Autoplay
Run Locally
possession_low_v3.1_e13.safetensors
293 MB · SafeTensor · bf16
Downloads are served by Civitai. Some creators ask for a Civitai login first.
About This Model
WAN · LoRAs
Possession 附身 憑依 for Wan2.2 Low 3.1 By: tsfxlover Type: I2V & T2V Trigger: The following is a sample training set prompt format: > possession, enter. [please describe in detail the prompts related to external forces entering the host, such as the possession process by ghost, parasite, slime, mist, wisp, etc.]. > possession, manifest. [please describe the specific prompts indicating sudden changes occurring on the host, such as aura, electricity, etc.]. > possession, transfer. [please describe the prompts specifically for the transfer process from A to B]. > possession, leave. [please describe the prompts specifically for leaving the host].
Creator notes
Guidance:
-
This lora can be used with t2v or i2v.
-
In the "About this version" section you can find a list of the most frequent nouns and verbs in the training set.
-
If the possession effect does not work, consider increasing the high-noise LORA; if there is contamination to other detail-enhancing LORAs, try lowering the low-noise LORA.
Preview Image Prompt:
I will mark prompts in the relevant preview images for your reference.
Version:
-
v3: Major update, complete replacement of the training set, featuring diverse abilities including possession forms (ghost, parasite, slime, mist, wisp, aura, electricity, etc.), transferring hosts, and leaving hosts. The following is a sample training set prompt format:
-
possession, enter. [please describe in detail the prompts related to external forces entering the host, such as the possession process by ghost, parasite, slime, mist, wisp, etc.].
-
possession, manifest. [please describe the specific prompts indicating sudden changes occurring on the host, such as aura, electricity, etc.].
-
possession, transfer. [please describe the prompts specifically for the transfer process from A to B].
-
possession, leave. [please describe the prompts specifically for leaving the host].
-
v2: Expand and clean the dataset, then adjust the training parameters and retrain with the help of Gemini Pro. Someone can possess a target in physical or ghost form.
-
v1: Attempted to train my first LORA, using Smiles_wx porting trainer, the Windows version of diffusion-pipe, from
说明:
-
这个lora可使用在t2v或是i2v。
-
在"About this version"部分,你可以看到训练集中频率最高的名词与动词清单。
-
如果没有附身效果,可以考虑增加高噪LORA;如果对其他细节增强型 LORA 造成污染,请尝试降低低噪LORA。
预览图提示词:
我会在各预览图中标记使用的提示词,作为参考使用。
版本:
-
v3:重大更新,完全更换训练集,此版本具有附身型态(ghost、parasite、slime、mist、wisp、aura、electricity等)、转移宿主与离开(leave)宿主等多样能力。可参考的训练集提示词格式如下:
-
possession, enter. [请具体描述提示词关于外界进入宿主。例如鬼魂ghost、寄生虫parasite、烟雾mist等等的附身过程]
-
possession, manifest. [请具体描述宿主身上突然发生一些变化的提示词。例如aura、electricity等]
-
possession, transfer. [请具体描述从A转移至B的转移过程提示词]
-
possession, leave. [请具体描述从宿主脱离的提示词]
-
v2:扩展并清理资料集,然后借助 Gemini Pro调整训练参数并重新训练。可以原形或鬼魂型态附身目标。
-
v1:尝试使用 Smiles_wx 整合的训练器(Windows 版的 diffusion-pipe)训练我的第一个 LORA,该训练器来自
Trigger words
possession
SHA256
A355480D43EB7F200A56D5FC83A94F6B00A944ADEB955DCE11C6C27DEEDCED58
ModelTrends.ai Model ID
#26874
Flag content
Example renders
Prompts and settings shared by the people who made these renders on Civitai. Pick a render to see how it was made.
Settings
Size
864x480
Prompt
possession, enter. The man rushes forward from the side and merges into the woman's body, and then disappears.
Negative prompt
色调艳丽,过曝,静态,细节模糊不清,字幕,风格,作品,画作,画面,静止,整体发灰,最差质量,低质量,JPEG压缩残留,丑陋的,残缺的,多余的手指,画得不好的手部,画得不好的脸部,畸形的,毁容的,形态畸形的肢体,手指融合,静止不动的画面,杂乱的背景,三条腿,背景人很多,倒着走
Similar Models
AI Models for Image, Video & Text Generation
possession low31 wan is a LoRA for the Wan family listed on ModelTrends.ai, a read-only catalog of open source AI image, video and text models. Compare it with other Wan models, check its heat score to see how it is trending, and open it on PirateDiffusion or BitVector to try it.
Browse by Model Family
Text / LLM Models
411
Anima Models
509
Chroma Models
14
Flux Models
1,071
Flux 2 / Klein Models
1,237
MiniMax H3 Models
331
Hunyuan Models
197
Ideogram Models
20
Krea2 Models
674
Qwen Models
17
Qwen2 Models
36
Ltx2 Models
150
Stable Diffusion 1.5 Models
6,465
Stable Diffusion XL 1.0 Models
15,432
Zimage Models
804
Wan Models
919

Model
Trends
.ai
© ModelTrends.ai
|
Made in Japan
|
© 2026
