Midjourney has been rolling out several new features to fight back against ChatGPT 4o’s powerful imaging capabilities. A new feature is “Omni-Reference” that lets you upload photos of “characters, objects, vehicles, or non-human creatures”.
I uploaded an older profile photo and prompted
“a cleanshaven bald man with glasses in a mecha suit standing outside a space station, dramatic lighting”.
“a 1940s movie poster of a bald man with glasses carrying a space gun, facing down many aliens”
“a portrait of a smiling bald man with glasses made with dramatic colorful paintstrokes, man is looking at camera”
Midjourney generated a good likeness about 20-30% of the time, but its strength is in the aesthetic styling of the image.
So, should you use ChatGPT or Midjourney? As of May 2025, this is my take:
ChatGPT
➕ Image generation available to free users
➕ Follows prompts well, even complex ones
➕ Generates text rather accurately
❌ Very slow (3-10 min)
❌ Poor editing features
❌ Images get generic fast, since millions are using it
Midjourney
➕ Beautiful aesthetics that is putting artists out of work
➕ Fast generation (usually 1 min) and an even faster “draft” mode
➕ Advanced editing features
❌ Only available to subscribers (from $10 USD monthly)
❌ Terrible with text generation
❌ Doesn’t always follow your prompt. (It keeps adding a goatee to my pix)
I think all marketers and communicators should use both apps. ChatGPT as an all-rounder LLM where image generation is a bonus, and Midjourney is for making differentiated images to get away from AI-slop.