• Base model: ERNIE-Image

• Strength: 0.85–1.0

• Resolution: trained at 512, generate at 1024

• CFG: 4.6 (Base) / 1.0 (Turbo)

• Steps: 35–50 (Base) / 8–9 (Turbo)

Vision Vanguard 23 [ERNIE-Image]

General-purpose film-grade realism for any subject: high contrast, visible grain, dramatic chiaroscuro and atmospheric depth.

• Base model: ERNIE-Image

• Trigger words: none. It responds well to photographic language:

• Style: photograph, documentary-style, sharp focus, high contrast

• Lighting: chiaroscuro, natural light, dramatic sidelight, soft diffused lighting

• Composition: eye-level, wide-angle, shallow depth of field, top-down perspective

• Strength: 0.8–1.0. Higher values sharpen focus and increase texture contrast.

• Negative prompt (optional): if results look too clean, add "digital art, rendering, illustration, oversaturated".

Prompting: subject → action/moment → environment and materials → lighting → micro-detail (for example skin pores, water beads, film grain).

Tips

• It handles grids and repeating patterns well. Try "a photographic grid showing…" or "a sequence of…".

• For landscapes, try atmospheric haze, pastel twilight, layered mountains, or turquoise water.

• It also renders unusual subjects (cybernetics, deep-sea life, abstract sculpture) photographically.