How to make this video clearer?

[poisondeathray]

"CUDA is not available"

Do you have Nvidia card with drivers updated ?

Do you have the dependencies installed ?
https://github.com/HolyWu/vs-realesrgan

NumPy , PyTorch with cuda ?

[kienkzz]

"CUDA is not available"

Do you have Nvidia card with drivers updated ?

Do you have the dependencies installed ?
https://github.com/HolyWu/vs-realesrgan

NumPy , PyTorch with cuda ?

No, I only have Intel HD Graphic 530

Numpy and pytorch have installed

[poisondeathray]

I think you can use device_type="cpu", but it will be slower

https://github.com/HolyWu/vs-realesrgan/blob/master/vsrealesrgan/__init__.py

device_type: Device type on which the tensor is allocated. Must be 'cuda' or 'cpu'.

Many anime have duplicate frames; one way to speed it up is to decimate the duplicates (VFR), instead of upscaling duplicates

[ridibunda]

The problem is that I don't know which script to use because those are too many, so I'm here to ask you guys

I have never done an "anime" encoding.

I said "search on the web", because I am sure you will find the exact script you need. Doom9 guys are the real experts.

[lollo]

I said "search on the web", because I am sure you will find the exact script you need. Doom9 guys are the real experts.

Some Doom9 experts post here too, and poisondeathray is one of them. If he's helping, there is no need to search elsewhere

[Selur]

Since RealESRFAN using the cpu usually is way to slow to be useful, another approach would be to use glsl filters (ignoring intro&outro) to increase the sharpness a bit.
(using Anime4Ks Darken, DeBlur and adaptive sharpening here)

# Imports
import vapoursynth as vs
# getting Vapoursynth core
core = vs.core
# Loading Plugins
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libvs_placebo.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SourceFilter/LSmashSource/vslsmashsource.dll")
# source: 'E:\clips\[NanakoRaws] Delicious Party♡Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4'
# current color space: YUV420P8, bit depth: 8, resolution: 1920x1080, fps: 29.97, color matrix: 709, yuv luminance scale: limited, scanorder: progressive
# Loading E:\clips\[NanakoRaws] Delicious Party♡Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4 using LibavSMASHSource
clip = core.lsmas.LibavSMASHSource(source="E:/clips/[NanakoRaws] Delicious Party♡Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4")
# making sure input color matrix is set as 709
clip = core.resize.Bicubic(clip, matrix_in_s="709",range_s="limited")
# making sure frame rate is set to 29.970
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# adjusting color space from YUV420P8 to YUV444P16 for vsGLSLAdaptiveSharpen
clip = core.resize.Bicubic(clip=clip, format=vs.YUV444P16, range_s="limited")
with open("I:/Hybrid/64bit/vsfilters/GLSL/parameterized/adaptive-sharpen.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define curve_height    1.0', '#define curve_height    1.0000')
glsl = glsl.replace('#define anime_mode      false', '#define anime_mode    true')
glsl = glsl.replace('#define overshoot_ctrl  false', '#define overshoot_ctrl  false')
glsl = glsl.replace('#define video_level_out false', '#define video_level_out false')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
with open("I:/Hybrid/64bit/vsfilters/GLSL/parameterized/Anime4K_Deblur_Original.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define REFINE_STRENGTH 1.0', '#define REFINE_STRENGTH 100')
glsl = glsl.replace('#define REFINE_BIAS 0.0', '#define REFINE_BIAS 0')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
with open("I:/Hybrid/64bit/vsfilters/GLSL/parameterized/Anime4K_Darken_HQ.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define STRENGTH 1.5', '#define STRENGTH 3')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
# adjusting output color from:  to YUV420P10 for x265Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, range_s="limited")
# set output frame rate to 29.970fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
# Output
clip.set_output()

(bonus: glsl filters are usually fast and only require opengl compatible cards)

Cu Selur

[kienkzz]

Since RealESRFAN using the cpu usually is way to slow to be useful, another approach would be to use glsl filters (ignoring intro&outro) to increase the sharpness a bit.
(using Anime4Ks Darken, DeBlur and adaptive sharpening here)

