RETROSPECTIVE RECORD · PREPARED 16 SEPTEMBER 2026The library · 100 retrospective records ↗
Screen Visual Lab

The library / Tools & pipelines

Tools & pipelines / Reference entry · Entry note · prepared 16 September 2026

Midjourney documents its style parameter as palette, not people

The company's own help article says style reference carries colors, medium and lighting, not objects or people, from an image.

Visual for this record: Midjourney documents its style parameter as palette, not people
Visual published by docs.midjourney.com, shown for identification of the record. Credit: docs.midjourney.com · source page ↗ Rights: owner-review-pending.

The image

Midjourney's own Style Reference help article, last updated 1 September 2026 at the time of this reading, is the document at hand rather than any single generated picture. It describes a feature reached by clicking the image icon in the Imagine bar or by appending the parameter "--sref" followed by an image URL in Discord, and it shows example pairings of a reference image beside new generations that carry over its palette or texture. As a living help page, its content reflects the product on the date retrieved, 16 September 2026, and can change again without a version number attached.

What the documents show

The article states plainly what the feature does and does not do: "It doesn't copy objects or people, just the overall style — like colors, medium, textures, or lighting." It documents a style-weight parameter, "--sw," ranging from 0 to 1000 with a default of 100, and a separate "Style Reference Code" system that applies a numbered look from what the article calls "the Midjourney internal style library" instead of an uploaded image. A second document, Midjourney's own Legacy Features page, corroborates the lineage: it records that a Discord-based "Style Tuner," used in version 5.2, was "one of the first steps" toward the personalization and style-code system the current article describes, and that the Style Tuner itself no longer accepts new submissions.

Production context

Read editorially, a documented style weight that a user can dial between 0 and 1000 offers a mood-board tool a numeric, repeatable setting rather than a vague verbal instruction, useful when a team needs to compare several strengths of the same reference against the same prompt. The article's own best-practice guidance — to keep text prompts focused on content rather than instructions like "copy this style" — indicates Midjourney documents the feature as sensitive to how it is prompted, not as a one-step transfer.

Reference versus imitation

The documentation is explicit that objects and people are not copied, only "overall style," which sets a documented boundary between reference and imitation for this specific parameter. A style-reference code, since it is drawn from an internal library rather than an uploaded image, carries no direct link to any external artist's work by the article's own account; an uploaded reference image, by contrast, is a specific file whose provenance and licensing remain the uploader's responsibility, a distinction the help article does not itself adjudicate.

  • Was the style conditioned from an uploaded image or from a numbered internal style code?
  • What style-weight value was used, and is it recorded alongside the output?
  • Does the reference image's own license or source permit this use, a question the tool's documentation does not answer?

A help-center article is a snapshot of a live product; treating today's description of a parameter as permanent, or as evidence about any one artist's style being reproduced, goes beyond what the document itself claims.

Sources & reading trail

Style Reference ↗

States that style reference conditions color, medium, texture and lighting rather than objects or people, and documents the --sref and --sw parameters.

Source published: Not established · Retrieved: 16 September 2026

Legacy Features ↗

Records that the version 5.2 Style Tuner preceded the current personalization and style-code system, corroborating the feature's documented lineage.

Source published: Not established · Retrieved: 16 September 2026

Records, documentation and rulings establish the entry; the reference-versus-imitation reading is Screen Visual Lab editorial analysis. This retrospective draft does not imply the site published on the event date.