Image Generation / DALL·E
Convert prompts between platforms, explaining what changes and why — cross-model porting intelligence.
Why This Prompt Exists
A prompt that works perfectly in Midjourney produces garbage in DALL·E — and vice versa. The syntax, parameter names, and interpretation styles are completely different.
You get:
- DALL·E prompts that Midjourney ignores (wrong parameter syntax)
- Midjourney prompts that DALL·E misinterprets (no weight syntax)
- no understanding of what needs to change when switching models
- wasted credits testing incompatible prompts
- inability to leverage both platforms’ strengths
But translation is systematic:
- parameter mapping: –ar (MJ) → size (DE)
- weight syntax: ::2 (MJ) → descriptive emphasis (DE)
- style control: –sref (MJ) → style keywords (DE)
- negative prompts: –no (MJ) → “without” descriptions (DE)
- text rendering: MJ poor → DE better (adjust expectations)
Without translation, you can’t use both tools effectively.
This prompt converts prompts between DALL·E and Midjourney.
The Prompt
Assume the role of a cross-platform prompt engineer who translates between models. Your task is to convert prompts between DALL·E and Midjourney formats. Generate: 1. SOURCE PROMPT - Platform: [MIDJOURNEY / DALL·E] - Full prompt: [paste] - Parameters used: [list] 2. TRANSLATION MAPPING | MJ Component | DE Equivalent | Translation Rule | |--------------|---------------|------------------| | --ar 16:9 | size=1792x1024 | Aspect ratio to pixel dimensions | | --ar 9:16 | size=1024x1792 | | | --c 50 | No direct equivalent | Use descriptive variation requests | | --s 500 | style=vivid + HD | Stylize to vivid style | | --no text | "without text" description | Negative to positive phrasing | | word::2 | "important word" emphasis | Weight to descriptive emphasis | | --sref | No equivalent | Describe style in words | | --seed | --seed (same syntax) | Works identically | 3. CONVERTED PROMPTS **Midjourney → DALL·E** `[subject description], [style keywords], [lighting], [color palette] --size [X]x[Y] --style [vivid/natural] --quality [standard/hd]` **DALL·E → Midjourney** `[subject description] --ar [X:Y] --s [value] --style [raw/creative] --no [negative terms]` 4. STRENGTH-BASED RECOMMENDATIONS | Task | Best Platform | Reason | |------|---------------|--------| | Text rendering | DALL·E | Significantly better at readable text | | Photorealistic | Midjourney | Better lighting, texture, composition | | Stylized illustration | Both | Different aesthetics, try both | | Character consistency | Midjourney | Has --cref (DALL·E lacks) | | Style reference | Midjourney | Has --sref (DALL·E lacks) | | Inpainting/editing | DALL·E | Native editor, easier workflow | | Fast iteration | DALL·E | Faster generations, lower cost | 5. TRANSLATION EXAMPLES **Example 1: MJ → DE** MJ: `cyberpunk city at night --ar 16:9 --s 500 --no text` DE: `Cyberpunk city at night, neon lights, rain on streets, without any text or signs, dramatic lighting, vibrant colors --size 1792x1024 --style vivid --quality hd` **Example 2: DE → MJ** DE: `A cozy coffee shop interior, warm lighting, wooden tables, steam rising from cups --size 1024x1024 --style natural` MJ: `A cozy coffee shop interior, warm lighting, wooden tables, steam rising from cups --ar 1:1 --s 200 --style raw` 6. PLATFORM-SPECIFIC NOTES **When translating TO DALL·E:** - Remove :: weight syntax (not supported) - Convert aspect ratios to pixel dimensions - Move negative prompts to positive "without" descriptions - Add specific style keywords (vivid/natural) **When translating TO Midjourney:** - Add --ar (default 1:1 if not specified) - Add --s 200-500 for stylization - Add --style raw for photorealism - Convert descriptive emphasis to :: weight syntax INPUTS: Source platform: [MIDJOURNEY / DALL·E] Target platform: [MIDJOURNEY / DALL·E] Original prompt: [PASTE THE PROMPT TO TRANSLATE] Intended use case: [E.G., "Text-heavy infographic" or "Fantasy character art"] RULES: - DALL·E has no direct equivalent for --c (chaos) or --sref (style reference) - Midjourney has no direct equivalent for DALL·E's inpainting/outpainting - Text rendering: always prefer DALL·E for text-heavy prompts - Photorealism: always prefer Midjourney for product shots, portraits - Test translated prompts on 2-3 variations before production - Keep a translation log for reuse (common conversions) - Some prompts cannot be perfectly translated; adjust expectations
How To Use It
- DALL·E has no direct equivalent for –c (chaos) or –sref (style reference) — describe these in words instead.
- Midjourney has no direct equivalent for DALL·E’s native inpainting/outpainting — use external tools.
- Text rendering: always prefer DALL·E for text-heavy prompts (logos, signs, book covers).
- Photorealism: always prefer Midjourney for product shots, portraits, realistic scenes.
- Test translated prompts on 2-3 variations before production.
- Keep a translation log for reuse — common conversions save time.
- Some prompts cannot be perfectly translated; adjust expectations accordingly.
Example Input
Source platform:
“MIDJOURNEY”
Target platform:
“DALL·E”
Original prompt:
`fantasy warrior with glowing sword –ar 16:9 –s 600 –no helmet`
Intended use case:
“Character art for a game banner”
Why It Works
Most users treat DALL·E and Midjourney as interchangeable — then get frustrated when prompts don’t transfer. They are not interchangeable.
This framework improves outcomes by forcing:
- parameter mapping (what translates directly vs. what needs transformation)
- strength-based recommendations (which tool for which task)
- translation examples (concrete before/after comparisons)
- platform-specific notes (what to add, remove, or adjust)
- expectation management (what can’t be perfectly translated)
Failure modes this prevents:
- DALL·E ignoring Midjourney parameters (wrong syntax)
- Midjourney misinterpreting DALL·E weight syntax (not supported)
- Text rendering in Midjourney (poor results)
- Photorealism in DALL·E (often less convincing)
This improves on: “Same prompt, different tool” guessing. Systematic translation produces usable results on both platforms.
Related to: MJ-01 (Midjourney Parameters), DE-01 (DALL·E Parameters) for reference.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.