
Originaldokument
GPT Image 2 · en
Scene:
Create a square dependent campaign visual inside the same romantic environment, with the same soft candle glow, blush-burgundy palette, and layered restaurant depth, but widen the framing to prioritize the table setting and atmosphere. The camera should still feel tied to the anchor scene, preserving the same lens behavior and visual polish while turning the composition into a hospitality hero suitable for reservations, social rollout, or event promotion.
Subject:
Show an elegant two-place dinner setting with florals, dessert plating, glassware, folded napkins, and one discreet wrapped gift or premium token that keeps gifting logic alive. Preserve the exact color balance, floral character, satin accents, and light direction from the reference model while emphasizing the dining experience. The scene must remain tasteful, premium, and free of any readable restaurant materials.
Important details:
Maintain the exact campaign grammar from the reference image. Only shift the hierarchy toward tablescape and mood. Make plate glaze, linen folds, candle wax, glass refraction, and petal texture feel physically convincing. Avoid generic wedding vibes; this should read as a refined Valentines campaign for restaurants or experiential brands. The viewer should see unmistakable continuity with the anchor image while still getting a distinct deliverable surface.
Use case:
Valentines series study 3/9.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, preserve exact campaign identity from the reference image, no readable menus, no visible people dining, no wedding rings, no bouquet excess, no cheesy red neon, no logo glassware, no banquet-hall clutter.Verwendung
- Den vollständigen Original-Prompt einschließlich Parametern kopieren.
- Motiv, Text und Komposition an die eigenen Anforderungen anpassen.
- Ein kompatibles externes Werkzeug nutzen; Zugangsbedingungen und Ergebnis prüfen.
Vor dem Start
- Prüfen, ob ein Referenzbild nötig ist. Kompatibilität belegt nicht das tatsächlich verwendete Modell.
- Namen, Zahlen und Bildtexte prüfen; Modellversionen können andere Ergebnisse liefern.





