AI Image & Design ToolsOpenArt AIOpenArt AIPhotoroomPhotoroomLetsEnhanceLetsEnhanceMagnificMagnificIdeogramIdeogramKittlKittlKrea AIKrea AIPhot.AIPhot.AICHATGPTCHATGPTStable DiffusionStable DiffusionLeonardo AILeonardo AICanva AICanva AIAdobe ExpressAdobe ExpressFotorFotorTopaz Photo AITopaz Photo AINightCafeNightCafeMidjourney AIMidjourney AI
Midjourney - AI image and video generator with Style References, Personalization, and the Edit Model

Midjourney

AI Image & Video Generator — Style References, Personalization, and Consistent Characters

What is Midjourney?

Midjourney is an AI image and video generator. You type a text prompt, and it returns a set of images built from that description. It started as a Discord bot in 2022 and has since added a full web interface, video generation, and tools for keeping a consistent visual style across many images.

Midjourney Today

Midjourney started as a Discord bot in 2022. Since then it's added a full web interface, video generation, and tools for keeping a consistent visual style across many images — Style References, Personalization, and the Edit Model.

Midjourney's current default model is V8.2 (released July 24, 2026). V8.2 replaced three older, separate tools — Omni Reference, Character Reference, and Retexture — with a single Edit Model. If you've read older Midjourney guides that describe those three as separate features, that's now out of date.

Midjourney Pricing

Midjourney is subscription-only. There's no free tier and no free trial — you need a paid plan to generate anything.

Basic
$10/mo

$8/mo billed annually ($96/yr).

  • 3.3 Fast GPU hours/month
  • No unlimited Relax mode
Pro
$60/mo

$48/mo billed annually ($576/yr).

  • 30 Fast GPU hours/month
  • Stealth mode (private generations)
Mega
$120/mo

$96/mo billed annually ($1,152/yr).

  • 60 Fast GPU hours/month
  • Stealth mode (private generations)

A few things worth knowing before picking a plan:

  • You're buying GPU time, not a fixed number of images. A basic image prompt uses about 1 minute of GPU time. Video uses much more: a batch of 4 standard-definition videos costs 8 minutes, and a batch of 4 high-definition videos costs 26 minutes.
  • Unused Fast GPU hours don't roll over — they reset every billing cycle, on both monthly and annual plans.
  • Basic is enough for occasional use. Standard gives you significantly more Fast GPU time and adds unlimited Relax mode for images, which queues generations instead of running them instantly — slower, but it doesn't use your Fast hours.
  • Stealth mode (keeping your generations private) is only available on Pro and Mega.
  • Annual billing works out to a 20% discount over paying monthly.

How It Works

You write a prompt describing what you want, and Midjourney generates four image variations. From there you can upscale one, generate more variations of it, or edit it further. The same idea extends to video: Midjourney turns a still image into a 5-second video, extendable up to 21 seconds.

Two features become especially useful once you start doing more than basic prompts:

Style References (--sref)

Upload a reference image and Midjourney pulls its color palette, lighting, and overall look, then applies that look to a new subject you describe.

Personalization (--p)

Midjourney builds a profile of what you tend to like, based on images you've ranked, and leans future generations toward that taste. Its results become more useful after you've ranked enough images to build a preference profile.

Edit Model (--edit)

As of V8.2, this replaced the older separate Omni Reference, Character Reference, and Retexture tools. It uses written instructions plus up to four reference images, and it can help keep a character, product, or other subject consistent across multiple generations, or restyle an existing image while keeping its composition.

Photorealism

Midjourney is widely used for photorealistic-looking images, and its current models can produce detailed lighting, skin texture, and material rendering. Results still vary by prompt, and complex hands and poses can still produce errors — "closely matches real-world photography" is a description of typical results, not a guarantee on every generation.

The Interface

Midjourney has a full web interface at midjourney.com, separate from the original Discord bot. It includes a dashboard, an Explore feed (public community generations, searchable — you can see the exact prompt and parameters behind any public image), a Create tab, an archive of your past generations, and a canvas-style editor.

Discord is still supported for people who prefer it, but it's no longer required — the web app covers full generation and editing. Draft Mode trades image quality for speed, useful for testing prompt directions quickly before committing GPU time to a full-quality render.

Moodboards

Moodboards let you collect reference images to define a visual direction — color palette, lighting, composition — before generating anything. Useful for teams aligning on a look before production starts, rather than iterating on individual prompts one at a time.

Prompt Parameters

ParameterWhat it does
--arSets aspect ratio (e.g. --ar 16:9)
--vSets model version (e.g. --v 8.2)
--qSets render quality/GPU cost
--s or --stylizeControls how much Midjourney's default aesthetic is applied
--iwSets how strongly an image prompt influences the result
--noExcludes elements from a generation
--chaosIncreases variation between the four results
--seedLocks a specific starting point for reproducible results
--srefApplies a Style Reference image
--pApplies your Personalization profile
--editEdit Model — modifies an image using written instructions plus up to 4 reference images
--style rawReduces Midjourney's default stylization, follows the prompt more literally
--weirdPushes results toward more unconventional output
  • Raw mode removes Midjourney's default aesthetic styling so results follow your prompt more literally.
  • Describe works in reverse: upload an image, and Midjourney suggests prompts that could recreate a similar look.
  • Vary (Subtle/Strong) generates controlled variations of an existing image — small tweaks (Subtle) or bigger reinterpretations (Strong).
  • Seed locks the random starting point of a generation, so re-running the same prompt with the same seed gives you a near-identical result.

