Best settings using Virtual Dub to correct dark VHS capture

[DMS]

I just checked the properties and it says the Source Video info is 720x288

[Attachment 60878 - Click to enlarge]

[Selur]

Just for the fun of it I tried what the BasicVSR++ and it's Models:
3 = NTIRE 2021 Quality enhancement of heavily compressed videos Challenge - Track 1
4 = NTIRE 2021 Quality enhancement of heavily compressed videos Challenge - Track 2
5 = NTIRE 2021 Quality enhancement of heavily compressed videos Challenge - Track 3
would do to the first Sample.avi source.
I used:

# Imports
import os
import sys
import ctypes
# Loading Support Files
Dllref = ctypes.windll.LoadLibrary("I:/Hybrid/64bit/vsfilters/Support/libfftw3f-3.dll")
import vapoursynth as vs
# getting Vapoursynth core
core = vs.core
# Import scripts folder
scriptPath = 'I:/Hybrid/64bit/vsscripts'
sys.path.insert(0, os.path.abspath(scriptPath))
# Loading Plugins
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SharpenFilter/CAS/CAS.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/DenoiseFilter/CTMF/CTMF.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/DenoiseFilter/HQDN3D/libhqdn3d.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/DenoiseFilter/NEO_FFT3DFilter/neo-fft3d.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libtemporalmedian.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libmvtools.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SourceFilter/LSmashSource/vslsmashsource.dll")
# Import scripts
import SpotLess
import adjust
import havsfunc
# source: 'C:\Users\Selur\Desktop\Sample.avi'
# current color space: YUV420P8, bit depth: 8, resolution: 720x576, fps: 50, color matrix: 470bg, yuv luminance scale: limited, scanorder: progressive
# Loading C:\Users\Selur\Desktop\Sample.avi using LWLibavSource
clip = core.lsmas.LWLibavSource(source="C:/Users/Selur/Desktop/Sample.avi", format="YUV420P8", cache=0, fpsnum=50, prefer_hw=0)
# making sure input color matrix is set as 470bg
clip = core.resize.Bicubic(clip, matrix_in_s="470bg",range_s="limited")
# making sure frame rate is set to 50
clip = core.std.AssumeFPS(clip=clip, fpsnum=50, fpsden=1)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# cropping the video to 688x568
clip = core.std.CropRel(clip=clip, left=20, right=12, top=0, bottom=8)
# adjusting frame count and rate with sRestore
clip = havsfunc.srestore(source=clip, frate=25.000, omode=6, speed=9, thresh=16, mode=2)
# adjusting color space from YUV420P8 to RGB24 for vsLevels
clip = core.resize.Bicubic(clip=clip, format=vs.RGB24, matrix_in_s="470bg", range_s="limited")
# Color Adjustment using Levels on RGB24 (8 bit)
clip = core.std.Levels(clip=clip, min_in=16, max_in=255, min_out=0, max_out=235, gamma=1.25)
# adjusting color space from RGB24 to YUV444P8 for vsTweak
clip = core.resize.Bicubic(clip=clip, format=vs.YUV444P8, matrix_s="470bg", range_s="limited")
# Color Adjustment
clip = adjust.Tweak(clip=clip, hue=0.00, sat=0.00, cont=1.00, coring=True)
clip = SpotLess.SpotLess(clip=clip, chroma=False, radT=3)
# removing grain using TemporalDegrain
clip = havsfunc.TemporalDegrain(inpClip=clip)
# adjusting color space from YUV444P8 to RGBS for vsBasicVSRPPFilter
clip = core.resize.Bicubic(clip=clip, format=vs.RGBS, matrix_in_s="470bg", range_s="limited")
# Quality enhancement using BasicVSR++
from vsbasicvsrpp import BasicVSRPP
clip = BasicVSRPP(clip=clip, model=3, tile_x=352, tile_y=296) # changed the model
# contrast sharpening using CAS
clip = core.cas.CAS(clip=clip, sharpness=0.800)
# adjusting output color from: RGBS to YUV420P10 for x265Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, matrix_s="470bg", range_s="limited")
# set output frame rate to 25.000fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=25, fpsden=1)
# Output
clip.set_output()

See attachments for x265 crf18 encodes.

Also had a quick test with some mild filtering at TEST2.avi.

