🎨 Image Prompt Anatomy
[Subject] + [Setting] + [Style] + [Camera] + [Lighting]
+ [Mood] + [Details] + [Aspect ratio]
Example:
"A young Bangladeshi woman in traditional saree,
standing on a rickshaw-filled Dhaka street,
documentary photography style,
35mm lens, shallow depth of field,
golden hour sunlight,
vibrant and nostalgic mood,
detailed textile patterns, rain-soaked ground
--ar 16:9 --v 6"
🖼️ Midjourney Parameter ⭐
| Param | উদ্দেশ্য |
|---|
| --ar 16:9 | Aspect ratio |
| --v 6 | Version |
| --style raw | Less stylized |
| --s 250 | Stylize (0-1000) |
| --chaos 50 | Variation (0-100) |
| --seed 12345 | Reproducibility |
| --niji | Anime style |
| --tile | Seamless pattern |
| --no text | Exclude element |
🎭 Style Keyword
- Photography: 35mm, portrait, macro, aerial, cinematic
- Art: oil painting, watercolor, pencil sketch, ukiyo-e, digital art
- 3D: octane render, Pixar style, unreal engine, low poly
- Retro: 1980s, vintage, polaroid, sepia
- Illustration: flat design, isometric, children's book
💡 Lighting Keyword
- Golden hour, blue hour
- Soft natural light
- Dramatic rim lighting
- Studio lighting, three-point
- Moody chiaroscuro
- Neon cyberpunk glow
- Volumetric god rays
👨🎨 Artist Reference
"...in the style of [Artist name]"
Photography: Annie Leibovitz, Ansel Adams, Steve McCurry
Painting: Van Gogh, Monet, Hokusai
Digital: Greg Rutkowski, Artgerm, Ross Tran
Animation: Studio Ghibli, Pixar, Disney
📐 Aspect Ratio ব্যবহার
- 1:1 — Instagram post
- 9:16 — Instagram/TikTok story
- 16:9 — YouTube thumbnail, banner
- 4:5 — Portrait photo
- 2:3 — Book cover
- 3:1 — LinkedIn banner
🔧 Advanced Technique
Image Reference (Midjourney)
[image URL] A woman in red dress
--iw 1.5 (image weight)
// Existing image-এর style/subject reuse
Style Reference
--sref [image URL] --sw 100
// Specific style copy
Character Reference (--cref)
--cref [image URL] --cw 100
// Same character maintain across scenes
Multi-prompt Weight
"bright city::2 rainy night::1"
// "bright city" 2x emphasis
🖌️ DALL-E 3 (ChatGPT-এ)
- Natural language — paragraph-style prompt
- Text render ভালো (sign, poster)
- ChatGPT expands prompt automatically
- Inpainting: "edit just the sky, keep rest"
🌊 Stable Diffusion (open source)
Positive + Negative prompt
Positive:
beautiful landscape, mountain, lake, masterpiece,
8k, detailed, sharp focus
Negative:
blurry, low quality, distorted, extra limbs,
watermark, signature, ugly
ComfyUI Workflow
- Node-based visual editor
- LoRA — fine-tuned style/character
- ControlNet — pose, depth, edge control
- Upscaler — 4K upscale
- Inpaint/outpaint
🎯 Use Case Example
Product Shot
"Minimalist product photography of a silver wireless
headphones on a white marble surface, soft studio
lighting from the left, subtle shadow, professional
e-commerce style, 4K, sharp focus --ar 1:1"
Social Media Post
"Instagram-worthy flat lay of healthy breakfast:
avocado toast, berries, coffee, on a wooden table,
top-down view, natural morning light, warm tones
--ar 1:1 --style raw"
Logo Concept
"Minimalist vector logo for a tech startup named Nexus,
abstract connected nodes forming an N shape,
blue gradient, clean sans-serif typography,
white background, corporate identity
--ar 1:1 --style raw --v 6"
❌ Common Mistake
- Vague prompt ("beautiful image")
- Conflicting style (oil painting + photorealistic)
- Too many subjects (confusing)
- Negative prompt missing
- Aspect ratio mismatch platform-এর সাথে
🏆 Power Tips ⭐
- ৪টা variant generate → best pick → upscale
- Prompt library বানান (category-wise)
- Seed save করে variation explore
- Hands/faces problematic → re-roll বা inpaint
- Reference image weight experiment
- Commercial use-এ license check
🔐 Copyright Concern
- "In the style of living artist" — ethical concern
- Trademark/logo generate avoid
- Celebrity likeness legal risk
- Adobe Firefly — commercially safe training data