README.md

metadata

license: apache-2.0
tags:
  - lora
  - style-transfer
  - somboy
  - wan2.2t2v
  - anime
  - 2.5d
  - text-to-video
  - video-generation

SomBoy Style LoRA for Wan2.2-T2V

Style LoRA trained to transfer the SomBoy v2.5 art style (2.5D anime aesthetic from CivitAI) to Wan2.2-T2V.

Trigger Word

somboy_2d5

Usage

Use the trigger word somboy_2d5 in your prompt to activate the style.

Example prompts:

• 1girl, portrait, long black hair, soft lighting, in the style of somboy_2d5
• 1boy, casual outfit, modern aesthetic, in the style of somboy_2d5

Sample Videos

Sample 1: 📹 sample_01.mp4

Sample 2: 📹 sample_02.mp4

Sample 3: 📹 sample_03.mp4

Sample 4: 📹 sample_04.mp4

Sample 5: 📹 sample_05.mp4

Sample 6: 📹 sample_06.mp4

Sample 7: 📹 sample_07.mp4

Sample 8: 📹 sample_08.mp4

Sample 9: 📹 sample_09.mp4

Sample 10: 📹 sample_10.mp4

Training Details

• Base Model: Wan2.2-T2V
• Training Steps: 750
• Learning Rate: 1e-4
• Network Dim: 64
• Dataset: 60 images generated with SomBoy v2.5 checkpoint

License

Apache 2.0