Optimize the generation logic
This commit is contained in:
puke
puke
174
reelforge/prompts/image_generation.py
Normal file
174
reelforge/prompts/image_generation.py
Normal file
@@ -0,0 +1,174 @@
|
||||
"""
|
||||
Image prompt generation template
|
||||
|
||||
For generating image prompts from narrations.
|
||||
"""
|
||||
|
||||
import json
|
||||
from typing import List, Optional
|
||||
|
||||
|
||||
# ==================== PRESET IMAGE STYLES ====================
|
||||
# Predefined visual styles for different use cases
|
||||
|
||||
IMAGE_STYLE_PRESETS = {
|
||||
"stick_figure": {
|
||||
"name": "火柴人简笔画",
|
||||
"description": "stick figure style sketch, black and white lines, pure white background, minimalist hand-drawn feel",
|
||||
"use_case": "通用场景，简单直观"
|
||||
},
|
||||
|
||||
"minimal": {
|
||||
"name": "极简抽象",
|
||||
"description": "minimalist abstract art, geometric shapes, clean composition, modern design, soft pastel colors",
|
||||
"use_case": "现代感、艺术感"
|
||||
},
|
||||
|
||||
"concept": {
|
||||
"name": "概念化视觉",
|
||||
"description": "conceptual visual metaphors, symbolic elements, thought-provoking imagery, artistic interpretation",
|
||||
"use_case": "深度内容、哲学思考"
|
||||
},
|
||||
}
|
||||
|
||||
# Default preset
|
||||
DEFAULT_IMAGE_STYLE = "stick_figure"
|
||||
|
||||
|
||||
IMAGE_PROMPT_GENERATION_PROMPT = """# 角色定位
|
||||
你是一个专业的视觉创意设计师，擅长为视频脚本创作富有表现力和象征性的图像提示词，将抽象概念转化为具象的视觉画面。
|
||||
|
||||
# 核心任务
|
||||
基于已有的视频脚本，为每个分镜的"旁白内容"创作对应的**英文**图像提示词，确保视觉画面与叙述内容完美配合，增强观众的理解和记忆。
|
||||
|
||||
**重要：输入包含 {narrations_count} 个旁白，你必须为每个旁白都生成一个对应的图像提示词，总共输出 {narrations_count} 个图像提示词。**
|
||||
|
||||
# 输入内容
|
||||
{narrations_json}
|
||||
|
||||
# 输出要求
|
||||
|
||||
## 图像提示词规范
|
||||
- 语言：**必须使用英文**（用于 AI 图像生成模型）
|
||||
- 画面风格：{style_description}
|
||||
- 描述结构：scene + character action + emotion + symbolic elements
|
||||
- 描述长度：确保描述清晰完整且富有创意（建议 50-100 个英文单词）
|
||||
|
||||
## 视觉创意要求
|
||||
- 每个图像都要准确反映对应旁白的具体内容和情感
|
||||
- 使用象征手法将抽象概念视觉化（如用路径代表人生选择，用锁链代表束缚等）
|
||||
- 画面要表现出丰富的情感和动作，增强视觉冲击力
|
||||
- 通过构图和元素安排突出主题，避免过于直白的表现方式
|
||||
|
||||
## 关键英文词汇参考
|
||||
- 象征元素：symbolic elements
|
||||
- 表情：expression / facial expression
|
||||
- 动作：action / gesture / movement
|
||||
- 场景：scene / setting
|
||||
- 氛围：atmosphere / mood
|
||||
|
||||
## 视觉与文案配合原则
|
||||
- 图像要服务于文案，成为文案内容的视觉延伸
|
||||
- 避免与文案内容无关或矛盾的视觉元素
|
||||
- 选择最能增强文案说服力的视觉表现方式
|
||||
- 确保观众能通过图像快速理解文案的核心观点
|
||||
|
||||
## 创意指导
|
||||
1. **现象描述类文案**：用直观的场景表现社会现象
|
||||
2. **原因分析类文案**：用因果关系的视觉比喻表现内在逻辑
|
||||
3. **影响论证类文案**：用后果场景或对比手法表现影响程度
|
||||
4. **深入探讨类文案**：用抽象概念的具象化表现深刻思考
|
||||
5. **结论启发类文案**：用开放式场景或指引性元素表现启发性
|
||||
|
||||
# 输出格式
|
||||
严格按照以下JSON格式输出，**图像提示词必须是英文**：
|
||||
|
||||
```json
|
||||
{{
|
||||
"image_prompts": [
|
||||
"[detailed English image prompt following the style requirements]",
|
||||
"[detailed English image prompt following the style requirements]"
