Personal reflections on AI image generation prompts and key insights from leading guides

As the Tech-Sharing Lead at BCP, I occasionally explore various AI-driven tools and methodologies to enhance our internal knowledge-sharing culture. Recently, I’ve taken an interest in AI image generation prompts and prompting guides as part of our Prompting Campaign —not as an expert, but as someone eager to understand best practices for crafting effective prompts across different platforms.

During this exploration, I came across three highly valuable prompting guides that offer structured insights into optimizing AI-generated images:

  • Adobe Firefly – A Guide to Effective Prompts (click)
  • DALL·E 3 Prompting Tips & Tricks (click)
  • Midjourney Prompting Basics & V6 Prompt Structure (click)

Since the BCP Prompt Library will include AI-generated image prompts in the future, I wanted to get a better sense of how different AI models interpret and process text inputs. The following is my personal assessment of these guides—based on their theoretical insights rather than extensive hands-on experience with the generators themselves.

If you have deeper experience with these tools, I’d love to hear your thoughts and corrections at the end!


Key Differences in Prompting Approaches

Adobe Firefly – Structured and Descriptive

Firefly suggests a structured approach to prompts:

Subject + Description + Context + Style = Image

  • Firefly prompts should be concise yet descriptive (max. 175 words).
  • Using photographic terms improves realism.
  • Compositional elements like depth, lighting, and mood should be explicitly mentioned.
DALL·E 3 – Narrative and Text-Heavy

DALL·E 3 focuses on detailed, long-form prompts, with emphasis on storytelling:

  • Rich descriptions with contextual depth improve output.
  • AI interprets prompts like a scene from a book, meaning storytelling language can be effective.
  • Photographic details help, but are not essential—DALL·E can generate highly stylized outputs as well.
Midjourney V6 – Concise and Parameter-Driven

Midjourney works best with short, structured prompts and style-focused modifiers:

  • Comma-separated adjectives influence aesthetics.
  • Special parameters (e.g., --ar for aspect ratio, --v 6 for engine version) fine-tune results.
  • Less emphasis on full sentences—Midjourney favors efficient, keyword-driven prompts over long narratives.


Comparing Example Prompts for Different Use Cases

I briefly tested different image categories across all three systems using insights from the guides. Below are my sample prompts for each AI model. All Images are first try results. No refinement was done. I did include all results, Firefly and Midjourney initially provides 4 suggestions, while Dall-E provides one.

Key Differences Marked:

  • Bold = Common elements across platforms
  • Italic = Significant differences in approach

Realistic Images (Photo-Style AI-Generated Visuals)

Firefly Prompt

"A professionally captured, ultra-detailed image of an elderly fisherman casting his net into a misty lake at sunrise. Soft morning light, photorealistic, National Geographic style."**
DALL·E 3 Prompt

"A highly detailed, cinematic photograph of a fisherman casting his net into a calm, mist-covered lake at sunrise. Captured with a Canon EOS R5, HDR quality, soft golden lighting, real-world depth of field."**
Midjourney V6 Prompt

"Realistic photo of a fisherman casting a net on a tranquil misty lake, sunrise, golden light, ultra-detailed, Canon EOS R5, 8K rendering, hyperrealistic, --ar 16:9 --v 6"**
Summary:

✅ All prompts include subject, lighting, and realism-enhancing elements.
🛑 Midjourney relies heavily on parameters, while DALL·E uses technical photography terms.

Illustrations (Digital Art, Vector-Style Graphics)

Firefly Prompt

"A colorful vector-style illustration of a futuristic city skyline with flying cars, clean geometric shapes, bold outlines, vibrant cyberpunk aesthetic."**
DALL·E 3 Prompt

"A highly stylized vector illustration of a futuristic cityscape with floating cars, inspired by Bauhaus and cyberpunk design, using isometric perspective."**
Midjourney V6 Prompt

"Minimalist vector-style digital illustration of a sci-fi city with flying cars, clean lines, bold colors, futuristic, isometric, neon-lit skyline --ar 16:9 --v 6"**
Summary:

✅ All emphasize isometric vector aesthetics and cyberpunk elements.
🛑 Midjourney is parameter-driven; DALL·E prefers artistic references.

