0.4.0

What's New

1. Set FFMPEG to only process 1 frame at the time and map it to the video instead of dissecting it, removing the decoding overhead for audio extraction during dissection phase saving at least 1 decode cycle time.
2. Added Upscayl's newest Ultrasharp AI model.
3. Added benchmark script and results for measuring FFMPEG commands' overall performances.
4. Prefix AI models' creator name for all AI models in the effort of granular ID and attribution.

Supported Models

For this version, the packaged models are:

ID        : upscayl-realesrgan-x4plus-anime
max scale : any
Purpose   : Upscayl's RealESRGAN for Anime Version 4+ Upscaling

ID        : upscayl-realesrgan-x4plus
max scale : any
Purpose   : Upscayl's RealESRGAN for General Image Version 4+ Upscaling

ID        : upscayl-remacri
max scale : any
Purpose   : Upscayl's EMACRI for General Image Upscaling

ID        : upscayl-ultramix_balanced
max scale : any
Purpose   : Upscayl's Ultramix balanced for General Image Upscaling

ID        : upscayl-ultrasharp
max scale : any
Purpose   : Upscayl's Ultrasharp for General Image Upscaling

ID        : upscayl-ultrasharp2
max scale : any
Purpose   : Upscayl's Ultrasharp V2 for General Image Upscaling

ID        : xintao-realesr-animevideov3-x2
max scale : 2
Purpose   : XinTao's Animated Video Version 3 x2 Upscaling

ID        : xintao-realesr-animevideov3-x3
max scale : 3
Purpose   : XinTao's Animated Video Version 3 x3 Upscaling

ID        : xintao-realesr-animevideov3-x4
max scale : 4
Purpose   : XinTao's Animated Video Version 3 x4 Upscaling

ID        : xintao-realesrgan-x4plus-anime
max scale : any
Purpose   : XinTao's RealESRGAN for Anime Version 4+ Upscaling

ID        : xintao-realesrgan-x4plus
max scale : any
Purpose   : XinTao's RealESRGAN for General Image Version 4+ Upscaling