[2ndfire] Plawres Sanshirou [DVD][480p][AV1][10bit]

Source: R2 DVDs

Video: AV1 Main 720x540@10bit CRF 13 Audio: Japanese - 2.0 Opus 96kbit Subs: English - SRT/ASS/SSA [AnimeClipse] (removed unreadable fonts and increased size of some styling. Inconsistent)

VS:

def postprocess(n, f, clip, deinterlaced):
   if f.props['_Combed'] > 0:
      return deinterlaced
   else:
      return clip


matched_clip = core.vivtc.VFM(video, order=0,mode=0,cthresh=8,micmatch=2,mi=60,scthresh=8)
deinterlaced_clip = haf.QTGMC(matched_clip, Preset='very slow', TR0=2, TR1=2, TR2=1, TFF=True, SourceMatch=3).std.SelectEvery( cycle=2, offsets=0)
postprocessed_clip = core.std.FrameEval(matched_clip, functools.partial(postprocess, clip=matched_clip, deinterlaced=deinterlaced_clip), prop_src=matched_clip)
video = core.vivtc.VDecimate(postprocessed_clip)
video = core.fmtc.resample (video, css="444")
video = core.std.Crop(video,3,2,0,0) [Varies by Ep]
video = core.fmtc.bitdepth(video, bits=16)
video2 = core.knlm.KNLMeansCL(video, d=0, a=1, s=4, h=2.5, wmode=0, wref=1.0, device_type="gpu",channels="YUV")
video = core.fmtc.bitdepth(video2, bits=32)
video = core.fmtc.matrix(video,mat="601",col_fam=vs.RGB)
video = core.ort.Model(video,network_path="_2xGT-v2-evA-01.onnx",fp16 = True,provider = "DML")
video = core.fmtc.matrix(video,mat="601",col_fam=vs.YUV)
video = core.resize.Spline64(video,width=720,height=540, matrix_s="170m", format = vs.YUV420P10)
video2 = core.placebo.Shader(video2,shader="AiUpscale_HQ_2x_LineArt.glsl",width= video2.width * 2 , height = video2.height * 2)
video2 = core.resize.Spline64(video2,width=720,height=540, format = vs.YUV420P10)
video = core.std.Merge(video2,video, weight=.50)
video.set_output()

svt-av1-psy

ffmpeg -i [infile] -c:v libsvtav1 -crf 13 -preset 2 -svtav1-params tune=3:enable-overlays=0:scd=1:scm=2:enable-tf=0:enable-cdef=1:variance-boost-strength=0:sharpness=2:lookahead=120:pin=0:lp=6 -g 1440 -sn -an -threads 8 [outfile]

Grain Synth: VS (denoised 10s sample)

video = core.resize.Spline64(video, format = vs.YUV444P16)
video = core.knlm.KNLMeansCL(video, d=3, a=6, s=4, h=0.6, wmode=0, wref=1.0, device_type="gpu",channels="YUV")
video = core.resize.Spline64(video,format = vs.YUV420P10)

VS (raw 10s sample)

video = core.resize.Spline64(video,format = vs.YUV420P10)

grav1synth

grav1synth diff [raw] [denoised] --output [grain table]

Film Grain Table (manually selected average table):

filmgrn1
E 0 9223372036854775807  1 10956 1
	p 3 7 0 11 0 1 128 192 256 128 192 256
	sY 7  0 80 54 80 81 82 94 88 121 96 148 96 255 100
	sCb 3 0 39 81 40 255 44
	sCr 4 0 39 54 39 148 43 255 43
	cY 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11 -30 56
	cCb 1 0 0 0 0 0 1 0 1 0 1 1 0 1 1 1 1 0 0 0 1 13 -20 69 0
	cCr 0 0 1 0 1 0 0 0 1 1 -1 2 1 0 1 1 0 0 0 -1 1 14 -21 63 -1

grav1synth apply [encode] --grain [grain table] --output [outfile]