2xHFA2kAVCOmniSR

Name: 2xHFA2kAVCOmniSR
License: CC BY 4.0
Network: OmniSR
Scale: 2
Purpose: 2x anime upscaling model that handles AVC (h264) compression
Iterations: 175,000
batch_size: 2-5
HR_size: 128-192
Dataset: HFA2k_h264
Number of train images: 2568
OTF Training: No
Pretrained_Model_G: None

Description: The second released community trained model on the OmniSR network. Trained with multiscale discriminator to fix decreased output brightness occuring with OmniSR. 2x anime upscale that handles AVC (h264) compression since h264 crf 20-28 degradation together with bicubic, bilinear, box and lanczos downsampling has been applied on the HFA2k dataset with Kim's dataset destroyer.

Examples:
https://imgsli.com/MTg2ODcz
https://imgsli.com/MTg2ODc1

2xHFA2kAVCOmniSR_Sharp

Name: 2xHFA2kAVCOmniSR_Sharp
License: CC BY 4.0
Network: OmniSR
Scale: 2
Purpose: 2x sharp anime upscaling model that handles AVC (h264) compression
Iterations: 229,000
batch_size: 2-5
HR_size: 128-192
Dataset: HFA2k_h264
Number of train images: 2568
OTF Training: No
Pretrained_Model_G: None

Description: A sharper version of 2xHFA2kAVCOmniSR since it has been further trained with some additional USM applied. I recommend using 2xHFA2kAVCOmniSR in general, but if more sharpness is desired or you simply like these results better for your input source, you can of course also give this model a go :)