A picture that googles for itself and knows what's happening today
Here's the idea in one line: you ask "make a poster about today's weather in my city" — and you get a finished image where the numbers are today's, not invented. A morning poster for the day, a "good morning" card with the real temperature, a social card with a current fact — and it's all one picture.
And here's what's fresh: a normal model can't do this. It draws the world as it memorized it during training and knows nothing about "today" — it gives you a plausible but made-up number. But Seedream 5.0 from ByteDance added a mode where the model searches the web itself during generation: it sees something about current events in your prompt, goes online, pulls a fresh fact and bakes it into the picture. As the developers put it, the result "feels timely instead of stale." That's what this project rides.
What you'll learn
The project is small, but it carries the "fresh data → image" link that a pile of real products rely on — from event posters to product cards.
- Giving the model a task with "freshness." Asking not just for something pretty, but tied to today.
- Checking the fact with your eyes. The model searches on its own, but you still verify the numbers — a key skill for working with AI.
- Getting the image as a file. Pulling the picture out of the response and saving it so you can post or forward it right away.
A ready starter prompt
Don't write "draw a poster about the weather" — the model will grab a random number out of its head. Say plainly that the fact must be today's and where it comes from:
Make a nice poster about the weather in London.The difference: the strong prompt explicitly demands a fresh fact and forbids making one up — so the poster comes out honest, not a pretty mistake.
What the result looks like
A today.png: a clean poster for the day with a real today's number in the center. Swap weather for anything tied to "now" — an exchange rate, yesterday's match score, a fact of the day — and you've got a card generator that's always current, not stuck in last year.
Start with one topic close to you, get it down to a picture with a real fact — and you've got a pipeline that draws not "in general" but about today.
Short story-lessons, an agent simulator and daily practice — in our mobile app. Free.
Source: Seedream 5.0 Guide: Real-Time Web Search — RunDiffusion
Read next
Drop in a clip — the model watches it and draws the cover
A translator with no key, no server, no bill — right in the browser
One photo and one line — and out comes a talking postcard
A partner you talk to out loud — and cut off mid-sentence, like a real person
Your first AI feature with no key at all — Apple now ships the model inside the phone, free
