I used ChatGPT (Creative Studio) + Nano Banana (google) to build the thumbnail for my pinned YouTube video.
No designer.
No guessing.
Just constraints + iteration.
Here’s exactly what I did.
1. Shot My Own Photos
I took clean photos of myself in front of a green screen.
Different poses.
Neutral expression.
Calm authority.
Pro tip:
If you can get green screen shots of yourself, do it.
It gives AI full control to replace the background cleanly.
2. Gave Nano This Prompt
(You can copy this and use it. Just change any section you would like but keep the format)
Use my attached real photo as the base.
Do NOT recreate my face.
Do NOT generate a new person.
Enhance only.
Objective:
Create a serious, authority-driven YouTube thumbnail.
Crop:
Chest-to-head framing.
Remove excess space.
Make face dominant.
Expression:
Neutral.
Calm authority.
Direct eye contact.
No smile.
Background:
Dark minimalist office.
Deep blue / charcoal tones.
Blurred shelves.
No clutter.
Lighting:
Increase contrast slightly.
Subtle rim light.
Natural skin texture.
No AI smoothing.
Text (left side):
STOP REACTING
START OPERATING
Make “START OPERATING” larger.
Emphasize “OPERATING.”
Bold modern sans-serif.
No glow.
No drop shadows.
No Canva overlays.
Make it 10% more intense without becoming dramatic.
3. Iterated
It wasn’t perfect the first time.
So I refined:
- Crop tighter
- Remove empty space
- Increase contrast 5%
- Strengthen text hierarchy
- Remove semi-transparent text box
- Darken background slightly
Refine.
Tighten.
Repeat.
That’s it.
You can use this same method for:
- YouTube thumbnails
- Reel covers
- Course graphics
- Banner images
AI works best when you control the constraints.
I’ll keep sharing more breakdowns like this.