KandinskyDreamsHaper
Accessible version of KandinskyDreams trained using Dream Shaper 8.
This LoRa was trained using the same dataset used for the original SDXL version. I'm offering it up here for use by those without access to SDXL models.
Like the original KandinskyDreams, this LoRa relies only on the styles of deceased artists.
Let me know what you think and win a free 🥪.
Usage guide
This LoRa version pulls strongly towards buildings and to a lesser extent, crowd scenes: Consider negative prompts to counteract.
Even at a weight of 0.6, this LoRa will swamp most prompts; 0.3 suffices for most applications.
Optimal prompting keywords
We recommend using these trigger words when prompting KandinskyDreamsHaper to get the best results:
- horror
- nightmare
- surreal
- gloom
- doom
- fantasy
No comments yet.
No comments yet