sempai from magical sempai
suggested checkpoints : based64 , abyss orange mix , Abased
trained on 1024 pixel res images so high res fix is suggested
examples images generated with 50/50 merge of based64 and abyss orange mix :
https://civitai.com/models/118513/abased
Trained on Dataset constructed from stitched panning shots extracted from anime
Tools used for construction of dataset:
for extraction and stitching of panning shots-
https://github.com/bdiaz29/Moving-Shot-Extraction-
for rapid captioning of images