[Checkpoint] YOZORA (PonyXL)版本V1 (ID: 705324)

Support me: https://ko-fi.com/elevatedone ☕

Please post suggestions for future models here: https://civitai.com/articles/5921/suggestion-list

The early access can be removed (for everyone) by donating to my Ko-Fi (Say the model name in the message field).

---

As I couldn't let go of many SD 1.5 model's artstyles, I decided to retrain them for PonyXL.

This is the fourth iteration of retraining styles, but the first trained checkpoint.

Information:

• The model is based on Night Sky YOZORA by @YozoRaAru

• Base model for training used was T-ponynai3 v5.5 by @Tonade

• The model was trained at 1536x1536 (V1) and 1024x1024 V2

• The model was trained with around 250 pictures and around 3000 steps after batch size of 8 I believe (V1), and around 500 pictures and around 2500 steps after a batch of 12 (V2)

• V2 is here.

---

If you want to check the other retrains of mine, here you go:

• flat2DAnimerge (Original by @bigbeanboiler)

• Animeliner (LoRA) (Original by @Goofy_Ai)

• SD15Lorium (LoRA trained on data from 34 checkpoints, with triggers for all of them. It's kinda bad since it was trained with rank of 16 I believe)

描述:

训练词语: