CivChan SD1.5

LORA
Original


Updated:

Disclaimer: This is an attempt to recreate 🩷 CivChan ~uWu~ 🩷utilizing her training data in Kohya SS for use on Stable Diffusion 1.5.

Training Steps: 1,185

Epochs: 20

Activation text is: civchan

Most models will get her eyes and hair color right, but if that's not the case, you can add the following prompts: purple eyes, pink hair

In my experiments, the best Weight has been 0.9 to 1, YMMV depending on the checkpoints you use. For compatible checkpoints and example prompts, please check the sample images.

I'm not entirely satisfied with this version as the number of steps seems oriented to the illustrious base model, but if there's demand for it, I can try to improve it in the future by adding more images to the dataset and ramping up the number of steps to around 2k.

Version Detail

SD 1.5
<p>First version. Tested only with semirealistic and anime checkpoints, SD1.5 only. Tested with textual inversions. Tested detail tweaking LoRA which can be found on the sample images.</p>

Project Permissions

    Use Permissions

  • Use in TENSOR Online

  • As a online training base model on TENSOR

  • Use without crediting me

  • Share merges of this model

  • Use different permissions on merges

    Commercial Use

  • Sell generated contents

  • Use on generation services

  • Sell this model or merges

Related Posts