Compare_Images

Topic: CNTower

Same prompt, six images across five different generators — side by side.

Context

I have a database with my GTA 5km and 10km races. Most races have a race route map. This is specifically for the CN Tower climb since there really is no race route map.

Prompt used

The CN Tower climb does not have a race route map, since there is not route, but 1776 steps over 144 flights of stairs. Generate an image that tells the user in a humorous way there is no race route map.

CNTower image generated by Adobe Express (Firefly 4)

Adobe Express (Firefly 4)

CNTowerAdobeXFirefly4400.jpg

Comments

CN Tower anybody? Is the white stuff snow?

CNTower image generated by Adobe Express (Gemini Nano Banana)

Adobe Express (Gemini "Nano Banana")

CNTowerAdobeXGemeniBanana.jpg

Comments

Not bad

CNTower image generated by ChatGPT

ChatGPT

CNTowerChatGPT400.jpg

Comments

I like it

CNTower image generated by Claude Opus

Claude (Opus)

CNTowerClaudeOpus400.jpg

Comments

But is is humorous?

CNTower image generated by Midjourney, scary variant

Midjourney (scary)

CNTowerMidjourney400scary.jpg

Comments

Oh yes, an outside staircase

CNTower image generated by Midjourney, down variant

Midjourney (down)

CNTowerMidjourney400down.jpg

Comments

Something tells ne Midjourney either needs different prompts or it is not my tools for visuals

Final thoughts

Originally all images were large png files, I resized them and saved them as jpg files. Gemini Nano Banana and ChatGPT do a pretty good job. While Claude is my main source of AI, it is not my first choice for non technical visuals. Midjourney is an entirely different kettle of fish.