What is a Midjourney seed?

A seed is the starting number Midjourney uses to initialise the noise pattern for a generation. Reusing the same seed with the same prompt and version produces a very similar (though not always pixel-identical) result, which is useful for consistency and controlled experiments.

How do I find the seed of an image I already made?

In Discord, react to the finished image with the envelope emoji and the bot will DM you the seed number. In the Midjourney web app the seed is shown in the image's details panel. Paste that number into this tracker to save it.

Does reusing a seed guarantee the same image?

Not exactly. The same seed plus the same prompt and version gives a very close result, but Midjourney is not fully deterministic across model updates. Treat seeds as a strong consistency tool, not a perfect reproduction guarantee, and pin the version too.

Where is my seed library stored?

Entirely in your browser's local storage. Nothing is uploaded to any server, so your seeds are private to this device and browser. Use the Export button to save a JSON backup or move your library to another machine.

How do I use a saved seed in a prompt?

Copy the --seed flag this tool generates and append it to your /imagine prompt, for example "/imagine prompt: a red fox --seed 12345". Combine it with the same version flag you originally used for the most consistent results.

What is the Midjourney Seed Manager?

Local browser tool to store Midjourney seed numbers alongside their prompt snippets and style tags. Filter by style, export your library as JSON, and copy a --seed flag with one click. Data stays in your browser only. It runs free in your browser on Gera Tools, with nothing uploaded.

Midjourney Seed Manager

Name: Midjourney Seed Manager
Creator: Gera Tools
License: https://creativecommons.org/licenses/by/4.0/

Seeds are the key to consistency in Midjourney — reuse one and you can iterate from the same starting point instead of rolling the dice each time. The catch is that seed numbers appear only briefly and are easy to lose. This manager keeps a private, browser-local library of your seeds with the prompt and style that produced them, so you can find and reuse the good ones in seconds.

Why seeds matter for creative workflows

Every Midjourney generation starts from a random noise pattern. The seed number determines which noise pattern is used. When you reuse a seed with the same prompt and the same model version, Midjourney initialises from the same noise and arrives at a very similar image — not pixel-identical, but close enough to iterate from.

This makes seeds valuable in several workflows:

Style exploration: Found an image with exactly the right atmosphere? Save the seed and start varying the prompt around it — changing subject, adding descriptors, adjusting parameters — while the underlying visual signature stays consistent.

Character consistency (without —cref): For older Midjourney versions that lack --cref, a seed paired with a detailed character prompt is the main tool for keeping a character recognisable across images.

Prompt A/B testing: Hold the seed constant and change one prompt variable at a time. The shared starting noise means you are genuinely comparing what changed in the prompt, not just seeing two different random outcomes.

Recovering a look you liked: Even if you have moved on from a generation, its seed number lets you return to the same starting point days or weeks later.

How the seed tracker works

Everything runs in your browser using local storage — no account, no server. Each entry stores:

Seed number — the integer Midjourney reports for a generation.
Prompt snippet — a short reminder of the prompt that produced a useful result.
Style tags — comma-separated labels such as portrait, neon, v6 so you can filter by workflow or look.

The filter box matches against both prompt text and tags. For any entry, copy a ready-formatted --seed N flag with one click, or export the entire library as JSON for backup or cross-device use.

How to find a seed you have already generated

In Discord: React to the finished image with the envelope (✉️) emoji. Midjourney’s bot sends you a DM with the job details, including the seed number. Do this immediately after upscaling — you cannot easily retrieve seeds for old jobs in bulk.

In the Midjourney web app: Open the image details panel; the seed is shown in the generation metadata alongside the prompt.

Tips for a useful seed library

Pin the model version in your prompt snippet. A seed produces consistent results only when paired with the same --v value it was originally run on. A seed from v5 will not produce the same image on v6. Note this as a tag or in the prompt text.
Save seeds from failed experiments too. Sometimes a generation that did not work for its original purpose has exactly the noise pattern you will want later.
Build a consistent tag vocabulary. A small set of tags — portrait, landscape, texture, abstract, character, product — is far more useful than different ad hoc labels for each entry.
Export to JSON periodically. Browser local storage can be cleared by browser settings, updates, or private-mode sessions. A JSON export takes two seconds and guarantees you never lose a seed you spent hours finding.