Prompts
Midjourney V7 wants a sentence, not a comma-separated keyword pile
Prompting advice for Midjourney V4 or V5 — stack keywords, separate with commas — actively hurts results on V7, which reads a genuine natural-language sentence far better than a keyword pile.
Last updated Aug 15 · 12 min read
Why V7 broke the old keyword-stacking advice
Older Midjourney prompting guidance recommended stacking descriptive keywords separated by commas — a habit from earlier model versions that read prompts more literally. Midjourney V7 understands a genuine natural-language sentence significantly better, and following the old keyword-stack advice on the current model actively produces worse results than just describing the image in a real sentence. The Midjourney prompt library is written for V7 specifically.
The foundational prompt, a natural-language portrait brief for V7, demonstrates the shift directly — a full descriptive sentence rather than a stacked keyword list.
Parameters that still matter: stylize, chaos, and aspect ratio
--stylize controls how much Midjourney's own aesthetic bias applies versus following the prompt literally, and --chaos controls variation between the four initial grid results. Exploring concept art with those two dialled deliberately — this prompt — treats them as genuine creative controls, not settings to leave at default and forget.
Consistency: character, style, and product references
Keeping a character consistent across a comic — this prompt — brand campaign visuals consistent via style reference — covered here — and a product consistent across multiple scenes via Omni Reference — this prompt — are three distinct reference mechanisms solving three distinct consistency problems.
Editing an existing generation: vary region, pan, and zoom out
Targeted region edits on a generated image — this prompt — fix one specific part without regenerating the whole composition. Panning and extending a scene — covered here — and zooming out to reveal more environment — this prompt — both extend an existing composition rather than starting fresh.
Clean product photography and negative prompting
Clean, distraction-free product photography — this negative-prompt-driven prompt — uses Midjourney's negative-prompt syntax to explicitly exclude common unwanted artifacts, complementing what the positive prompt already describes.
Batches, patterns, and icon sets
A permutation prompt generating a whole batch of icon variants at once — this prompt — beats generating each icon individually. Seamless tile pattern design — covered here — needs specific tiling-aware phrasing most single-image prompts skip, and locking a seed for consistent icon sets — this prompt — keeps a whole set visually unified.
Text-free logo concepts and image-to-video
Midjourney remains unreliable at rendering legible text — a text-free logo mark concept prompt works around that limitation by designing a mark that needs no text at all. And animating a still image into motion — this prompt — extends a generated still into Midjourney's newer video capability.
When generated visuals need to become a real brand system
These prompts produce excellent individual images. Turning consistent visual output into an actual, coherent brand identity across every touchpoint is bigger design work. That's exactly what Scult's branding team does, or book a meeting to talk through your visual identity.
Need this built into your business?
The free tools and prompts on this site handle the small, solved problems. If what you need is bigger — branding & design, built and maintained for you — that's Scult's day job.
