Experimental Selfie LoRA (SDXL)

This is an experimental model, trained using 1000 very low-quality images. Consequently, output results are likely to be poor. Use is at your own discretion.

The goal of this LoRA is to produce natural-looking selfies.
Post your pictures with this LoRA below.

The three sample images currently in the gallery were generated automatically by the Civitai system without the recommended settings (no LCM&TurboMix).

For optimized and faster results, I highly recommend using this LoRA in conjunction with the LCM&TurboMix LoRA.
Recommended Settings (with LCM&TurboMix):
Sampler: DPM++ SDE Karras
Steps: 8
CFG Scale: 2.0
LCM&TurboMix Strength: 1.0