Midjourney now lets beginners create and refine AI images directly on its website, so you do not need to begin in a public Discord channel. The basic workflow is simple: subscribe, open the Create page, describe the image, compare the results, refine the strongest option, and download the finished asset.
This current Midjourney tutorial covers that workflow with the default V8.2 model, a practical prompt formula, the controls that matter most, and the quality and rights checks to complete before publishing an image.
The original video above was published in 2023 and demonstrates an earlier Midjourney workflow. Its “free generator” wording is no longer current: Midjourney’s official Discord quick-start says a subscription is required before generating. The source post did not contain verified chapters or a transcript, so this update preserves the video without inventing timestamp links.
Midjourney Tutorial: The Current Workflow
- Create or sign in to a Midjourney account.
- Choose a subscription that fits your expected usage.
- Open the Create page on midjourney.com.
- Enter a focused text prompt in the Imagine bar.
- Compare the generated images for prompt adherence and structure.
- Create variations or change the prompt.
- Upscale, edit, pan, zoom, or remix when needed.
- Inspect the final image and download it.
The web interface is the clearest starting point for most new users. Discord remains supported for people who prefer bot commands or already use Midjourney there.
Step 1: Create an Account and Review the Plans
Start with Midjourney’s official getting-started guide. At this update, Midjourney lists four subscription tiers: Basic, Standard, Pro, and Mega. Pricing, GPU allocation, video access, concurrency, privacy, and plan features can change, so use the live plan comparison before purchasing.
Choose a plan from your actual workflow:
- How many images and videos do you expect to generate?
- Do you need Relax Mode for additional image generation?
- Will you work with confidential concepts that require Stealth Mode?
- Are you generating as part of a company with more than $1 million in annual gross revenue?
- How much Fast GPU time and concurrency do you need?
Do not choose a plan from an old tutorial’s price or feature list.
Step 2: Open the Create Page
Sign in to midjourney.com and open Create. According to Midjourney’s current web guide, the Imagine bar accepts prompts, provides image-upload and settings controls, and displays completed jobs in the creation feed.
Use the settings button to review the active model, image size, aspect ratio, stylization, variety, speed, and other available options before generating. Defaults may change when Midjourney releases a new model.
Step 3: Confirm the Current Model
Midjourney’s official version documentation lists V8.2 as the current default. It became the default on July 24, 2026. You can select a default model in settings or use the version parameter where supported.
An old prompt written for V4, V5, V6, or V7 may behave differently in V8.2. Some parameters and reference systems are version-specific. Start with a clean prompt, confirm feature compatibility, and add older parameters only after checking the current documentation.
Step 4: Write a Strong Midjourney Prompt
A useful prompt communicates the subject and visual decisions without becoming a random keyword list. Use this structure:
[Main subject], [action or pose], in [setting], [composition and camera], [lighting], [visual medium or treatment], [important colors or materials], [layout requirement] –ar [ratio]
Example: product photograph
Matte-black wireless headphones on a pale stone pedestal, three-quarter product view, soft gray studio background, broad diffused key light with a narrow rim light, premium commercial photograph, accurate material texture, clean negative space on the right –ar 16:9
Example: editorial illustration
Small community restoring an urban rooftop garden after a storm, wide elevated composition, early morning haze, layered editorial illustration, restrained green and terracotta palette, clear human silhouettes, room at the top for an editable headline –ar 3:2
Example: original character concept
Original solar-station engineer, cropped dark hair, amber visor, white utility jacket with cobalt fasteners, neutral three-quarter pose, full-body character design on a plain warm-gray backdrop, even studio lighting, detailed garment construction, no logo, no text –ar 2:3
Prompt-writing rules that improve iteration
- Lead with the main subject.
- Describe one understandable action.
- Specify close-up, portrait, full body, wide shot, top-down, or another clear composition.
- Name a lighting condition rather than adding “beautiful lighting.”
- Use a small, intentional style vocabulary.
- Add the output layout or aspect ratio required by the final platform.
- Change one major variable at a time after the first generation.
Negative instructions are less useful than clearly describing what should appear. When exact text matters, add the final copy in a design editor. Midjourney’s current text-generation guidance recommends putting a short phrase in double quotation marks, but generated lettering still needs proofreading.
Step 5: Choose the Right Aspect Ratio
Aspect ratio controls the shape of the generated frame, not its final print quality. Use the destination to choose it:
| Destination | Common starting ratio | Composition note |
|---|---|---|
| Square social post | 1:1 | Keep the focal point near the center |
| Vertical post or character | 4:5 or 2:3 | Leave room above and below the subject |
| Story or short-form cover | 9:16 | Protect interface-safe areas |
| YouTube thumbnail | 16:9 | Reserve negative space for editable copy |
| Landscape illustration | 3:2 or 16:9 | Define foreground, subject, and background |
Midjourney’s version page documents model-specific maximum ratios and HD limitations. Check compatibility before using an extreme ratio.
Step 6: Compare the First Results
Evaluate the outputs against the prompt rather than choosing the most visually busy image. Check:
- Is the main subject immediately clear?
