50 lines
2.5 KiB
Markdown
50 lines
2.5 KiB
Markdown
# Art Style Format Example
|
|
|
|
Below is the standard format for `art_style.md`.
|
|
|
|
```markdown
|
|
Art Style: Cinematic Realism
|
|
|
|
Style Description:
|
|
Warm film tones, primarily natural lighting, 35mm film grain texture.
|
|
Referenced from the visual language of Into the Wild and Nomadland.
|
|
Widescreen ratio, anamorphic lens oval bokeh and soft edges.
|
|
|
|
Keywords: cinematic, film grain, natural lighting, warm amber tones, 35mm film, anamorphic
|
|
Palette: warm yellow, amber, deep brown, golden hour gold
|
|
Lighting: natural light, golden hour, backlit silhouette, Tyndall effect
|
|
|
|
Recommended Model: Midjourney V7 (best for cinematic texture) / Kling (character realism)
|
|
|
|
Reference:
|
|
- style_cache/cinematic/01_highway.png
|
|
- style_cache/cinematic/02_bar.png
|
|
- style_cache/cinematic/03_canyon.png
|
|
- style_cache/cinematic/04_starry.png
|
|
|
|
Style JSON:
|
|
{
|
|
"visual_style_analysis": {
|
|
"color_palette": ["#D4A574", "#8B6914", "#2C1810", "#E8C98E", "#4A3728"],
|
|
"color_mood": "warm amber and golden tones with deep brown shadows",
|
|
"lighting_and_shadows": "natural golden hour lighting, strong rim light, Tyndall effect, soft diffused shadows",
|
|
"texture": "35mm film grain, slight lens aberration, organic noise",
|
|
"composition_style": "wide anamorphic framing, rule of thirds, deep depth of field for landscapes",
|
|
"line_style": "soft organic edges, no hard outlines, photographic realism",
|
|
"art_style_keywords": ["cinematic", "film grain", "golden hour", "anamorphic", "warm amber", "natural lighting", "35mm film", "photorealistic"]
|
|
}
|
|
}
|
|
```
|
|
|
|
## Format Notes
|
|
|
|
- First line format: `Art Style: ` + style name
|
|
- `Style Description:` 2-4 sentences describing the visual language, may reference films/works
|
|
- `Keywords:` comma-separated English keywords, used for subsequent image generation prompts
|
|
- `Palette:` list main color tendencies
|
|
- `Lighting:` list lighting and shadow characteristics
|
|
- `Recommended Model:` recommend the most suitable model based on style, with brief rationale
|
|
- `Reference:` list selected reference image paths (relative to project directory), one per line, starting with `- `. Built-in presets are auto-cached to `style_cache/{style}/{filename}.png` by the script; custom styles may point to local files under `style_references/`
|
|
- `Style JSON:` structured style data returned by `read_media` analysis of reference images (JSON format), injected into prompts for character and prop image generation
|
|
- Content indented 2 spaces for visual alignment
|