Skip to content

2xHFA2kAVCCompact

Compare
Choose a tag to compare
@Phhofm Phhofm released this 24 Apr 15:22
· 6 commits to main since this release
c664eb5

Here two (relatively) fast anime sr models that are meant to handle AVC (h264) compression (as can be found on some anime from the web). Did an EDSR M (network settings in the textfile) and a Compact (SRVGGNet) one.

Name: 2xHFA2kAVCEDSR_M
Author: Philip Hofmann
Release Date: 18.06.2023
License: CC BY 4.0
Network: EDSR M
Scale: 2
Purpose: Fast 2x anime upscaling model that handles AVC (h264) degradation
Iterations: 186,000
batch_size: 2-5
HR_size: 384
Epoch: 317 (require iter number per epoch: 1284)
Dataset: HFA2k_h264
Number of train images: 2568
OTF Training: No
Pretrained_Model_G: None

Description: A 2x EDSR M anime upscale model that handles AVC (h264) degradation. Applied h264 crf 20-28 degradation and bicubic, bilinear, box and lanczos downsampling on the HFA2k dataset with Kim's dataset destroyer.


Name: 2xHFA2kAVCCompact
Author: Philip Hofmann
Release Date: 18.06.2023
License: CC BY 4.0
Network: SRVGGNet
Scale: 2
Purpose: Fast 2x anime upscaling model that handles AVC (h264) degradation
Iterations: 111,000
batch_size: 4
HR_size: 256
Dataset: HFA2k_h264
Number of train images: 2568
OTF Training: No
Pretrained_Model_G: None

Description: A 2x Compact anime upscale model that handles AVC (h264) degradation. Applied h264 crf 20-28 degradation together with bicubic, bilinear, box and lanczos downsampling on the HFA2k dataset with Kim's dataset destroyer.

image
324388155-7cd72469-4631-4f20-b5af-227692b1879a