## 中文说明

LTX-Video (2.3) IC-LoRA: 双人分镜头对话增强模型

本模型是基于 Lightricks LTX-2.3 底模训练的 IC-LoRA，专为双人同框对话、角色互动及分镜头视频生成场景深度优化。

---

### 一、 模型核心提升

1. 角色参考稳定性：显著提升双人同框时的人物特征一致性，减少角色漂移。

2. 分镜构图稳定性：针对影视化对话构图进行了加固，支持更精准的镜头控制。

3. 叙事连贯性：增强了对多段描述的理解力，使分镜间的过渡衔接更自然。

4. 风格兼容性：完美支持古风仙侠、现代都市、3D 动漫等主流视觉风格。

---

### 二、 模型基本信息

1. 基础模型：Lightricks/LTX-2.3

2. 许可证：Apache-2.0

3. 管道标签：image-to-video, text-to-video

4. 模型用途：仅供学习交流使用

5. 开发者：麻雀 AI

---

### 三、 运行指南

1. 推荐平台：ComfyUI

2. 支持工作流：ComfyUI 官方 LTX 工作流、KJ-LTX 插件工作流

3. 生成模式：文生视频 (T2V) 与 图生视频 (I2V) 均支持

4. 硬件参考：RTX 5090 显卡在 720P 分辨率下，单条视频生成耗时约 2 分钟

---

### 四、 推荐参数配置

1. 分辨率：建议使用 16:9 (如 1280x720)

2. 时长与帧率：建议时长 ≥10 秒，帧率设定为 24 FPS

3. LoRA 权重设定：

- 独立使用建议：0.6 - 1.0

- 叠加其他 LoRA 使用时建议：0.3 - 0.5

---

### 五、 Prompt 编写规范

1. 编写逻辑：需包含完整的场景描述 + 角色设定 + 分镜设计 + 镜头语言，强化双人对话互动逻辑。

2. 触发词说明：无需特定触发词。

3. 参考提示词示例（古风）：

[场景] 古风仙侠，云端仙境，粉紫霞光与翻涌云海交织，暖金日光穿透云层洒落，仙气缥缈朦胧，氛围梦幻唯美，尽显东方仙侠的空灵意境。

[角色] 女子：古典发髻挽起，点缀银质花叶发饰；身着天蓝色轻纱广袖裙，裙身绣有鎏金莲花纹样，肩带与袖摆薄如蝉翼，手持琉璃净瓶，瓶中盛有清透仙水，宛如瑶池临凡的仙子。男子：墨色长发束起，银质发冠点缀；身着月白色广袖仙袍，袖口与衣摆绣有银线云纹，手持白玉柄拂尘，尘丝垂落随风轻晃，仙风道骨，仿若天界执掌时序的仙君。

分镜 1 (全景 3 秒) 俯拍全景。云海翻涌如浪，粉紫霞光铺满天际，暖金日光斜洒而下，两人并肩立于云端，衣袂随仙气轻扬，与云雾霞光融为一体，尽显仙境的辽阔与缥缈。

分镜 2 (中景 4 秒) 正面中景。镜头缓缓推近，女子微微侧身，指尖轻托琉璃净瓶，瓶中仙水泛着细碎微光，她抬眸望向身旁男子，眉眼含柔，轻声道：“仙君，今日凡间春汛已平，这净瓶仙水可滋养人间草木，唤回生机。”

分镜 3 (近景 3 秒) 侧拍近景。男子手持拂尘，目光落在女子手中的净瓶上，语气平和温润：“有劳仙子费心，人间得此仙泽，定能岁岁繁茂，不负春日。”

分镜 4 (特写 2 秒) 特写镜头。聚焦女子手中的琉璃净瓶，瓶身通透如冰，仙水在光下流转泛着莹光，裙身鎏金莲花纹样若隐若现，与瓶中微光相映，细节尽显东方仙意的精致。

分镜 5 (全景 4 秒) 拉远全景。镜头缓缓拉远，两位仙子的身影渐渐融入云海霞光之中，拂尘尘丝与净瓶微光在云雾中闪烁，最终化作天际一点流光，只留缥缈仙踪于天地间。

---

### 六、 效果说明与局限性

1. 优势风格：在古风、现代、3D 动漫类双人对话场景中表现最佳。

2. 已知限制：受限于 LTX-2.3 底模性能，极其复杂的双人肢体互动（如缠绕、打斗）可能出现形变。

3. 运动幅度：建议以对话和微动作为主，大动态动作的连贯性仍有提升空间。

## English Description

This model is an IC-LoRA trained based on the Lightricks LTX-2.3 base model, specifically optimized for two-person dialogue, character interaction, and multi-shot cinematic video generation.

