This LORA is really an experimental one. Although trained with SDXL base model, not all models can get an accurate representation of V4x variant of Megurine Luka, so only few of them managed to get it almost accurate.Strength settings should be at least between 0.7 and 1.0. Use refiner when necessary.Compatible with Automatic1111's and any other SD-based offline image generators. Use what's best for you. Working models that tested gives accurate results will be shown under suggested resources.

Megurine Luka V4x LORA

Second version, trained with RaemuXL_v30 and this is far better version compared to the first experimental release.

Files with SHA256: 1eb12bc42f0917f5f8c4b8aa55bec78af2033efca3542b508c6cdfee43193cca