Midjourney vs. DALL-E 3

FeatureMidjourneyDALL-E 3
Photorealistic imagesPhotorealistic image generationSupported
Text in imagesCan require several attemptsGenerally good
Style controlStyle References, Moodboards, PersonalizationDifferent, prompt-driven workflow
Character/subject consistencyEdit Model with reference imagesImage editing/reference workflow
EditingEdit ModelBuilt-in image editing
Learning curveMore product-specific termsFamiliar chat interface

Midjourney vs. Other Image Generators

FeatureMidjourneyChatGPT ImagesAdobe FireflyIdeogram
Photorealistic imagesPhotorealistic image generationSupportedSupportedSupported
Exact textCan require several attemptsGenerally goodSupportedStrong focus
EditingEdit ModelConversational editingBuilt-in editing toolsEditing tools
Text/typographyCan struggle with exact textHandles textHandles textDesigned for text-heavy images
Learning curveMore product-specific termsChat-basedFamiliar editor workflowFamiliar interface

Who It's For

Midjourney is useful for people creating visual concepts and working with a consistent style:

🎨
Designers

Working on visual direction for a project.

📣
Creative agencies

Creating campaign concepts before committing a production budget.

🛒
Brands

Generating product images in different settings without new photography.

🎬
Filmmakers & directors

Building moodboards and lighting references before a shoot.

🚀
Marketers

Who need campaign images without arranging a new photo shoot.

Who Should Look Elsewhere

  • Want an interface that isn't a separate ecosystem to learn — try something with a simpler, more conventional editor.
  • Need pixel-accurate text or logos in your output — Midjourney is weaker here than tools built for typography, like Ideogram.
  • Need a free tier to try before paying — Midjourney has none.
  • Want a traditional photo editor with AI features added on, not an AI-first generator — Adobe Firefly fits that better.

Pros & Cons (from G2)

G2 currently lists Midjourney at 4.4/5 from 102 reviews.

What users like
  • Realistic imagery (tagged in 5 reviews)
  • Ease of use (5)
  • Time-saving (4)
  • Helps with idea generation (4)
  • Creativity/output quality (3)
What users don't like
  • AI limitations in output quality and detail accuracy (3)
  • Prompt issues — getting exact details right often takes several tries (3)
  • Image errors — some outputs don't follow instructions closely (2)
  • Learning curve for new users (2)
  • Inconsistent output/design quality (2)

The written reviews also mention a learning curve, confusion around the Discord-based workflow, the cost of higher-tier plans, and the lack of traditional manual editing tools.

Source: Midjourney reviews on G2.

Version History

VersionReleasedNotes
V6Dec 2023Better prompt accuracy on longer prompts, improved image prompting
V6.1Jul 2024~25% faster renders, more coherent detail
V7Apr 2025Introduced Draft Mode and Omni Reference
V8.1Apr 20264–5x faster rendering, added 2K "HD" still images
V8.2 (current)Jul 2026Replaced Omni Reference/Character Reference/Retexture with a single Edit Model; improved Personalization

Midjourney also runs Niji, a separate model line built with Spellbrush for anime and Eastern art styles, currently on Niji 7 (Jan 2026).

Ownership, Commercial Use & Privacy

Midjourney's Terms generally give subscribers ownership of content they create, subject to a couple of specific exceptions:

  • If your company makes more than $1,000,000 USD a year in revenue, you (or your employer) need a Pro or Mega plan to own what you create.
  • If you upscale someone else's image, that image stays owned by its original creator.

By default, everything you generate is public and can be remixed by other users — Midjourney's Terms grant Midjourney a broad license to display and reuse what you create. Stealth mode (Pro and Mega only) keeps your generations private and out of the public feed. Even with Stealth mode on, anything you generate in a shared space like a Discord server is visible to everyone in that server.

FAQ

Is Midjourney good for photorealism?

Yes. Midjourney can produce detailed, photorealistic-looking images, although results still vary by prompt.

What's different about Midjourney compared to other AI image tools?

One notable difference is Personalization: Midjourney builds a profile of your preferences over time and can use it to influence future generations.

Can Midjourney keep a character or product consistent across multiple images?

Yes. The V8.2 Edit Model (--edit) can use up to four reference images to help keep a character, product, or other subject consistent across generations.

Can I use Midjourney images commercially?

Generally yes — you own what you create, subject to the ownership exceptions above (the $1M-revenue/Pro-or-Mega rule in particular). Check Midjourney's current Terms of Service before using images in client or commercial work.

Is Midjourney available outside Discord?

Yes. The web app at midjourney.com covers generation, editing, and your history — Discord is optional now, not required.

What are Moodboards?

A way to set visual direction — palette, lighting, composition — by collecting reference images before you start generating.

About this review

We checked Midjourney's own documentation (subscription plans, parameter list, video generation costs, and Terms of Service) and its G2 reviews page when preparing this review. We update this page as Midjourney's pricing, models, or features change — its V8.2 update in July 2026 changed how several older reference tools work, so we've flagged that clearly above.

This page contains affiliate links: if you sign up through them, we may earn a commission at no extra cost to you — which never affects what we recommend.

The Bottom Line

Midjourney is a paid-only image and video generator with strong control over visual style and reference-based generation. The main trade-offs are the subscription cost, a real learning curve, and limits around exact text and manual image editing. Its current V8.2 model also changed how some older reference features work — Omni Reference and Character Reference are gone, folded into the Edit Model — so older Midjourney guides describing those as separate tools are out of date.

Try Midjourney for yourself

You can explore Midjourney's current plans and see whether Basic or Standard fits your needs.

Affiliate link. We may earn a commission at no extra cost to you. Our review is always independent.

Try Midjourney →
Back to Top