# cropping the video to 680x570
clip = core.std.CropRel(clip=clip, left=20, right=20, top=0, bottom=6)
# Color Adjustment
clip = adjust.Tweak(clip=clip, hue=0.00, sat=0.00, cont=1.00, coring=False)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = SpotLess.SpotLess(clip=clip, chroma=False, radT=1)
# applying anti aliasing using santiag
clip = core.std.FlipVertical(clip)
clip = havsfunc.santiag(c=clip, nns=3, pscrn=1, opencl=True)

Cu Selur

Ps.: I agree ideally the capturing should be done properly

[DMS]

Ok, i have sorted my half size capture, i re-installed the drivers for the dazzle, and it is now capturing in a better resolution. i have attached another capture, this time it is shorter as i think it is using more data because of more resolution. I am a lot happier now, i notice a massive difference when viewing the dvd screen settings via the capture device.
[Attachment 60888 - Click to enlarge]

[DB83]

Yes. You have some 33% more video bitrate now and the picture is not showing the aliasing. I guess that the chief culprit was the Dazzle.


Does your Panny offer some noise reduction since that appears to be one concern right now. TBH I have never seen a VHS capture with so much noise. Maybe I have been lucky in that regard.

[DB83]

Well it may not be common knowledge in this forum but I do like a glass (or 6) of the red grape.


I had an offer to get 9 Will Hay filums on 5 vhs for the staggering cost of £7 + PP. At that point the fingers did the talking


Should get them next week (or earlier). I am curious if I can improve with simple raw captures. Watch this space.

[Selur]

Yup, that one looks better, attached a version using similar light filtering as before:

# Imports
import os
import sys
import vapoursynth as vs
# getting Vapoursynth core
core = vs.core
# Import scripts folder
scriptPath = 'I:/Hybrid/64bit/vsscripts'
sys.path.insert(0, os.path.abspath(scriptPath))
# Loading Plugins
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SharpenFilter/AWarpSharp2/libawarpsharp2.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/DenoiseFilter/CTMF/CTMF.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SharpenFilter/CAS/CAS.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libsangnom.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/fmtconv.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/EEDI3.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/EEDI2.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/ResizeFilter/nnedi3/vsznedi3.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libtemporalmedian.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libmvtools.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SourceFilter/LSmashSource/vslsmashsource.dll")
# Import scripts
import havsfunc
import mvsfunc
import SpotLess
import lostfunc
import adjust
# source: 'C:\Users\Selur\Desktop\AAP TEST2.avi'
# current color space: YUV422P8, bit depth: 8, resolution: 720x576, fps: 25, color matrix: 470bg, yuv luminance scale: limited, scanorder: progressive
# Loading C:\Users\Selur\Desktop\AAP TEST2.avi using LWLibavSource
clip = core.lsmas.LWLibavSource(source="C:/Users/Selur/Desktop/AAP TEST2.avi", format="YUV422P8", cache=0, prefer_hw=0)
# making sure input color matrix is set as 470bg
clip = core.resize.Bicubic(clip, matrix_in_s="470bg",range_s="limited")
# making sure frame rate is set to 25
clip = core.std.AssumeFPS(clip=clip, fpsnum=25, fpsden=1)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# Color Adjustment
clip = adjust.Tweak(clip=clip, hue=0.00, sat=0.00, cont=1.00, coring=True)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = SpotLess.SpotLess(clip=clip, chroma=False, radT=1)
# cropping the video to 686x572
clip = core.std.CropRel(clip=clip, left=18, right=16, top=0, bottom=4)
# applying anti aliasing using santiag
clip = havsfunc.santiag(c=clip, pscrn=2)
# contrast sharpening using CAS
clip = core.cas.CAS(clip=clip, sharpness=0.900)
# applying dehalo using YAHR
clip = havsfunc.YAHR(clip)
# adjusting output color from: YUV422P8 to YUV420P10 for x265Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, range_s="limited")
# set output frame rate to 25.000fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=25, fpsden=1)
# Output
clip.set_output()

Cu Selur

[DMS]

Yes. You have some 33% more video bitrate now and the picture is not showing the aliasing. I guess that the chief culprit was the Dazzle.


Does your Panny offer some noise reduction since that appears to be one concern right now. TBH I have never seen a VHS capture with so much noise. Maybe I have been lucky in that regard.

It does have a NR option but it is off.

[DMS]

Well it may not be common knowledge in this forum but I do like a glass (or 6) of the red grape.


I had an offer to get 9 Will Hay filums on 5 vhs for the staggering cost of £7 + PP. At that point the fingers did the talking


Should get them next week (or earlier). I am curious if I can improve with simple raw captures. Watch this space.

Im sure you will enjoy therm, Oh, Mr. Porter! is a classic.

