Inputs
optional · up to 14 imagesLabel each picture so the model knows what to keep: up to 4 characters and 10 objects. With nothing added it draws from the prompt alone.
Prompt
System instruction — none
Frame
Aspect ratio
Size
Images
Format
Thinking & search
Thinking — always on and billed; more plans the composition better
Search is free for the first 5,000 searches a month across Gemini 3 models, then $14 per 1,000. Google's suggestion chips show under the still.
People & safety
Google defaults
People in pictures
Well-known people — celebrities and public figures
Safety filters — Google's adjustable ones. Block none still scores; Off doesn't. Some things are blocked whatever you set
Filter scoring
Advanced
all defaults
Input detail — how finely it reads your inputs (more = more input tokens)
Seed — same seed + same settings ≈ same picture; ×2–4 use seed, seed+1…
Max output tokens — caps text + thinking; too low can stop the image
Stills
0No stills yet.
Write a prompt and press Shoot — or drop in pictures to edit or combine.