Everything a new LLM artist needs to know, in plain English.
Why V7 Is Worth Your Time
Midjourney’s seventh model isn’t just a minor tweak—it changes how you prompt, how fast you iterate, and how much control you have over style and characters. Key additions like Draft Mode, Conversational Mode, an Omni Reference system, and extended parameters (
--q 4
,
--exp
) all aim to shrink the gap between imagination and final render.
Quick-Start Checklist
- Switch to V7 – Open
/settingsin Discord and pick Version 7, or add--v 7to any prompt. - Toggle Draft Mode for rapid prototyping.
- Use Conversational Mode when you’d rather let MJ craft the prompt from natural language.
- Reserve
--q 4and--expfor your final, polished pass. - For consistent characters or props, upload an image and add
--oref 1. - Finish with Upscale, Pan, or Zoom buttons under your chosen image.
Work Faster & Cheaper: Draft Mode + Conversational Mode
Draft Mode renders about 10× faster at half the GPU cost—perfect for testing composition or color palettes. Flip on Conversational Mode inside Draft to “talk” your way to a prompt; Midjourney reformats your thoughts into
/imagine
syntax while frames appear in real time.
Pro tip: Stay in Draft until you like what you see, then disable it and rerun the same prompt for production-quality output.
Dial In Detail: New Quality & Experimental Parameters
-
--q 4doubles the previous fidelity ceiling, giving razor-sharp textures and micro-detail—at a steep GPU cost. -
--expactivates experimental stylization, letting the model push lighting, grading, and composition beyond what--stylizecan achieve. Use sparingly with--q 4to avoid over-rendered chaos.
Keep Characters On-Model: Omni Reference (
--oref
)
The old Character Reference is gone; Omni Reference replaces and expands it. Drop in any reference image and append
--oref
plus an optional weight (
--ow 0.5-2
). Midjourney anchors that exact subject—even across wildly different scenes or styles.
/imagine epic space opera movie poster --oref https://imgur.com/myhero.png --ow 1.2 --v 7
Style on Autopilot: Personalization & Moodboards
If you’ve been ranking images, V7 automatically applies your saved tastes through personalization profiles—no extra flag needed. Combine this with classic
--sref
or moodboard links to steer colors and lighting without locking in content.
Edit, Pan & Zoom (Still V6.1 Under the Hood)
Buttons for Pan, Zoom-out, Variations, and the Full Editor let you in-paint or out-paint after generation. These steps temporarily drop to the V6.1 engine, so results may look slightly different—don’t worry, that’s normal.
V7 vs V6 at a Glance
| 🆕 Feature | What It Unlocks | Old Status |
|---|---|---|
| Draft Mode | 10× faster, ½ GPU | N/A |
| Conversational Mode | Voice/text prompt drafting | N/A |
| Omni Reference | Consistent characters/objects | Replaced Character Ref |
--q 4
| Ultra-high fidelity | Ceiling was
--q 2
|
--exp
| Experimental stylization | Unavailable |
| Personalization default-on | Auto style matching | Opt-in |
Example Workflow for Beginners
- Brainstorm in Draft + Conversational: “Make a cozy reading nook with a cat and soft morning light.”
- Refine by adjusting text prompt edits while keeping Draft on.
- Lock in your cat: upload a reference photo and add
--oref 1. - Finalize: turn Draft off, add
--q 4 --exp, and upscale the winner.
Final Thoughts
Midjourney V7 turns prompt art into an iterative playground—perfect for LLM artists who think in words first and tweak visuals later. Start cheap with Draft, anchor your characters with Omni Reference, and only pay the GPU premium when your concept sings. Happy generating!
Related
Discover more from Fornelli by Design | FxD
Subscribe to get the latest posts sent to your email.



