This is an photographic image I took IRL on an iPhone 13 Pro Max.

    Puppy Dog Source Image
    Puppy Dog Source Image

    OpenAI Image Generation 4-o – Variations on Source Image

    https://openai.com/index/introducing-4o-image-generation

    create this puppy image as an anime

    Notes:

    • 1 minute generation time
    • ✅Consistent character and setting
    • ✅Image aspect ratio preserved
    • ✅Object on top of box preserved
    • ✅Well known target style ‘anime’ was understood

    create that image in claymotion

    Notes:

    • 1 minute generation time
    • ✅Consistent character and setting
    • ❌Image aspect ratio not preserved. Became square
    • ✅Object on top of box preserved (but clipped)
    • ✅Fuzzy description ‘claymotion’ was sufficient for the LLM to understand the intention.