Closer #012

Imagine a cat behind a closed door, only paws and nose peeking from under the small crack between the door and the floor. Now imagine it again. How different are those results? Quite a lot, as it turns out.
The image shared with ChatGPT as input.

Welcome to Closer,

A weekly experiment in AI interpretation: I share a photo, ask AI to describe it, then generate an image from that description. This is edition #012.

Small adjustments in the input gets you small adjustments in the output. No surprises there, unfortunately. As we proceed...

Human experience → photo → AI description → AI image. Prompts are in grey boxes. Descriptions are in italics.

LLMs used

GPT-4o for image interpretation and image generation.

The input

✍️
Please describe the attached image in detail.
The image shared with ChatGPT as input.
The image shared with ChatGPT as input.

The output

The image humorously captures a cat partially hidden behind or under a white surface, likely a door or piece of furniture. Only the lower portion of the cat is visible, showing its pink nose, a bit of its mouth, and its front paws peeking out from the narrow gap.

Details:

  • The cat appears to be ginger or orange and white, as indicated by the color of the visible fur.
  • Its paws are positioned flat on the ground, with white fur on the toes.
  • The surface above the gap is clean and minimal, possibly a cabinet or a door, emphasizing the simplicity of the scene.
  • The lighting is soft and natural, creating a calm and intimate feel.

Overall, it's a playful, cozy, and slightly mysterious photo that evokes curiosity—capturing the classic cat behavior of hiding in small, unexpected spaces.

✍️
Generate an image with an aspect ratio of 3:2 based on this description: {the description previously generated}
The image generated by ChatGPT based on its description of my image alone.
The image generated by ChatGPT based on its description of my image alone.

Impressions

Last week we 'dumbed things down' and simply asked the image to be described. This week I've asked it to be described in detail. I was hoping that would give me a more elaborate description, as we've seen in previous weeks, but all we got is a 'Details:' header at the top of the bullet points.

The resulting image? Vastly different from the input. The composition is not at all similar, the subject is shown from a different angle, and the overall result is much more... boring? I personally think my original photo is a more interesting image to begin with, though personal preferences can differ there.

That being said, the AI-gen cat does look a lot like my actual cat though. Perhaps that's the most surprising take away this week.

Is there anything else that caught your eye, that I missed? Feel free to send a reply.

See you next week.

Mitch

Subscribe to the monthly newsletter

Just you, me, and some occasional notes from the field. No spam.

Join the conversation