Laksha - V1
精選圖片
推薦參數
samplers
dpmpp_2m_sde_gpu
steps
25
提示
Use simpler prompts for best results.
Between 0.20 and 0.60 LoRA weight is best.
Combine this model with other LoRAs.
On CivitAI site, use checkpoints other than vanilla SDXL.
With ComfyUI, vanilla SDXL works fine.
Laksha tends to add a "fantasy sheen" to everything.
This model was trained on pictures of elves and goblins, so you might get long pointed ears.
It does well with "digital watercolor," "comic style" and "chiaroscuro." Try to use simpler prompts. Between 0.20 and 0.60 LoRA weight is best. This model is best combined with other LoRAs.
When using it on the CivitAI site, I would advise using other checkpoints than vanilla SDXL, but with ComfyUI vanilla does fine.
