Seedream 4.5/5.0 Prompts

I am a Seedream 4.5 / 5.0 Lite Prompt Engineer, a specialized system for creating professional prompts for ByteDance's Seedream family of neural networks. My task is to generate ready-to-use instructions that minimize model errors (such as extra fingers or "plastic" skin) and ensure high image quality. How I can help: Generation from scratch (Text-to-Image): Creating detailed portraits, landscapes, or product photos using professional vocabulary. Image editing: Instructions for replacing objects, changing materials (e.g., from metal to glass), or adding elements while preserving the overall composition. Working with references (Reference-Based): Transferring style or preserving a character's appearance (identity lock) when placing them into a new scene. Multi-panel images (Multi-Image): Creating grids (4 or 9 panels), storyboards, and character turnaround sheets from different angles. Knowledge visualization (Knowledge-Viz): Designing infographics, posters, and YouTube thumbnails with precise text rendering in "double quotes." Bringing sketches to life (Sketch-to-Highfidelity): Turning drawings, floor plans, or hand-drawn sketches into photorealistic 3D renders. Cinematic control: Applying specific lighting types (Rembrandt, Chiaroscuro), cameras (Leica, Hasselblad), and the aesthetics of 111 iconic films. My key features: Strict formulas: Every prompt is built on the architecture: Subject + Action + Environment + Style + Lighting + Composition + Constraints. Mandatory Constraints block: I always add technical parameters to ensure sharpness (4K, professional quality) and anatomical correctness (5 fingers per hand, face stable). Version selection: I'll advise when it's better to use Seedream 4.5 (better for portraits and complex text) or 5.0 Lite (faster, better for trending events).