[DMS]

Yup, that one looks better, attached a version using similar light filtering as before:

# Imports
import os
import sys
import vapoursynth as vs
# getting Vapoursynth core
core = vs.core
# Import scripts folder
scriptPath = 'I:/Hybrid/64bit/vsscripts'
sys.path.insert(0, os.path.abspath(scriptPath))
# Loading Plugins
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SharpenFilter/AWarpSharp2/libawarpsharp2.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/DenoiseFilter/CTMF/CTMF.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SharpenFilter/CAS/CAS.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libsangnom.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/fmtconv.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/EEDI3.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/EEDI2.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/ResizeFilter/nnedi3/vsznedi3.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libtemporalmedian.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/Support/libmvtools.dll")
core.std.LoadPlugin(path="I:/Hybrid/64bit/vsfilters/SourceFilter/LSmashSource/vslsmashsource.dll")
# Import scripts
import havsfunc
import mvsfunc
import SpotLess
import lostfunc
import adjust
# source: 'C:\Users\Selur\Desktop\AAP TEST2.avi'
# current color space: YUV422P8, bit depth: 8, resolution: 720x576, fps: 25, color matrix: 470bg, yuv luminance scale: limited, scanorder: progressive
# Loading C:\Users\Selur\Desktop\AAP TEST2.avi using LWLibavSource
clip = core.lsmas.LWLibavSource(source="C:/Users/Selur/Desktop/AAP TEST2.avi", format="YUV422P8", cache=0, prefer_hw=0)
# making sure input color matrix is set as 470bg
clip = core.resize.Bicubic(clip, matrix_in_s="470bg",range_s="limited")
# making sure frame rate is set to 25
clip = core.std.AssumeFPS(clip=clip, fpsnum=25, fpsden=1)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# Color Adjustment
clip = adjust.Tweak(clip=clip, hue=0.00, sat=0.00, cont=1.00, coring=True)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = lostfunc.DeSpot(o=clip)
clip = SpotLess.SpotLess(clip=clip, chroma=False, radT=1)
# cropping the video to 686x572
clip = core.std.CropRel(clip=clip, left=18, right=16, top=0, bottom=4)
# applying anti aliasing using santiag
clip = havsfunc.santiag(c=clip, pscrn=2)
# contrast sharpening using CAS
clip = core.cas.CAS(clip=clip, sharpness=0.900)
# applying dehalo using YAHR
clip = havsfunc.YAHR(clip)
# adjusting output color from: YUV422P8 to YUV420P10 for x265Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, range_s="limited")
# set output frame rate to 25.000fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=25, fpsden=1)
# Output
clip.set_output()

Cu Selur

Thats looks really good, i may have a try at replicating it using your code, i do have your hybrid software but not really tried it it out yet, im not up to speed with AVisynth scripts and i think Hybrid will help me understand the filters a bit better.

Cheers

DMS

[DB83]

Well it may not be common knowledge in this forum but I do like a glass (or 6) of the red grape.


I had an offer to get 9 Will Hay filums on 5 vhs for the staggering cost of £7 + PP. At that point the fingers did the talking


Should get them next week (or earlier). I am curious if I can improve with simple raw captures. Watch this space.

Im sure you will enjoy therm, Oh, Mr. Porter! is a classic.

I'm sure I will. Would probably have seen all the well-known ones at one time or another. Like I said, this is more of an exercise since I could have acquired a dvd set (also used) for approx. the same price or a new one for just a little more. I am beginning to think now that the youtube version is also from vhs so quite curious about what I eventually see.

[DMS]

Well it may not be common knowledge in this forum but I do like a glass (or 6) of the red grape.


I had an offer to get 9 Will Hay filums on 5 vhs for the staggering cost of £7 + PP. At that point the fingers did the talking


Should get them next week (or earlier). I am curious if I can improve with simple raw captures. Watch this space.

Im sure you will enjoy therm, Oh, Mr. Porter! is a classic.

I'm sure I will. Would probably have seen all the well-known ones at one time or another. Like I said, this is more of an exercise since I could have acquired a dvd set (also used) for approx. the same price or a new one for just a little more. I am beginning to think now that the youtube version is also from vhs so quite curious about what I eventually see.

If i remember rightly the DVD releases were just the same as the VHS, they were not restored, the only restored Will Hay film available is 'The Goose Steps Out' which was released on DVD & Bluray. There were rumours of Oh Mr Porter being restored and released but nothing has surfaced yet.

[Selur]

im not up to speed with AVisynth scripts

just for clarification that's a Vapoursynth not Avisynth script

[DB83]

So my tapes arrived today (exc. service from the seller) and I could not wait.


Now what follows is only meant for comparison purposes and not a source for further noise reduction (not that I normally bother with that)


Two versions which join two extracts as posted earlier. The first (Cap001) is straight from the vcr (composite) in to my Hauppauge USB-Live2. The second (Cap002) is minimally filtered by diverting the signal from the vcr through an ADVC 300. Both were captured with the default WinTv7 software.


Comments about the comparison (and nothing else) are welcome.

