Midjourney personalization: create and reuse a taste profile
Select images to build a personalization profile, apply it deliberately, and save the code and test notes needed to return to an accepted look.
Updated 5 min read
Set up Midjourney personalization at Personalize > Global V7 Profile, select images you like from the grid until the profile unlocks, then apply it with the Personalization button beside the Imagine bar or with --p. Save the code from an accepted generation if you want to revisit that profile state later. Midjourney’s Personalization guide (opens in a new tab) distinguishes a changing profile ID from a code that identifies an older state. fenbs keeps the accepted code, brief, and review notes together when the next session starts.
How do you unlock the first profile?
Use the current image-grid workflow. Older instructions about choosing between pairs do not describe the current Global Profile unlocking steps. Before you start, decide whether you are teaching a broad personal taste or looking for a narrow visual direction for a particular piece of work.
- Open Personalize from the website navigation and find Global V7 Profile.
- Select images you like from the grid until the profile unlocks.
- After unlocking, make a small test with a simple prompt whose subject is easy to judge.
- Write down what you were selecting for, such as quiet compositions, soft light, or textured illustration.
The selection process teaches an aesthetic preference. Your selection note helps you remember the intention without pretending that the profile is a written specification. For the generation entry point and account requirements, use Midjourney’s Getting Started Guide (opens in a new tab), which documents Create, the Imagine bar, and the subscription requirement.
How do you apply the profile to a prompt?
Use the Personalization button beside the Imagine bar to turn personalization on or off, or add --p to a prompt. Use the selector to choose the intended profile. For additional profiles, the documented creation control is New Profile. Keep separate purposes clearly named in your own notes so you do not confuse a personal favorite with a work-approved look.
Test the same subject with personalization off and on. This is a review method, not a promise that the outputs will differ in only one visual detail. It gives you a useful comparison: did the profile move the result closer to the intended appearance, and did any important image requirement become weaker?
Purpose: calm illustrations for onboarding Prompt: a small desk lamp beside a closed notebook Model version: record the version used Profile: record the selected profile ID Accepted code: copy the code from the accepted prompt Review: quiet composition works; check object clarity next Next test: a different subject using the accepted code
Keep the test prompt simple enough that someone else can understand what you were checking. Our guide to saving AI prompts covers the longer-term record, while AI for graphic designers addresses the broader review workflow.
Should you save the profile ID or the generated code?
Save both when you need both behaviors. A profile ID follows the latest state as you keep selecting and liking images. A historical code selects an older state. That difference matters when an approved look must remain available after you continue developing your taste profile.
| Keep this | Use it when | What to write beside it |
|---|---|---|
| Profile ID | You want the profile’s latest state | Purpose and date of your latest review |
| Historical code | You want to revisit an accepted state | Accepted output and the reason it passed |
| Model version | You need to understand the generation setup | The version actually used in the test |
| Prompt and review note | You want another person to continue | What worked, what failed, and the next test |
Midjourney documents /list_personalize_codes in Discord for retrieving old codes. Existing codes from a deleted additional profile remain usable, but that should not replace a deliberate record of the code you accepted. Copy it when you approve the output, while the reason for choosing it is still clear.
For a team, write down who approved the use of a particular look and where it should apply. Keeping that choice beside the test in fenbs gives the next assistant a reason to reuse the code instead of choosing a new favorite.
How do model versions affect reuse?
The Personalization guide documents support for versions 6 and later. Its Global V7 Profile also works with V8.1 and V8.2, while V8 profiles do not work with V7. Record the version alongside the profile; a profile name alone is not enough to explain a successful generation.
When you change versions, run the short review again before treating the look as approved. Keep the old result for comparison and describe any difference that matters to the actual work. “Background became busier and crowded the heading” is a useful finding. “New version feels different” gives the next person much less to act on.
Avoid making an unchanged output the acceptance test unless that is truly required. A practical review can instead ask whether color, spacing, medium, and subject clarity still meet your requirements. Separate a successful aesthetic from the correctness of a face, logo, or product detail.
When is a moodboard a better starting point?
A preference profile is built by selecting images you like. A moodboard instead starts from an explicit collection of source images, as Midjourney explains in its Moodboards guide (opens in a new tab). Choose based on what you are trying to communicate: broad taste or a particular reference collection.
For a collection built around specific source images, follow creating a reusable Midjourney moodboard. Keep that separate from training the profile in this guide. Liking numerical style codes is also different: Midjourney’s Style Reference guide (opens in a new tab) says those likes do not train Personalization.
That distinction avoids a common recordkeeping mistake: saving a page of favorite looks and assuming you have documented the profile that produced an accepted image. Keep the actual code, model, and result. The prompts topic hub has more guidance on turning one good result into a reusable recipe.
Keep the prompts that work
Save your best prompts in fenbs and use them in any AI, on any device. No more digging through old chats.
Create a free accountFree for up to 5 people and 50 open tasks · No credit card · Works with Claude, ChatGPT, Cursor and more
How to keep an accepted profile recipe with fenbs
Keep the recipe so you can carry on from exactly where you left off, on any device and with another AI. You should be able to find the approved look without remembering which evening you generated it.
- Create a free account.
- Save the accepted profile code, its model version, the test prompt, and a link to the result.
- Write what the look is for and who approved it.
- Before you stop, write the next test and anything that still needs a person to check.
fenbs does not train Midjourney profiles or apply them to an image. A text file is enough for a one-time personal test. fenbs helps when the accepted recipe and the next step need to remain clear across people, sessions, or connected AI apps. For a larger collection, use prompt library management to plan what to keep current.