---

### 1. Core Enhancements

1. Character Consistency: Significantly improves character feature stability when two people are in the same frame, reducing character drift.

2. Compositional Stability: Reinforced for cinematic dialogue compositions, supporting more precise camera control.

3. Narrative Coherence: Enhanced understanding of multi-segment descriptions, making transitions between shots more natural.

4. Style Compatibility: Perfectly supports various styles including Ancient Chinese Fantasy, Modern Urban, and 3D Anime.

---

### 2. Basic Information

1. Base Model: Lightricks/LTX-2.3

2. License: Apache-2.0

3. Pipeline Tags: image-to-video, text-to-video

4. Usage: For educational and exchange purposes only

5. Developer: Maque AI (麻雀 AI)

---

### 3. Running Guide

1. Recommended Platform: ComfyUI

2. Supported Workflows: ComfyUI Official LTX Workflow, KJ-LTX Plugin Workflow

3. Generation Mode: Supports both Text-to-Video (T2V) and Image-to-Video (I2V)

4. Hardware Reference: On an RTX 5090 at 720P resolution, a single video takes approximately 2 minutes to generate.

---

### 4. Recommended Parameters

1. Resolution: 16:9 (e.g., 1280x720) is recommended.

2. Duration & FPS: Duration ≥ 10s, Frame Rate set to 24 FPS.

3. LoRA Weight:

- Standalone use: 0.6 - 1.0

- When stacked with other LoRAs: 0.3 - 0.5

---

### 5. Prompting Guide

1. Writing Logic: Prompts should include a complete Scene Description + Character Settings + Storyboard Design + Camera Language to reinforce dialogue logic.

2. Trigger Words: No specific trigger word required.

3. Reference Example (Ancient Chinese Fantasy):

- [Scene] Ancient Chinese fantasy, celestial realm above clouds, pink and purple sunset glow blending with rolling clouds, warm golden sunlight piercing through, ethereal and dreamy atmosphere.

- [Characters] Female: Classical bun hairstyle, silver floral hairpins, wearing a sky-blue translucent flowing gown with gold lotus patterns, holding a glazed jade bottle. Male: Black hair tied up with a silver crown, wearing a moon-white wide-sleeved immortal robe with silver cloud embroidery, holding a white jade whisk.

- Shot 1 (Wide Shot, 3s): Overhead view. Two immortals stand side by side above the clouds, robes fluttering in the mist, showing the vastness of the celestial realm.

- Shot 2 (Medium Shot, 4s): Camera slowly pushes in. The female character turns slightly, looking at the male with gentle eyes, whispering: "The spring floods in the mortal world have subsided; this celestial water can restore life to the land."

- Shot 3 (Side Medium-Close Shot, 3s): The male character looks at the bottle in her hand, replying warmly: "Thank you for your kindness. With this blessing, the mortal world will surely flourish."

- Shot 4 (Close-up, 2s): Focus on the glazed jade bottle. The body is translucent like ice, with light reflecting off the water inside and the golden lotus patterns on the dress flickering in the background.

- Shot 5 (Wide Shot, 4s): Camera pulls back. The two figures gradually merge into the sunset glow and clouds, eventually turning into a streak of light in the sky.

---

### 6. Limitations & Notes

1. Best Styles: Performs best in Ancient Fantasy, Modern, and 3D Anime dialogue scenes.

2. Known Limitations: Due to the constraints of the LTX-2.3 base model, extremely complex physical interactions (e.g., wrestling or hugging) may cause deformation.

3. Motion Range: Recommended for dialogue and subtle movements; high-intensity action consistency still has room for improvement.

---