Concept Art (Game Design, Storytelling Visuals)

Firefly Prompt

"A fearless cyber-samurai standing atop a neon-lit skyscraper, dark rainy atmosphere, detailed armor, digital concept art, sci-fi cyberpunk aesthetic."**
DALL·E 3 Prompt

"A cinematic scene of a cyber-samurai warrior on a futuristic rooftop, illuminated by glowing neon signs. Inspired by Akira and Blade Runner, with intense reflections and rain-soaked streets."
Midjourney V6 Prompt

"Cyber-samurai on a neon-lit skyscraper, futuristic cityscape, dark rain-soaked atmosphere, ultra-detailed, sci-fi cyberpunk concept art --v 6 --ar 16:9"
Summary:

✅ Common cyberpunk theme, neon lighting, and dark atmosphere.
🛑 DALL·E focuses on cinematic storytelling, while Midjourney uses compact parameters.

Marketing & Ads (Social Media Creatives, Banners)

Firefly Prompt

"A sleek advertisement featuring a luxury perfume bottle, placed on a reflective marble surface, elegant lighting, minimalist branding, photorealistic studio shot."
DALL·E 3 Prompt

"A high-end perfume ad featuring a glass bottle on a white marble counter, surrounded by soft glowing light. Cinematic studio lighting, Vogue-inspired, fine art photography."
Midjourney V6 Prompt

"Luxury perfume bottle, placed on reflective marble, high-end advertisement, elegant minimalism, photorealistic product render, studio shot --ar 16:9 --v 6"
Summary:

✅ All prompts focus on luxury branding, elegant lighting, and photorealism.
🛑 DALL·E leans into artistic inspiration (Vogue style), Midjourney requires structured parameters.

Memes & Fun Content (AI-Generated Memes, Comics)

Firefly Prompt

"A funny cartoon illustration of a cat in a business suit, sitting at a desk, caption: 'CEO of Naps.' Vibrant colors, exaggerated comic style."
DALL·E 3 Prompt

"A meme-style comic of a cat in a suit, sitting behind a desk with a smug expression. Hand-drawn effect, meme format, simple flat colors, text overlay."
Midjourney V6 Prompt

"A cat in a business suit, sitting at a desk, exaggerated expressions, funny comic art, vibrant cartoon style, meme-like aesthetic --v 6 --ar 1:1"
Summary:

✅ All focus on humor, cats, and a business theme.
🛑 DALL·E highlights hand-drawn meme styles, while Midjourney depends on style parameters.

3D Renders (Product Visualization, Architecture)

Firefly Prompt

"A high-detail 3D render of a modern tiny house on a cliff, large glass windows, breathtaking sunrise reflections, architectural photography."
DALL·E 3 Prompt

"A photorealistic 3D-rendered image of a futuristic tiny home on a cliff, surrounded by nature. Ultra-detailed, smooth textures, concept architecture rendering."
Midjourney V6 Prompt

"Modern tiny house, 3D-rendered, large glass windows, futuristic design, photorealistic, architectural visualization, ultra-detailed --ar 16:9 --v 6"
Summary:

✅ All focus on modern tiny house architecture with photorealistic 3D rendering.
🛑 DALL·E emphasizes artistic architecture concepts, while Midjourney requires explicit rendering parameters.


Key Takeaways: Which AI is Best for Which Use Case?

🚀 Firefly (Adobe) – Best for structured, professional work, especially in photography, marketing, and 3D renders, where predefined styles and realism are needed.

🚀 DALL·E 3 – Best for narrative-driven prompts, creative meme formats, and conceptual storytelling in marketing and concept art. Handles artistic inspirations well and is great for detailed scene descriptions.

🚀 Midjourney V6 – Best for artistic freedom, rapid style experimentation, and detailed product/architecture renders. Its parameter-based fine-tuning makes it the most flexible for branding visuals, ads, and hyperrealistic 3D renders.

🔹 If you want marketing creatives, Firefly and Midjourney work best.
🔹 For humor/memes, DALL·E's simple, hand-drawn aesthetic is ideal.
🔹 For high-end product and architecture renders, Midjourney's structured approach excels.

Would you agree with this breakdown? Let me know if I missed any key aspects! 🚀

Final Thoughts: Your Feedback Needed!

This is my personal interpretation of the differences between Firefly, DALL·E 3, and Midjourney, based on the best practices outlined in their respective guides. Since my insights are theory-based rather than hands-on expertise, I would love to hear from those of you who have practical experience with these tools:

  • Does my assessment match your experience?
  • Are there key factors I overlooked?
  • Which AI do you prefer for specific image-generation tasks?

Let me know in the comments! Your input will help refine our BCP Prompt Library and improve AI-powered creativity across our platform. 🚀

Special THX to Max Beier, for assistance with creating the images!

p.s. did you find the totally odd image, not matching the prompt? ;)