The most common Midjourney mistakes in commercial work are vague prompts that produce generic images, inconsistent style across a campaign, relying on the model to render text and logos, ignoring licensing and likeness rules and skipping the design pass that turns a generated image into a finished asset. Design teams avoid them with reference systems, style controls, clear usage rules and a fixed finishing workflow.
What is Midjourney good at, and what is it not?
Midjourney is an image generation service run by the independent research lab of the same name. It is known for strong aesthetics, rich lighting and painterly or photographic styles from short prompts. Users work through its web app or Discord, and the service offers tools such as style references, character references, variations, upscaling, inpainting through its editor and video generation from images at the time of writing.
It is weaker at exact control. Precise product shapes, readable text, accurate hands in complex poses and brand logos still need care or another tool. Knowing this up front avoids most of the mistakes below.
Which mistakes come up most often?
The list below is ordered roughly by how often each problem appears in real projects, starting with the ones that cost the most to fix after publication.
Mistake 1: prompting like a search engine
A prompt such as “modern office marketing team” returns polished but generic stock-like images. They are easy to make and easy to forget.
Better prompts describe the scene like a photographer or illustrator would: subject, setting, lighting, lens or medium, mood and composition. They also include what makes the brand different, such as a colour palette or a recurring visual motif.
A quick prompt structure
- Subject and action: who or what, doing what.
- Setting: place, time of day, season.
- Visual language: photographic or illustrated, lens or medium, lighting.
- Brand cues: palette, texture, recurring elements.
- Parameters: aspect ratio for the target placement, stylisation level and references.
Mistake 2: a different style in every image
Campaigns fall apart when each image looks like it came from a different artist. This happens when every designer prompts from scratch.
Use style references to anchor the look and keep a shared library of reference images and the exact parameters that produced approved results. For recurring characters or mascots, character references help keep features consistent across scenes, though results still need checking. Some teams also train their eye by building a “do and don’t” board for each client.
Mistake 3: expecting clean text and logos
Midjourney has improved at rendering short words, but it remains unreliable for brand names, slogans, prices and logos. Small letter errors are common and easy to miss on a phone screen.
Generate the image without text and add typography and logos in Figma, Photoshop or Canva. This also keeps the brand’s actual typeface, which a model cannot reproduce exactly.
Mistake 4: ignoring rights and likeness
Paid Midjourney plans grant broad usage rights to the images users create, subject to the service’s terms, and large companies above a revenue threshold need a higher tier. Terms change, so read the current version before a commercial project. Copyright protection for purely AI-generated images is limited in several jurisdictions, which matters if a client wants exclusive ownership of a visual identity.
Likeness is a separate issue. Prompting for a real, recognisable person or a living artist’s distinctive style creates legal and ethical risk. Avoid it in client work, and keep records of prompts in case questions come up later.
Mistake 5: publishing the raw output
A generated image is a starting point. Raw outputs often have small flaws: extra fingers, warped background objects, odd reflections or colours slightly off the brand palette.
A finishing checklist
- Zoom in on hands, faces, edges and background details.
- Fix flaws with inpainting in Midjourney’s editor or retouching in Photoshop.
- Colour-correct to the brand palette.
- Upscale to the resolution needed for the placement.
- Add typography, logos and legal lines in a design tool.
- Export in the right formats and sizes for each channel.
Mistake 6: wrong formats for the placement
An image that looks great as a square does not always crop well into a 9:16 story or a wide website banner. Set the aspect ratio in the prompt for each placement from the start, or plan a composition that leaves safe space for cropping.
How Alien Road applies Midjourney in graphic design work
A typical engagement begins with a visual discovery session. The team reviews the client’s brand guidelines, existing photography and the placements the campaign needs, and decides which visuals suit generation and which need a photoshoot or illustration.
Setup produces a reference kit per client: approved style reference images, a prompt template, parameter notes and a do-and-don’t board. Generated images go through the finishing checklist, and every final asset has its prompt and edits logged.
Measurement depends on use. For ad creative, AI-assisted variants are tested against existing creatives in Meta Ads Manager or Google Ads and compared on click-through and cost per result. For website imagery, the team tracks production time per asset and revision rounds. The report keeps creative performance and production efficiency in separate sections.
A short scenario: a café chain’s seasonal menu
Picture a café chain launching an autumn menu across social channels and in-store screens. The first batch of Midjourney images looks lovely but inconsistent: one set feels like film photography, another like a glossy magazine, a third like an illustration. Cups show garbled lettering where the logo should be.
The team picks the two strongest images as style references, rewrites the prompt with fixed lighting and lens notes, and leaves the cups blank so designers can add the real logo later. The second batch matches. Each placement gets its own aspect ratio, and the finishing checklist catches a warped spoon before anything goes live.
Where to start this week
Choose one upcoming visual and write a prompt using the five-part structure above. Generate a few options, pick the best and save it as a style reference. Then run the image through the finishing checklist before showing it to anyone. Keep the prompt, parameters and edits in a shared document; that record is the start of your reference kit.
Keep reading: Graphic Design · AI Marketing