L

Learn AI

Track progress · learn offline

Open

Lesson 4 of 7

Getting the same picture twice

7 min read

You finally get the perfect image. You ask for "one more like that" — and get something completely different: a new pose, a new face, a new mood. The magic won't repeat. Why not, and how do the pros pin a good result down?

The same words, a new picture every time

Remember how a picture grows out of a field of random static? Every run starts from a fresh patch of that static, so the very same words clear into a slightly different picture each time — a different fox, a different pose, a different sky. That's why "one more like that" gives you a stranger. It isn't a bug; the randomness is baked into how these tools work.

The same prompt gives a different image every run because each one starts from a new patch of random noise. Sameness isn't the default — you have to ask for it.

Meet the seed

Here's the lever. That starting patch of static isn't truly random to the computer — it's picked by a single number called the seed, like the roll of a die that decides where the noise begins. Same prompt plus the same seed always clears into the exact same picture. Most tools show you the seed of any image and let you paste it back in. So the moment you get a keeper, you grab its seed — now you can return to that exact picture any time.

A seed is the number that fixes the starting noise. Same prompt + same seed = the same picture, every time. It's your "save" button for a result you like.

Lock it, then nudge it

The seed is also how you make small changes instead of starting over. Lock the seed and the composition holds still; now change one word in the prompt and only that one thing moves — the same scene, at night instead of at noon. Unlock it and every roll is a fresh gamble again. Try both below.

Lock the seed to hold the picture steady, then change one word for a controlled tweak. Unlocked, every roll is a brand-new gamble; locked, you're editing one thing at a time.

Keeping a character the same

The hardest version of this is keeping the same character across many images — the same face in scene after scene. A locked seed keeps one picture stable, but the moment you change the scene much, the face drifts. That's why modern tools add a character reference: you give them a picture of your character and they carry that likeness across new images — the same idea as a style reference, aimed at who's in the frame instead of the overall look.

When you land an image you love, save its seed and its exact prompt together. That pair is the recipe — without it, that specific picture is gone for good.

The gist

You generated a great portrait and want the exact same one back tomorrow to keep working on it. What do you need to save?

Continue in the app

Take the whole AI Images course — tracked

Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.