Genvizu

AI creations made with Midjourney 8.2

Midjourney

Midjourney 8.2 has been Midjourney's default version since 24 July 2026. On Genvizu one run returns four images, at 1K or 2K — there is no 4K tier, because the upstream resolution flag is a boolean. Prompts must be written in English and are capped at 1,024 characters, trimmed at 1,023 on a word boundary. Parameters such as --ar, --stylize, --chaos, --raw, --iw and --sref are read from the prompt; resolution and version come from the panel, not from flags.

Verified on 2026-09-10

Start free · no credit card

Best for

  • Art-directed stills where the default look matters: editorial covers, campaign keyframes, concept art, illustration series.
  • Getting four comparable options from a single prompt, since every run returns four images from one submission.
  • Style continuity across a set — one --sref code can be reused across prompts, and Midjourney parses it natively rather than through prompt text.
  • Iterating on one variable at a time: --stylize sets how far the default aesthetic is applied, --chaos sets how much the four results differ, and --raw reduces that default styling.

Not for

  • Work you have to clear for commercial use without checking rights. Midjourney output is commercially restricted by Midjourney’s own terms, so confirm your licence before publishing.
  • Long, exhaustive briefs. The prompt budget is 1,024 characters and longer text is trimmed at 1,023 characters on a word boundary, so the tail is what gets lost.
  • Prompts you want to keep in Chinese. The Midjourney models on Genvizu require English input.
  • Reproducing an exact person, product or previous composition. --sref transfers style, not identity, and --seed is not sent as an API parameter.
  • 4K delivery. The upstream flag is a boolean that maps to a native 2K output; there is no 4K tier.

Specifications and limits

FieldValue
ProviderMidjourney (routed through Genvizu)
VersionV8.2, the Midjourney default since 24 July 2026; V8.1 remains available separately
InputEnglish text prompt; optional single reference image, plus one style reference (--sref)
Output4 images per run
Resolutions1K / 2K (1K is the default; there is no 4K tier)
Aspect ratios1:1, 4:3, 3:2, 16:9, 3:4, 2:3, 9:16
Prompt languageEnglish required
Prompt limit1,024 characters; longer prompts are trimmed at 1,023 on a word boundary
Parameters read from the prompt--ar, --stylize / --s, --chaos / --c, --quality / --q (1 or 4), --raw, --iw (0-3), --sref, --sw
Parameters ignored--hd is stripped so billing matches the panel resolution; --v is unused because the version follows the model you pick
BillingPer run of four images; the 2K tier costs more than 1K (see the pricing page)
Plan requirementAvailable on every plan, subject to your balance
Failed runsRefunded automatically

Prompt examples

Editorial architecture still

Prompt

solitary coastal hotel on a Mediterranean cliff,
editorial architectural photography, wide establishing view,
late-afternoon side light, chalk white walls and faded cobalt blue,
quiet summer atmosphere, restrained textures --ar 16:9 --stylize 150

Subject, medium, viewpoint, light, palette, mood — one clause each, parameters last. This is the baseline you change one variable at a time.

Style reference carrying a series

Prompt

Airy Hazy Blue --sref 4250870612
Soft, delicate illustration style bathed in faded sky blues and muted whites.
Gentle minimalist aesthetic, soft edges, nostalgic and tranquil mood.

Text defines content and mood, --sref supplies the visual language. Paste several --sref values and only the first survives, so keep one code per series.

Product-safe brand frame

Prompt

flagship sneaker on a poured concrete plinth, premium editorial campaign,
symmetrical centred composition, deep burgundy with warm sand accent,
hard key light from the upper left creating a sharp diagonal shadow,
ample negative space on the right for layout, no readable text --ar 3:2 --raw

--raw reduces the default Midjourney styling, so the render stays closer to what the text says. Asking for no readable text keeps the space clean for real typography added later.

How it compares

Midjourney 8.1

Same parameter surface, same four images per run and the same 1K / 2K tiers, but the older rendered look. Pick it to keep an existing set visually consistent instead of re-baselining --stylize on 8.2.

GPT Image 2

Pick it when the image must contain correct typography, when the brief is long, when you need 4K, or when you would rather not write in English.

Grok Image 2

Text-to-image only and single-tier, with no English requirement. Use it to explore a poster idea in your own language before committing to a Midjourney run.

Examples from the gallery

Common failures and fixes

The end of a long prompt seems to be ignored.

The budget is 1,024 characters and the tail is what gets cut. Delete adjectives that do not change your acceptance criteria, and move parameters to the end where they belong.

Style codes that worked on 8.1 produce a different look.

The parameter surface is unchanged but the rendered look shifted. Re-baseline --stylize and re-test each --sref code before regenerating a whole set.

A style reference does not preserve a face or a product shape.

It never did. State identity, quantity and composition in the text, and treat --sref as a palette and texture instruction only.

A --hd or --v flag in the prompt has no effect.

Both are removed on purpose. Resolution comes from the panel selector so billing matches the output, and the version comes from choosing Midjourney 8.2 or 8.1 in the model picker.

A Chinese prompt returns something unrelated.

These models require English. Draft in your own language, then convert with the Midjourney prompt helper and check counts, directions and colours by hand.

Frequently asked questions

How many images does one Midjourney 8.2 run return, and how is it billed?

Four images per run, and the run is billed once for all four rather than per image. Resolution is 1K or 2K, with the 2K tier costing more; the current tiers are on the pricing page. There is no 4K tier, because the upstream resolution flag is a boolean that maps to a native 2K output.

Do I need to write --v 8.2 in the prompt?

No. The version comes from the model you select in the picker, and a --v flag would only consume part of the 1,024-character budget.

Which Midjourney parameters does Genvizu actually send?

--ar, --stylize, --chaos, --quality (1 or 4, defaulting to 1), --raw and --iw are read from the prompt; --sref and --sw stay in the prompt so Midjourney parses style codes natively, with duplicates stripped. --hd and --v are removed, and --seed is not extracted as an API parameter.

Can I use Midjourney 8.2 images commercially?

Midjourney output is commercially restricted by Midjourney’s own terms. Confirm your rights with Midjourney before commercial use; Genvizu runs the generation but does not grant that licence.

Is Midjourney 8.2 limited to a paid plan?

No. Both 8.1 and 8.2 are available on every plan, as long as your balance covers the run.

Sources