- Does the composition fit the intended crop?
- Are the pose and camera angle believable?
- Do materials and object geometry make sense?
- Are hands, faces, text, and fine edges usable?
- Does the image contain accidental logos, signatures, or protected content?
If none of the results solves the composition, revise the prompt. Generating many variations of a structurally weak image usually preserves the same problem.
Step 7: Use Variations, Remix, and Upscale
Midjourney’s current creation controls support several ways to continue:
- Subtle variation: keep the main direction and explore smaller changes.
- Strong variation: create a broader interpretation.
- Remix: change the prompt or parameters while creating a variation.
- Subtle upscale: enlarge while attempting to preserve the source.
- Creative upscale: enlarge while introducing additional interpretation.
Always inspect the upscaled result. More pixels do not guarantee more accurate anatomy, typography, texture, or product geometry.
Step 8: Edit, Pan, and Zoom
The Midjourney web Editor combines tools for expanding and changing an image. Current documentation describes:
- Vary Region or Erase: regenerate a selected area.
- Pan: extend the canvas in a direction.
- Zoom Out: add context around the original frame.
- Remix: steer the revised generation with changed instructions.
Use local editing for a defined problem. Select enough surrounding context for the model to blend the replacement, but avoid erasing the entire image when only one object needs revision.
Step 9: Use Images as References Carefully
Midjourney provides multiple reference workflows, and they are not interchangeable:
| Reference type | Purpose |
|---|---|
| Image Prompt | Influence content, composition, colors, or general direction |
| Style Reference | Carry visual qualities such as color, medium, texture, or lighting |
| Omni Reference | Use a person or object reference in compatible models |
| Moodboard | Build broader personalized aesthetic direction from selected images |
Only upload material you have the right to use. A reference feature does not grant permission to copy a person’s likeness, a copyrighted character, a product design, or someone else’s artwork.
Step 10: Download and Run Final Quality Control
- Open the final image at full size.
- Check anatomy, perspective, geometry, reflections, shadows, and repeated patterns.
- Proofread every visible character of generated text.
- Remove accidental trademarks, signatures, and watermarks.
- Confirm the pixel dimensions required by the publishing platform.
- Add exact copy, logos, and brand elements in a controlled design tool.
- Save the prompt, settings, model version, source references, and editing record.
For print or high-stakes commercial work, order a proof or perform a destination-specific test instead of relying only on the browser preview.
Website vs. Discord
| Midjourney website | Discord workflow |
|---|---|
| Visual Create feed and settings panel | Command-based interaction with the Midjourney Bot |
| Convenient image organization and web editing | Useful for users already comfortable with Discord |
| Imagine bar accepts prompts and image inputs | /imagine accepts a prompt in supported channels or servers |
| Best default starting point for many beginners | Still supports command parameters and bot workflows |
The result quality depends on the model and settings, not on treating Discord as inherently more advanced.
Commercial Use and Privacy
Midjourney’s current terms say users own assets they create to the fullest extent possible under applicable law, subject to the agreement and third-party rights. They also state that a company—or an employee creating for a company—with more than $1 million in annual gross revenue must use a Pro or Mega subscription for the stated ownership terms.
Default creations may be publicly viewable and remixable. Midjourney documents Stealth Mode for Pro and Mega plans, while warning that content generated in shared or open spaces remains visible there. Review the current terms before uploading confidential references or using outputs commercially.
You remain responsible for permissions, privacy, publicity rights, trademarks, copyright, and the receiving platform’s rules. AI generation is not a legal clearance process.
Common Midjourney Beginner Mistakes
- Following an old free-trial or Discord-only tutorial.
- Using legacy parameters without checking V8.2 compatibility.
- Writing a keyword pile with conflicting scenes and camera angles.
- Changing the entire prompt after every result.
- Choosing detail over composition and prompt adherence.
- Trusting generated words without proofreading.
- Upscaling without checking newly invented details.
- Uploading reference images without permission.
- Assuming private client work is hidden on every plan and in every space.
Frequently Asked Questions
Is Midjourney free?
Midjourney’s current official quick-start requires a subscription before generating. Plan prices and features can change, so check the live plan page.
What is the current Midjourney model?
The official version page lists V8.2 as the current default at this update. You can choose compatible versions through settings or a version parameter.
Do I need Discord to use Midjourney?
No. You can create images on midjourney.com. Discord remains an optional supported workflow.
What should my first prompt include?
Start with the subject, one action, setting, composition, lighting, visual treatment, and target aspect ratio. Add only details you can evaluate.
Can Midjourney generate consistent characters?
Reference tools and stable prompt anchors can improve continuity, but model and feature compatibility differs by version and no workflow guarantees perfect identity. Compare every output with an approved reference.
Final Takeaway
The best beginner workflow is to create on the Midjourney website, confirm the current model and settings, write one focused prompt, and improve the result through controlled variations and edits. Finish with a full-size quality check and a rights review.
Use the original video for historical context, but rely on current Midjourney documentation for subscriptions, V8.2 features, reference compatibility, privacy, and commercial-use terms.