Trigger words optional use to enhance the effect.
Trained on a very colorful dataset, I've used weight 0.5-1 depending on model,
Noise offset
embeds used in preview images:
https://huggingface.co/Xynon/models/blob/main/experimentals/TI/bad-image-v2-39000.pt
https://civitai.com/models/5224/bad-artist-negative-embedding