|
||||
]
|
||||
}}
|
||||
```
|
||||
|
||||
# 重要提醒
|
||||
1. 只输出JSON格式内容，不要添加任何解释说明
|
||||
2. 确保JSON格式严格正确，可以被程序直接解析
|
||||
3. 输入是 {{"narrations": [旁白数组]}} 格式，输出是 {{"image_prompts": [图像提示词数组]}} 格式
|
||||
4. **输出的image_prompts数组必须恰好包含 {narrations_count} 个元素，与输入的narrations数组一一对应**
|
||||
5. **图像提示词必须使用英文**（for AI image generation models）
|
||||
6. 图像提示词必须准确反映对应旁白的具体内容和情感
|
||||
7. 每个图像都要有创意性和视觉冲击力，避免千篇一律
|
||||
8. 严格遵守上述指定的画面风格要求（{style_description}）
|
||||
9. 确保视觉画面能增强文案的说服力和观众的理解度
|
||||
|
||||
现在，请为上述 {narrations_count} 个旁白创作对应的 {narrations_count} 个**英文**图像提示词。只输出JSON，不要其他内容。
|
||||
"""
|
||||
|
||||
|
||||
def build_image_prompt_prompt(
|
||||
narrations: List[str],
|
||||
min_words: int,
|
||||
max_words: int,
|
||||
image_style_preset: Optional[str] = None,
|
||||
image_style_description: Optional[str] = None
|
||||
) -> str:
|
||||
"""
|
||||
Build image prompt generation prompt
|
||||
|
||||
Args:
|
||||
narrations: List of narrations
|
||||
min_words: Minimum word count
|
||||
max_words: Maximum word count
|
||||
image_style_preset: Preset style name (e.g., "minimal", "stick_figure", "concept")
|
||||
Available presets: see IMAGE_STYLE_PRESETS
|
||||
image_style_description: Custom style description (overrides preset if provided)
|
||||
Example: "warm scenes, soft lighting, professional photography"
|
||||
|
||||
Returns:
|
||||
Formatted prompt
|
||||
|
||||
Examples:
|
||||
# Use preset style
|
||||
>>> build_image_prompt_prompt(narrations, 50, 100, image_style_preset="minimal")
|
||||
|
||||
# Use custom style
|
||||
>>> build_image_prompt_prompt(
|
||||
... narrations, 50, 100,
|
||||
... image_style_description="cyberpunk style, neon colors, futuristic"
|
||||
... )
|
||||
|
||||
# Use default style (stick_figure)
|
||||
>>> build_image_prompt_prompt(narrations, 50, 100)
|
||||
"""
|
||||
# Determine style description
|
||||
if image_style_description:
|
||||
# Custom description takes priority
|
||||
style_desc = image_style_description
|
||||
elif image_style_preset:
|
||||
# Use preset
|
||||
if image_style_preset not in IMAGE_STYLE_PRESETS:
|
||||
raise ValueError(
|
||||
f"Unknown preset '{image_style_preset}'. "
|
||||
f"Available presets: {list(IMAGE_STYLE_PRESETS.keys())}"
|
||||
)
|
||||
style_desc = IMAGE_STYLE_PRESETS[image_style_preset]["description"]
|
||||
else:
|
||||
# Use default preset
|
||||
style_desc = IMAGE_STYLE_PRESETS[DEFAULT_IMAGE_STYLE]["description"]
|
||||
|
||||
narrations_json = json.dumps(
|
||||
{"narrations": narrations},
|
||||
ensure_ascii=False,
|
||||
indent=2
|
||||
)
|
||||
|
||||
return IMAGE_PROMPT_GENERATION_PROMPT.format(
|
||||
narrations_json=narrations_json,
|
||||
narrations_count=len(narrations),
|
||||
min_words=min_words,
|
||||
max_words=max_words,
|
||||
style_description=style_desc
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user