中文介绍

这是一个基于 FLUX.2 [klein] 9B 训练的双参考图人物替换 LoRA。

它可以将 Image 2 中的人物身份和外貌迁移到 Image 1 的指定人物身上，同时尽可能保留 Image 1 原有的姿势、动作、表情、服装、构图、背景和光线。

本模型主要用于人物替换和身份迁移。它不是单纯的换脸模型，而是尝试替换完整的人物外貌，同时保持原始画面的其他内容不变。

参考图片的作用

• Image 1： 提供需要保留的原始场景，包括目标人物的位置、姿势、动作、表情、服装、身体方向、镜头角度、构图、背景和光线。

• Image 2： 提供需要迁移的人物身份、面部特征、发型和整体外貌参考。

主要特点

• 将 Image 2 中的人物身份迁移到 Image 1 的指定人物

• 尽量保留 Image 1 的姿势、动作和面部表情

• 尽量保留 Image 1 原有的服装和服装结构

• 保留 Image 1 的镜头角度、构图、背景和光线

• 减少 Image 2 中服装、姿势、构图和背景向结果中的泄漏

• 改善人物面部、发型和整体外貌的一致性

• 支持单人物画面和明确指定位置的多人物画面

使用方法

1. 将需要进行人物替换的原始图片加载为 Image 1。

2. 将目标人物参考图片加载为 Image 2。

3. 加载本 LoRA。

4. 使用自然语言描述需要替换的人物及其位置。

5. 根据人物相似度和原场景保留效果调整 LoRA 权重。

原始测试提示词

Replace the girl on the right in image 1 with the girl in image 2, while keeping the same actions, expressions, and background as in image 1.

推荐提示词

Replace the adult woman on the right in image 1 with the adult woman in image 2, while keeping the same pose, action, facial expression, outfit, camera angle, composition, background, and lighting as in image 1.

如果 Image 1 中只有一个人物：

Replace the adult woman in image 1 with the adult woman in image 2, while keeping the same pose, action, facial expression, outfit, camera angle, composition, background, and lighting as in image 1.

详细版本：

Replace the adult woman on the right in image 1 with the adult woman shown in image 2. Transfer the identity, facial features, hairstyle, and overall appearance from image 2, while preserving the pose, action, facial expression, body position, outfit, accessories, camera angle, framing, composition, background, and lighting of image 1.

推荐设置

• Base model： FLUX.2 [klein] 9B

• Model type： LoRA

• Reference images： 2

• Image 1： 原始场景及需要被替换的人物

• Image 2： 目标人物身份参考

• Trigger word： 无需触发词

• Workflow： ComfyUI

• LoRA strength： 使用展示图中的实际权重

如果展示图使用的是 1.0 权重：

• 推荐权重： 1.0

• 建议范围： 0.8–1.2

较低权重通常更容易保留 Image 1 的服装、背景和原始构图，但人物身份迁移效果可能较弱。

较高权重可能增强 Image 2 的人物特征，但也可能增加 Image 2 中发型、服装、姿势或背景元素的泄漏。

适用场景

• 写实人物替换

• 人物身份迁移

• 单人物肖像替换

• 全身人物替换

• 多人物场景中的指定人物替换

• 保留原服装的人物替换

• 保留原姿势、表情和背景的人物编辑

• 双参考图图像编辑工作流

展示图说明

每组对比图片包含：

• Image 1： 原始场景和需要被替换的人物

• Image 2： 目标人物身份参考

• No LoRA： 仅使用 FLUX.2 [klein] 9B 底模生成

• + LoRA： 在相同条件下加载本 LoRA 后生成

No LoRA 和 + LoRA 使用相同的输入图片、提示词、种子、分辨率和工作流，仅改变是否加载本 LoRA。

部分展示测试素材来自 Z-Image-Turbo 返图区或社区展示区。相关图片均为 AI 生成内容，仅用于模型功能测试、技术比较和效果展示。

English Description

This is a two-reference character-replacement LoRA trained for FLUX.2 [klein] 9B.

It transfers the identity and appearance of the person in Image 2 onto the specified person in Image 1, while preserving as much of the original pose, action, expression, outfit, composition, background, and lighting from Image 1 as possible.

This model is designed for character replacement and identity transfer. It is not limited to a simple face swap. It attempts to replace the overall appearance of the person while preserving the rest of the original image.

Reference Image Roles

• Image 1: Provides the original scene, target-person position, pose, action, facial expression, outfit, body direction, camera angle, composition, background, and lighting that should be preserved.

• Image 2: Provides the target identity, facial features, hairstyle, and overall appearance.

Main Features

• Transfers the identity from Image 2 onto the specified person in Image 1

• Preserves the pose, action, and facial expression from Image 1

• Preserves the original outfit and garment structure from Image 1

• Preserves the camera angle, composition, background, and lighting

• Reduces outfit, pose, composition, and background leakage from Image 2

• Improves facial, hairstyle, and overall identity consistency

• Supports single-person scenes and clearly specified subjects in multi-person scenes

How to Use

1. Load the original image containing the person to be replaced as Image 1.

2. Load the target identity reference as Image 2.

3. Enable this LoRA.

4. Describe the target person and position using natural language.

5. Adjust the LoRA strength according to identity accuracy and scene preservation.

Original Test Prompt

Replace the girl on the right in image 1 with the girl in image 2, while keeping the same actions, expressions, and background as in image 1.

