Image to Midjourney Prompt Generator
Upload an image and get a paste-ready Midjourney V7 prompt in about 5 seconds — with --ar, --style, and --stylize already tuned. Works with V7 and V6.1, output drops straight into Discord or alpha.midjourney.com. Two free runs a day, no signup.
Drop an image here to generate a Midjourney prompt
Max 5MB · PNG, JPG, WEBP · takes about 5 seconds
Anatomy of a good Midjourney prompt
Midjourney V7 (default since June 2025) responds best to a tight natural-language sentence followed by explicit parameters. Parameters ALWAYS go at the end, after a double-hyphen `--`. The tool below outputs prompts in exactly this structure.
1. Subject phrase
close-up portrait of a cybernetic woman with translucent skinV7 weighs the first 5–8 tokens heavily; open with the single most important visual noun.
2. Medium / style tag
cinematic 35mm photography, shot on Arri AlexaAnchors the aesthetic. Camera/film names unlock specific texture profiles V7 recognizes.
3. Lighting + mood
neon reflections on wet skin, cold blue rim lightDescribe how light *interacts* with the subject — don't just say 'good lighting'.
4. --ar aspect ratio
--ar 16:9Default is 1:1. V7 supports up to 4:1 but ratios over 2:1 are experimental.
5. --stylize (--s)
--stylize 1500–1000 range, default 100. Lower = closer to your prompt, higher = more Midjourney 'house style'.
6. --style raw (optional)
--style rawFor photorealistic output without Midjourney's default artistic bias.
Real image, real prompt
The prompt below is the raw output of this tool on the image shown — not hand-written samples.
Young woman with long brown hair, looking at the camera, smiling slightly, standing, wearing button-down shirt with vertical black and white stripes, long sleeves, no footwear visible, set against outdoors, dark green foliage, blurred horizon line, light blue and orange sky, shown in a close-up portrait, with soft, warm light from the side, creating shadows on the face --ar 16:9 --style rawThings most beginners miss
1. Keep the subject in the first 8 words
V7 token-weighting favors early tokens. Lead with the noun, then describe. Burying the subject after 20 words of adjectives dilutes attention.
2. Use --style raw for photoreal
Without --style raw, V7 applies a subtle artistic finish that can read as 'AI-looking'. Raw mode produces more candid, less processed output.
3. --stylize 100–250 is the sweet spot
Below 50 you lose aesthetic cohesion; above 400 Midjourney takes over your prompt. Tool outputs tuned defaults; tweak only if needed.
4. Omni-Reference (--oref) for character consistency
For the same subject across scenes, use --oref <image-url> --ow 100. Higher --ow (up to 1000) preserves face/clothing more strictly.
5. Avoid old V5/V6 tricks in V7
Negative prompts via --no still work but are less needed in V7. Weight syntax like `subject::2 background::1` is also deprecated — V7 handles balance natively.
Midjourney vs other image models
| Capability | Midjourney | Flux | DALL·E 3 |
|---|---|---|---|
| Text in image | ⚠️ Improving in V7 | ✅ Best in class | ✅ Good |
| Photoreal skin | ✅ V7 excellent | ✅ Excellent | ⚠️ Plastic-looking |
| Style control via --sref | ✅ Unique strength | ❌ No native --sref | ⚠️ IP-Adapter hack |
| Runs on your own GPU | ❌ Cloud only | ✅ Open weights | ✅ Open weights |
Frequently asked questions
Does this work with Midjourney V7 and V8.1 alpha?
Yes for V7 (the current default since June 2025). V8.1 alpha launched April 14, 2026 on alpha.midjourney.com — our output is compatible; just swap --v 7 for --v 8.1 if you're testing the alpha. Prompt structure is identical across both.
Why does the tool include --ar automatically?
Because your input image's aspect ratio is the cheapest piece of information to extract and the most common parameter people forget. We match your source image ratio to the nearest Midjourney-supported --ar value.
Can I use the prompt directly on Discord?
Yes — paste the full output after `/imagine` in the #newbies channel or any server with the Midjourney bot. The tool outputs Discord-ready syntax. For alpha.midjourney.com just paste into the prompt box, no slash prefix.
What's the difference between --stylize and --chaos?
--stylize controls how much Midjourney's aesthetic filter is applied (0–1000). --chaos controls variation between the 4 grid images (0–100). High --chaos gives you wildly different interpretations per tile; low --chaos keeps them close together.
Does --style raw work with --oref and --sref?
Yes, all V7 parameters compose. A typical power-user prompt stacks: --ar 16:9 --style raw --sref <style-image> --oref <subject-image> --ow 200 --stylize 100. Our tool only outputs --ar + --stylize + optional --style raw by default — you add --sref/--oref manually with your own reference URLs.
Is 2 free runs per day enough to test the tool?
Yes, for deciding if the output matches what you'd write by hand. Run 2 images through this page, compare with your usual prompts, and if you want more, sign in for 5 bonus credits (no payment required for signup).
Other model guides
- Image to SDXL Prompt— Prefer local generation on your own GPU? SDXL gives you total control and no subscription.
- Image to Flux Prompt— Need better text rendering or photoreal skin? Flux 1.1 Pro handles both better than Midjourney V7.
Ready to generate your Midjourney prompt? Open the workspace and drop an image — first two runs each day are free.