why soooo SLOW? 30x slower than normal sdxl loras

[fritol]

when i run SDXL it takes few secs per image
with this pipeline it takes OVER a MINUTE per image??!

[maszhongming]

Hi, thank you for your interest in our work.

For the inference speed, LoRA Composite does require more processing time. Typically, the time increase should be linear, so if you're combining k LoRAs, you'd expect it to take about k times longer. The 30x slower performance you're experiencing seems unusual. Could you share more about your experiments? Additionally, LoRA Switch should have comparable inference speed to LoRA Merge.