I’ve been running side-by-side tests of ChatGPT’s new 4o image generator (launched just two weeks ago) and Midjourney — and the results were full of surprises.
Here’s what I discovered, along with a few tips if you’re experimenting with these tools yourself:
1. Branded Product Images
I used both tools to create a mock product for Sustain Well — a fictional, environmentally conscious skincare brand I invented for an AI talk in Istanbul last year.
- ChatGPTrendered a clean product shot with the correct brand name on the label.
- Midjourney offers far more customisation and creative freedom — it excels in lighting, texture and atmosphere — but still struggles with text accuracy, even when using precise prompts.
If your visual needs include actual copy, ChatGPT’s image generator is currently more reliable.
Midjourney can get it right with very specific prompting (and I did manage a good example), but even when the brand name appears correctly, smaller supporting text often turns out as gibberish.
2. People-Centric Scenes
I prompted both tools to create a diverse group of people unboxing Sustain Well products.
Both produced strong results:
- ChatGPT made it easy to iterate. I could remove individual people, change objects or tweak the layout without restarting from scratch.
- Midjourney delivered a more artistic, cinematic result — but once an image is generated, it’s less straightforward to make changes (like removing one person from the scene, whilst keeping the same image).
That said, I love Midjourney’s creativity and how easily it blends people and products into artistic compositions.
3. Personalised Model (Midjourney Beta v7)
Midjourney’s new beta version allows you to “train” it on your visual preferences by selecting from 200 image pairs.
It’s a fascinating feature, but in my case, the final outputs didn’t reflect my chosen style very closely.
Still, it’s early days for this feature and definitely worth keeping an eye on.
4. Combining AI with Real Photos (Coming Soon)
Next, I’ll be testing how each tool handles blending a real photo of me into AI-generated content, for example, creating a scene where I’m holding Sustain Well products.
This is where things get especially interesting and where storytelling and personal branding can really come to life. I’ll share results from that experiment soon.
5. Top Tips for Using ChatGPT + Midjourney
- Prompt like a pro: The more detail you include (style, lighting, background, colour palette, mood), the better the output. Check out: Best Midjourney Prompts for Product Photography
- Refine iteratively: ChatGPT allows quick, edit-friendly adjustments.
- Master Midjourney parameters for a more artistic lookeird parameter! (Tip: look up the full list of Midjourney parameters)
- Use reference images – Both tools allow you to guide generation using image inputs.
- Know the limitations – Midjourney = poor at text. ChatGPT = not yet as stylised or artistic.
Want to Learn How to Use These Tools to Build a Full Campaign in 4 Hours?
I teach a practical, hands-on course called AI Marketing Mastery
You’ll learn how to use AI to create:
- Campaign visuals and promotional imagery
- Video scripts and assets
- Scroll-stopping ad creatives
- Entire content packs generated fast
Perfect for marketers, founders, freelancers, and creators looking to level up with GenAI (no prior experience needed).
Have you been experimenting with AI image tools? I’d love to hear what’s working for you — drop your thoughts or questions in the comments ⬇️
#genai #marketing #midjourney #chatgpt #branding #aiimagegeneration #creativetools #aimarketing
