An experiment about combining multiple LoRAs with a/DARE
Base image (SDXL):
Adding pixel_art, fluffy and vibrant dream with additive (usual):
Adding pixel_art, fluffy and vibrant dream with DARE (this repo):
An experiment about combining multiple LoRAs with DARE https://arxiv.org/pdf/2311.03099.pdf
See this workflow for an example assets/dare_lora.json
For example applicable LoRAs, see https://civitai.com/user/ntc/models or https://huggingface.co/ntc-ai
Open a ticket or message me on civit / huggingfaces / patreon / here.
Support me on patreon https://www.patreon.com/NTCAI and get access to hundreds of SDXL LoRAs (~8 mb each) that this technique works great on.