补充分镜视频相关文档说明

This commit is contained in:
puke
2025-11-12 20:12:28 +08:00
parent f7e3162a4a
commit 830436d8dc
8 changed files with 156 additions and 8 deletions

View File

@@ -21,6 +21,10 @@ comfyui:
default_workflow: "runninghub/image_flux.json"
prompt_prefix: "Minimalist illustration style"
video:
default_workflow: "runninghub/video_wan2.1_fusionx.json"
prompt_prefix: "Minimalist illustration style"
tts:
default_workflow: "selfhost/tts_edge.json"
```
@@ -48,6 +52,13 @@ comfyui:
- `default_workflow`: Default image generation workflow
- `prompt_prefix`: Prompt prefix
### Video Configuration
- `default_workflow`: Default video generation workflow
- `runninghub/video_wan2.1_fusionx.json`: Cloud workflow (recommended, no local setup required)
- `selfhost/video_wan2.1_fusionx.json`: Local workflow (requires local ComfyUI support)
- `prompt_prefix`: Video prompt prefix (controls video generation style)
### TTS Configuration
- `default_workflow`: Default TTS workflow

View File

@@ -154,15 +154,39 @@ Suitable for Instagram, WeChat Moments, and other platforms.
---
## Template Naming Convention
Templates follow a unified naming convention to distinguish different types:
- **`static_*.html`**: Static templates
- No AI-generated media content required
- Pure text style rendering
- Suitable for quick generation and low-cost scenarios
- **`image_*.html`**: Image templates
- Uses AI-generated images as background
- Invokes ComfyUI image generation workflows
- Suitable for content requiring visual illustrations
- **`video_*.html`**: Video templates
- Uses AI-generated videos as background
- Invokes ComfyUI video generation workflows
- Creates dynamic video content with enhanced expressiveness
## Template Structure
Templates are located in the `templates/` directory, grouped by size:
```
templates/
├── 1080x1920/ # Portrait (11 templates)
├── 1920x1080/ # Landscape (2 templates)
└── 1080x1080/ # Square (1 template)
├── 1080x1920/ # Portrait
│ ├── static_*.html # Static templates
│ ├── image_*.html # Image templates
│ └── video_*.html # Video templates
├── 1920x1080/ # Landscape
│ └── image_*.html # Image templates
└── 1080x1080/ # Square
└── image_*.html # Image templates
```
---

View File

@@ -16,10 +16,42 @@ Pixelle-Video is built on the ComfyUI architecture and supports custom workflows
Located in `workflows/selfhost/` or `workflows/runninghub/`
Used for Text-to-Speech, supporting various TTS engines:
- Edge-TTS
- Index-TTS (supports voice cloning)
- Other ComfyUI-compatible TTS nodes
### Image Generation Workflows
Located in `workflows/selfhost/` or `workflows/runninghub/`
Used for generating static images as video backgrounds:
- FLUX series models
- Stable Diffusion series models
- Other image generation models
### Video Generation Workflows
Located in `workflows/selfhost/` or `workflows/runninghub/`
**New Feature**: Supports AI video generation to create dynamic video content.
**Preset Workflows**:
- `runninghub/video_wan2.1_fusionx.json`: Cloud workflow (recommended)
- Based on WAN 2.1 model
- No local setup required, accessed via RunningHub API
- Supports Text-to-Video generation
- `selfhost/video_wan2.1_fusionx.json`: Local workflow
- Requires local ComfyUI environment
- Requires installation of corresponding video generation nodes
- Suitable for users with local GPU
**Use Cases**:
- Works with `video_*.html` templates
- Automatically generates dynamic video backgrounds based on scripts
- Enhances visual expressiveness and viewing experience
---
## Custom Workflows