# Imports
import vapoursynth as vs
# getting Vapoursynth core
core = vs.core
# Loading Plugins
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libvs_placebo.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SourceFilter/LSmashSource/vslsmashsource.dll")
# source: 'E:\clips\[NanakoRaws] Delicious Party♡Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4'
# current color space: YUV420P8, bit depth: 8, resolution: 1920x1080, fps: 29.97, color matrix: 709, yuv luminance scale: limited, scanorder: progressive
# Loading E:\clips\[NanakoRaws] Delicious Party♡Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4 using LibavSMASHSource
clip = core.lsmas.LibavSMASHSource(source="E:/clips/[NanakoRaws] Delicious Party♡Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4")
# making sure input color matrix is set as 709
clip = core.resize.Bicubic(clip, matrix_in_s="709",range_s="limited")
# making sure frame rate is set to 29.970
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# adjusting color space from YUV420P8 to YUV444P16 for vsGLSLAdaptiveSharpen
clip = core.resize.Bicubic(clip=clip, format=vs.YUV444P16, range_s="limited")
with open("I:/Hybrid/64bit/vsfilters/GLSL/parameterized/adaptive-sharpen.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define curve_height    1.0', '#define curve_height    1.0000')
glsl = glsl.replace('#define anime_mode      false', '#define anime_mode    true')
glsl = glsl.replace('#define overshoot_ctrl  false', '#define overshoot_ctrl  false')
glsl = glsl.replace('#define video_level_out false', '#define video_level_out false')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
with open("I:/Hybrid/64bit/vsfilters/GLSL/parameterized/Anime4K_Deblur_Original.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define REFINE_STRENGTH 1.0', '#define REFINE_STRENGTH 100')
glsl = glsl.replace('#define REFINE_BIAS 0.0', '#define REFINE_BIAS 0')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
with open("I:/Hybrid/64bit/vsfilters/GLSL/parameterized/Anime4K_Darken_HQ.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define STRENGTH 1.5', '#define STRENGTH 3')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
# adjusting output color from:  to YUV420P10 for x265Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, range_s="limited")
# set output frame rate to 29.970fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
# Output
clip.set_output()

(bonus: glsl filters are usually fast and only require opengl compatible cards)

Cu Selur

I have tried this but...

[Attachment 63327 - Click to enlarge]

[poisondeathray]

The glsl shaders for placebo.Shader have to be in mpv compatible syntax

The path that you're using glsl-shaders-master/sharpen... suggests to me they were from a collection of shaders, but not in mpv format , probably this one
https://github.com/libretro/glsl-shaders

Use this one for adaptive-sharpen.glsl
https://gist.github.com/HolyWu/72fbe92c2b4700243adb502526a88591

[kienkzz]

The glsl shaders for placebo.Shader have to be in mpv compatible syntax

The path that you're using glsl-shaders-master/sharpen... suggests to me they were from a collection of shaders, but not in mpv format , probably this one
https://github.com/libretro/glsl-shaders

Use this one for adaptive-sharpen.glsl
https://gist.github.com/HolyWu/72fbe92c2b4700243adb502526a88591

The green video:


[Attachment 63332 - Click to enlarge]

# Imports
import vapoursynth as vs
# getting Vapoursynth core
core = vs.core
# Loading Plugins
core.std.LoadPlugin(path=r"E:\Video\Script\libvs_placebo.dll")
core.std.LoadPlugin(path=r"E:\Video\Script\L-SMASH-Works_20210423\plugins64\LSMASHSource.dll")
# source: 'E:\Video\[NanakoRaws] Delicious Party Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4'
# current color space: YUV420P8, bit depth: 8, resolution: 1920x1080, fps: 29.97, color matrix: 709, yuv luminance scale: limited, scanorder: progressive
# Loading E:\Video\[NanakoRaws] Delicious Party Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4 using LibavSMASHSource
clip = core.lsmas.LibavSMASHSource(source="E:/Video/[NanakoRaws] Delicious Party Precure - 01 (WEB-DL 1920x1080 x264 AAC).mp4")
# making sure input color matrix is set as 709
clip = core.resize.Bicubic(clip, matrix_in_s="709",range_s="limited")
# making sure frame rate is set to 29.970
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# adjusting color space from YUV420P8 to YUV444P16 for vsGLSLAdaptiveSharpen
clip = core.resize.Bicubic(clip=clip, format=vs.YUV444P16, range_s="limited")
with open("E:/Video/Script/adaptive-sharpen.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define curve_height    1.0', '#define curve_height    1.0000')
glsl = glsl.replace('#define anime_mode      false', '#define anime_mode    true')
glsl = glsl.replace('#define overshoot_ctrl  false', '#define overshoot_ctrl  false')
glsl = glsl.replace('#define video_level_out false', '#define video_level_out false')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
with open("E:/Video/Script/Anime4K_v4.0/Anime4K_Deblur_Original.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define REFINE_STRENGTH 1.0', '#define REFINE_STRENGTH 100')
glsl = glsl.replace('#define REFINE_BIAS 0.0', '#define REFINE_BIAS 0')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
with open("E:/Video/Script/Anime4K_v4.0/Anime4K_Darken_HQ.glsl") as glslf:
  glsl = glslf.read()
glsl = glsl.replace('#define STRENGTH 1.5', '#define STRENGTH 3')
clip = core.placebo.Shader(clip=clip, shader_s=glsl, width=1920, height=1080)
# adjusting output color from:  to YUV420P10 for x265Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, range_s="limited")
# set output frame rate to 29.970fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
# Output
clip.set_output()

[poisondeathray]

"Failed creating GPU textures"

Not sure , that script works for me with those shaders

BTW you should decimate to 23.976 after the source filter , for any script version
clip = core.tivtc.TDecimate(clip)

That particular shader combination does not improve the aliasing (the broken / jaggy line problems) . I think Selur was using it to illustrate line darkening , deblur, sharpen filters

[Selur]

Yes, I mainly posted this to show an alternative route.
libplacebo should work, without cuda.
-> try updating your gpu drivers. (Windows defaults drivers may lack some features.)

Cu Selur