"Studio product photo of a [object], [light direction] light, [angle] angle, neutral seamless background, 4 variants"
Names the angle + light explicitly. gpt-image-2 understands camera vocabulary precisely.
NEWPOPULAR
gpt-image-2
Best for editorial portraits, product photography, and brand work that needs precise typography. Understands camera vocabulary and renders in-image text reliably.
Selected works.
What it can
do well.
- ✓Generate from text promptsupported
- ✓Edit existing imagessupported
- ✓Reference image inputup to 8 refs
- ✓In-image text renderingsupported
- ✓Inpaint with masksupported
- ✓Expand canvassupported
- ×Native 4K outputuse upscale tool
- ×Video generationnot supported
Prompts that hit.
"A poster reading "[exact text]" in [font style] on [background]. The text must be sharp and legible."
Quote the text. Specify font feel. The "sharp and legible" clause helps a lot.
"Replace the background with [scene]. Keep the subject untouched. Match the original lighting direction."
For edits, the "keep X untouched" clause is the difference between good and clean.
Compare gpt-image-2.
All comparisons →FAQ.
Q.01
When does the auto-router pick gpt-image-2?
When the prompt's intent matches gpt-image-2's strengths and the cost is lower than alternatives that meet the same quality bar.
Q.02
Can I pin this model as default?
Yes. The model picker in any chat lets you pin per-turn or per-session. Set as default in /settings.