2x_ModernSpanimationV1
2x_ModernSpanimationV1
Scale: 2
Architecture: SPAN
Links: Github Release
Author: TNTwise
License: MIT
Purpose: Compression Removal, Deblur, Denoise, JPEG, Restoration
Subject: Anime
Input Type: Images
Date: 04.06.2024
Size: 64
Dataset: Modern Animation V2
Dataset Size: 2200
OTF (on the fly augmentations): No
Pretrained Model: None
Iterations: 100,000
Batch Size: 3-12
GT Size: 256-512
Description: Upscale modern animation images/videos. Compression, blur, and un-sharpening degradations were used on the dataset. First model ever trained and I think it came out pretty well.
Showcase: https://slow.pics/s/NfSOp6L6