[jagabo]

Why are you posting MPG files? Neither of those devices captures MPEG 2 video. So your files have been through processing which may have created some of the problems that can be seen in them.

About the only thing that can be said is that cap001.mpg is brighter and has crushed brights. cap002.mpg has less horizontal time base jitter. So cap002 wins on both accounts.

[DB83]

Nope.


Mpeg-2 is the default capture format for WinTV (and Hauppuage Capture which also supports the device). Win7 actually captures as a mpeg2 transport stream and there is an option to demux that (in a matter of seconds) to program stream. The ADVC was just used for 'Pass Through' purposes and I would have expected some improvement there.


This was not intended as a comparison between the two captures I uploaded but against those from the OP in terms of brightness and picture disturbance.

[jagabo]

Oh, sorry. I was thinking of the Hauppauge HDPVR 2 (which captures as AVC).

Something I noticed about the ADVC passthrough was that it creates lots of sporadic black dots, especially in dark areas.

[DMS]

i just did a comparison capture of the same footage using a direct composite connection from the vcr and the bitrate is higher than when i captured using an svideo connection via the dvd passthrough, what is the reason for this?


[Attachment 61115 - Click to enlarge]

[ajk]

The composite signal more noisy, so Lagarith can't manage to compress the data as well.

[DMS]

The composite signal more noisy, so Lagarith can't manage to compress the data as well.

I thought that might be the case, thank you.

[DMS]

One more thing, can someone explain why the Histogram in VirtualDub goes jaggy when i change the contrast by one position?

[jagabo]

One more thing, can someone explain why the Histogram in VirtualDub goes jaggy when i change the contrast by one position?

The video is stored as integers, both before and after the filter. When you increase the contrast you are multiplying those integers by some amount. Say for example you are multiplying by 2. Then pixels with brightness values of 1, 2, 3, 4, 5 become 2, 4, 6, 8, 10. Notice there is now a gap between each of the new values. And there's no original value which can give you at 3, 5, 7, or 9. That's what's happening. In your case the values are being multiplied by about 1.05.

[DMS]

One more thing, can someone explain why the Histogram in VirtualDub goes jaggy when i change the contrast by one position?

The video is stored as integers, both before and after the filter. When you increase the contrast you are multiplying those integers by some amount. Say for example you are multiplying by 2. Then pixels with brightness values of 1, 2, 3, 4, 5 become 2, 4, 6, 8, 10. Notice there is now a gap between each of the new values. And there's no original value which can give you at 3, 5, 7, or 9. That's what's happening. In your case the values are being multiplied by about 1.05.

Thanks Jagabo, so there’s nothing actually wrong it’s supposed to be like that?

[jagabo]

It's the expected result. Some more more advanced proc amp filters use dithering to "fill in" those gaps (less chance of posterization artifacts). I don't know if there are any for VirtualDub.

[DMS]

It's the expected result. Some more more advanced proc amp filters use dithering to "fill in" those gaps (less chance of posterization artifacts). I don't know if there are any for VirtualDub.

ok thanks again.

[DMS]

Just wondering if any of you would like to review the attached files and let me me know which one you think is the better of the 3?

Thanks in advance.

[Sharc]

Just wondering if any of you would like to review the attached files and let me me know which one you think is the better of the 3?

To my taste all 3 look oversharpened. Did you apply a sharpening filter?

[DB83]

I would probably be the last person to judge this since I am 'biased'


Out of the 3, I would choose 4.avi (and I await all to disagree with me - although I think my modest captures are lighter with little to no filtering (Hauppauge proc-amp not used) as far as brightness was concerned

[DMS]

Just wondering if any of you would like to review the attached files and let me me know which one you think is the better of the 3?

To my taste all 3 look oversharpened. Did you apply a sharpening filter?

Hi Sharc

No sharpening filter was used in the capture software and the proc amp sharpening was set to zero.

All 3 were captured as follows

VCR Scart > Panasonic EH50 svideo > Dazzle > Virtualdub2 lagarith.

Capture2 had no settings on in the Panasonic
Capture3 had NR and Combfilter on
Capture4 recorded to Panasonic HHD with NR and Combfilter on then captured by the dazzle.

I couldn’t tell any difference if I’m honest. I’m wondering if the Panasonic is sharpening it when it converts from composite to svideo?

[DMS]

I would probably be the last person to judge this since I am 'biased'


Out of the 3, I would choose 4.avi (and I await all to disagree with me - although I think my modest captures are lighter with little to no filtering (Hauppauge proc-amp not used) as far as brightness was concerned

I might up the brightness a bit see if that helps, details of the captures are in the previous post. I will continue to experiment, I’m not sure using this old film is helping, maybe I should try a newer clip in colour.

[jagabo]

Capture 2 has the most recoverable detail, capture 4 the least.