Create an ultra-photorealistic IMAX-style cinematic image in 8K, formatted in a PERFECT 1:1 SQUARE ASPECT RATIO (e.g., 8192x8192), optimized for viral Facebook engagement — humorous, iconic, and full of “inside joke” energy.
Tone:
Epic + humorous + slightly sarcastic + “quem é do grupo sabe”
========================
GLOBAL HARD CONSTRAINT
========================
TOTAL_SUBJECTS = 5 (ABSOLUTE)
UNIQUE_FACES = 5
Each subject must map exactly to one of the provided faces:
(photo 1), (photo 2), (photo 3), (photo 4), (photo 5)
Rules:
- One face per person
- No duplication
- No face alteration
- Identity preservation = MAXIMUM
========================
FRAME ADAPTATION (1:1 CRITICAL)
========================
- Aspect ratio MUST be 1:1 (square)
- All 5 subjects must fit comfortably inside the square frame
- No cropping of heads or limbs
- Maintain visual balance within square composition
Safe framing:
- Top margin: no head cutoff
- Bottom margin: visible body grounding
- Side margins: no subject clipping
========================
SCENE SEGMENTATION (CRITICAL)
========================
ZONE_L (left third of square) → EXACTLY 3 people
ZONE_R (right third of square) → EXACTLY 2 people
ZONE_C (center vertical band) → EMPTY
Rules:
- No crossing zones
- No limbs invading other zones
- Center must stay visually clean
========================
VIRAL CONCEPT (FACEBOOK ENERGY)
========================
Scene represents:
👉 “Quando o grupo entra na discussão”
👉 “Os que sabem vs os que acham que sabem”
👉 “O ADM chegou… acabou a discussão”
Left side (3 people):
- Chaotic expert energy
- Expressions:
• “deixa que eu explico”
• “isso tá errado”
• “já vi isso mil vezes”
Right side (2 people):
- Calm authority
- ADM + especialista dominante
- Expressions:
• olhar analítico
• leve ironia
• confiança silenciosa
All subjects looking directly at camera
========================
FACIAL EXPRESSIONS
========================
- 5 unique expressions
- Mix:
• sarcasm
• confiança
• humor
• superioridade leve
• “eu avisei”
Rules:
- No distortion
- Natural anatomy
- High micro-detail
========================
COMPOSITION (SQUARE OPTIMIZED)
========================
- Central corridor aligned vertically
- Left third: 3 stacked people
- Right third: 2 stacked people
- Center vertical strip empty (visual tension)
Camera:
- Eye-level
- Slight wide-angle (28mm feel)
- Framing adjusted to fit square without distortion
========================
WARDROBE
========================
- 5 unique outfits
- No repetition
- Realistic textures
========================
ENVIRONMENT
========================
Abandoned corridor:
- Worn wooden floor
- Cracked dirty walls
- Spider webs
Wall text:
"Gemini Epic Prompts"
- Large
- Black
- Slightly worn
- Centered within visible square area
========================
BACKGROUND
========================
- Broken window centered in square
- Moon visible
- Tree branches entering
- Subtle fog for depth
========================
LIGHTING
========================
- Moonlight from background
- Soft frontal fill
Rules:
- All faces evenly visible
- No dark faces
- Catchlight in eyes
========================
ANATOMY LOCK
========================
- No fusion
- No extra limbs
- No distortion
- Natural posture
========================
CAMERA SPECS
========================
- IMAX digital style
- 28mm anamorphic look
- f/2.8
- ISO 400
- 1/48 shutter
Focus:
- All faces sharp
- Background slightly softer
========================
AUTO-CORRECTION SYSTEM
========================
If any issue:
- subjects > 5 → remove
- subjects < 5 → add
- duplicate faces → fix
- wrong faces → restore
- anatomy error → correct
- fusion → separate
If zone fails:
→ regenerate ONLY that zone
========================
NEGATIVE CONSTRAINTS
========================
DO NOT GENERATE:
- More or less than 5 people
- Wrong faces
- Duplicate faces
- Blurry faces
- Extra limbs
- Deformations
- Merged bodies
- Repeated outfits
- Cartoon / CGI look
========================
FINAL VALIDATION
========================
- Exactly 5 people
- Left = 3, Right = 2, Center empty
- All inside square frame (no cropping)
- 5 unique correct faces
- All faces visible and sharp
- No anatomical errors
- No fusion
If ANY fails:
→ auto-correct
========================
FINAL OBJECTIVE
========================
Generate a highly viral 1:1 cinematic image optimized for Facebook feed, combining humor, authority, and inside-joke energy, with perfect technical execution and balanced square composition.