Recommended Prompt

Replace the adult woman on the right in image 1 with the adult woman in image 2, while keeping the same pose, action, facial expression, outfit, camera angle, composition, background, and lighting as in image 1.

For an Image 1 containing only one person:

Replace the adult woman in image 1 with the adult woman in image 2, while keeping the same pose, action, facial expression, outfit, camera angle, composition, background, and lighting as in image 1.

Detailed version:

Replace the adult woman on the right in image 1 with the adult woman shown in image 2. Transfer the identity, facial features, hairstyle, and overall appearance from image 2, while preserving the pose, action, facial expression, body position, outfit, accessories, camera angle, framing, composition, background, and lighting of image 1.

Recommended Settings

• Base model: FLUX.2 [klein] 9B

• Model type: LoRA

• Reference images: 2

• Image 1: Original scene and person to be replaced

• Image 2: Target identity reference

• Trigger word: No trigger word required

• Workflow: ComfyUI

• LoRA strength: Use the actual value shown in the showcase examples

If the showcase images were generated at strength 1.0:

• Recommended strength: 1.0

• Suggested range: 0.8–1.2

Lower strengths may preserve more of the original outfit, background, and composition, but the identity transfer may become weaker.

Higher strengths may improve the target identity, but they can also increase hairstyle, clothing, pose, or background leakage from Image 2.

Suitable Use Cases

• Photorealistic character replacement

• Identity transfer

• Single-person portrait replacement

• Full-body character replacement

• Replacement of a specified person in multi-person scenes

• Character replacement while preserving the original outfit

• Character edits that preserve the original pose, expression, and background

• Two-reference image-editing workflows

Showcase Format

Each comparison contains:

• Image 1: The original scene and person to be replaced

• Image 2: The target identity reference

• No LoRA: Generated with the FLUX.2 [klein] 9B base model only

• + LoRA: Generated under the same conditions with this LoRA enabled

The No LoRA and + LoRA comparisons use the same input images, prompt, seed, resolution, and workflow. The only difference is whether this LoRA is enabled.

Some showcase references were obtained from the Z-Image-Turbo community return-image or gallery section. All displayed references are AI-generated and are used only for model evaluation, technical comparison, and demonstration.

免责声明 / Disclaimer

本资源仅用于 AI 图像编辑研究、学习、测试和合法创作。

使用者应确保自己对所有输入人物图片、身份参考图片、服装、角色设计和生成结果拥有必要的权利、许可、授权或有效同意。

不得将本模型用于欺骗、诈骗、冒充真实人物、身份盗用、骚扰、诽谤、侵犯隐私、非自愿私密内容、未经授权的肖像使用、涉及未成年人的不当内容或其他违法用途。

不得在未经本人明确同意的情况下，使用本模型制作可能被误认为是真实人物拍摄的敏感、误导性或损害性内容。

本模型不会自动向使用者授予输入人物、角色、服装、品牌、标志或其他第三方素材的版权、肖像权、商标权或商业使用权。

生成结果可能包含身份偏差、人物混合、面部结构错误、发型变化、服装变化、手部瑕疵或其他非预期结果。公开发布或商业使用前，请进行人工检查，并在必要时明确标注内容由 AI 生成或编辑。

展示素材仅用于非商业技术测试和模型效果比较。上传者不对原始展示素材主张所有权，相关权益归原上传者或相应权利人所有。

使用者还应遵守 FLUX.2 [klein] 9B 的底模许可证、Civitai 平台规则以及所在地适用法律。

This resource is intended only for AI image-editing research, learning, testing, and lawful creative use.

Users must possess the necessary rights, permissions, authorization, or valid consent for all input portraits, identity references, clothing, character designs, and generated outputs.

The model must not be used for deception, fraud, impersonation of real individuals, identity theft, harassment, defamation, privacy violations, non-consensual intimate content, unauthorized likeness use, inappropriate content involving minors, or any unlawful activity.

The model must not be used without clear consent to create sensitive, misleading, or harmful content that could reasonably be mistaken for authentic imagery of a real person.

This model does not grant copyright, personality, publicity, likeness, trademark, or commercial rights to any person, character, clothing, brand, logo, or third-party material shown in the references.

Generated results may contain identity drift, mixed identities, facial-structure errors, hairstyle changes, clothing changes, hand artifacts, or other unintended results. All outputs should be manually reviewed before publication or commercial use. AI-generated or AI-edited content should be disclosed where appropriate.

The showcase materials are used only for non-commercial technical testing and model comparison. The uploader does not claim ownership of the original showcase materials. Any applicable rights remain with the original uploaders or respective rights holders.

Users must also comply with the FLUX.2 [klein] 9B base-model license, Civitai platform rules, and all applicable laws.