Sign In

Mirror CCTV Chinese Male Concept LoRA - 镜面/监控lora Krea2

0

Updated: Aug 23, 2026

concept

Download

1 variant available

SafeTensor

218 MB

Verified:

Type
LoRA
Stats

90

0

Reviews
Published

Aug 23, 2026

Base Model

Krea 2

Training
Steps: 2,000
Epochs: 10
Usage Tips
Clip Skip: 1
Hash
AutoV2
27A3D24F40
default creator card background decoration
Followers - 206

206

Likes - 554

554

Created on Civitai

krea_n3_c_sfw_mirror_wardrobe_00001_.png

镜面监控男性 Concept LoRA

中文描述:

这是一个面向写实男性场景的镜面与监控视角概念 LoRA。训练集包含 200 张独立图片:100 张镜面场景和 100 张固定监控机位场景,单人、双人画面均有覆盖。

镜面部分重点学习人物本体与镜中倒影同时出现的构图,包括全身镜、浴室镜、衣柜镜、酒店镜和健身房镜等场景。适合生成自然站立、整理衣服、交谈以及双人同框等画面。

监控部分重点学习高位固定摄像头、广角透视和非摆拍构图,包括走廊、电梯、楼梯间、停车场、商店入口和公寓大厅等场景。

镜面触发词:mirrormaleconcept

监控触发词:cctvmaleconcept

两个触发词建议分开使用,不建议在同一张图片里同时调用。推荐从 0.75 左右的 LoRA 权重开始测试。

镜面提示词示例:

mirrormaleconcept
Caption: A young Chinese man stands naturally in front of a full-length apartment mirror. His reflection matches his face, clothes, pose and limbs. The room is softly lit and realistic, with no phone and no additional person.

监控提示词示例:

cctvmaleconcept
Caption: A fixed high-angle security camera observes a young Chinese man walking through a quiet apartment corridor. The lens is wide, the lighting is ordinary, and the man does not look at the camera.

本模型为 SFW 概念模型。复杂动作可能增加倒影错误;如果镜面结构不稳定,建议先使用简单站姿和清晰的完整镜面。生成双人画面时,可明确写出两人拥有不同的脸、发型和衣服。

English description:

This is a realistic male concept LoRA designed for mirror-reflection compositions and fixed CCTV perspectives. The training set contains 200 individual images: 100 mirror scenes and 100 surveillance-camera scenes, with both solo and two-person compositions.

The mirror section focuses on compositions where the real subject and his reflection appear together. It covers full-length apartment mirrors, bathroom mirrors, wardrobe mirrors, hotel mirrors, and gym mirrors. It works well for natural standing poses, adjusting clothes, casual conversations, and two-person mirror scenes.

The CCTV section focuses on fixed high-angle cameras, wide-angle perspective, and unposed surveillance-style framing. Common environments include corridors, elevators, stairwells, parking garages, shop entrances, and apartment lobbies.

Mirror trigger: mirrormaleconcept

CCTV trigger: cctvmaleconcept

Use the two triggers separately. Using both triggers in the same image is not recommended. A LoRA weight around 0.75 is a good starting point.

Mirror prompt example:

mirrormaleconcept
Caption: A young Chinese man stands naturally in front of a full-length apartment mirror. His reflection matches his face, clothes, pose and limbs. The room is softly lit and realistic, with no phone and no additional person.

CCTV prompt example:

cctvmaleconcept
Caption: A fixed high-angle security camera observes a young Chinese man walking through a quiet apartment corridor. The lens is wide, the lighting is ordinary, and the man does not look at the camera.

This is an SFW concept model. Complicated poses may produce less consistent reflections. Start with simple poses and a clearly visible mirror. For two-person images, describe the subjects as having different faces, hairstyles, and clothes.