VideoHelp Forum
+ Reply to Thread
Page 2 of 4
FirstFirst 1 2 3 4 LastLast
Results 31 to 60 of 115
Thread
  1. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Is it possible to toggle time/date overlays on some of the captures ? That would be "best"
    I am unaware that for video8 itīs possible to toggle time /date overlays during capture , I thought that time and date were hard recorded on tape if there is any chance do disable the overlays I was very much interested to know how to do it.

    To avoid halos due to over sharpening probably I canīt do any better even in the captures, I noticed that when I select edit on the Hi8 deck all the picture controls including brightness go to itīs default values , so most probably these default values are the neutral values if not I must not use edit to be able to decrease brightness. With the D8 camcorder I donīt have any option and I think that by default it over sharps as the image captured on the D8 seems to have a little more detail.
    I will have to stay with the deHalo plug-in and tweak a little to reduce detail loss.

    I think that eventually an area were I can still do some enhancement is denoizing I have to see the vhs scripts options to try to improve this area a little.If you have any specifically advice in this area Iīm all ears.

    I found a long technical and detailed script recently made and specifically for video8 that I was interested to test I will probably post it hear to have your opinion , However it uses avisynth MT and that poses one question , can I have 2 versions of avisynth the standard and MT on the same pc. I have some applications like videocleaner and others that depend on the standard avisynth so I could eventually run into problems or can they co-exist pacifically and the applications find which one they need?
    Quote Quote  
  2. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    The script that I had spoke about . However it seems to apply to video8 /hi8 already captured by third part and transferred to DVD in mpeg2 so it could not be used as is but can eventually be adapted or some ideas could be exploited

    Script is written for 32-bit AviSynth 2.6 MT
    Plugins can be found at avisynth.nl
    The VidMod plugins are part of QTGMC.
    The script below assumes that the DVD is ripped, and the resulting .vob file is indexed using dgindex.
    To transfer the PAL Video8 tapes to PAL DVD, I used SunrayVideo in the US.

    Code:
    # http://www.coertvonk.com/technology/videoediting/restoring-video8-hi8-10849
     
    Global cwd = "C:\Path\To\File\Root\"  # UPDATE!
    Global pd = cwd + "plugins\"  # UPDATE!
     
    function loadSource(string "fname") {
        SetMTMode(5, 2)
        SetMemoryMax(2048)
        LoadPlugin(pd + "dgindex\DGDecode.dll")
        LoadPlugin(pd + "colormatrix\ColorMatrix.dll")
        LoadPlugin(pd + "nicaudio_206\NicAudio.dll")
        V=DGDecode_mpeg2source(cwd + fname + ".d2v", info=3).ConvertToYV12(interlaced=true).AssumeTFF() 
        A = NicAC3Source(cwd + fname + " T80 2_0ch 256Kbps DELAY -40ms.ac3")  # UPDATE!
        c = AudioDub(V, A)
        DelayAudio(c, -0.040)
        SetMTMode(2)
        return c
    }
     
    function restoreChroma( clip c ) {
        LoadPlugin(pd + "cnr2_v261\cnr2.dll")
        Cnr2(c)
    }
     
    function stabFast( clip c, int "range", int "dxmax", int "dymax", int "mirror" ) {  # http://avisynth.nl/images/Stab.avsi
        LoadPlugin(pd + "rgTools\RgTools.dll")
        LoadPlugin(pd + "depanestimate192\DePanEstimate.dll")
        LoadPlugin(pd + "depan1101\DePan.dll")
        range = default(range, 1)
        dxmax = default(dxmax, 4)
        dymax = default(dymax, 4)
        mirror = default(mirror, 0)
        temp = c.TemporalSoften(7,255,255,25,2)
        inter = Interleave(temp.Repair(c.TemporalSoften(1,255,255,25,2)), c)
        mdata = DePanEstimate(inter, range=range, trust=0, dxmax=dxmax, dymax=dymax)
        DePan(inter, data=mdata, offset=-1, mirror=mirror )
        SelectEvery(2,0)
    }
     
    function stabSlow( clip c) {
        LoadPlugin(pd + "depanestimate192\DePanEstimate.dll")
        LoadPlugin(pd + "depan1101\DePan.dll")
        maxstabH = 20 maxstabV = 20  # maximum values for the stabiliser (in pixels) 20 is a good start value
        trust = 1.0  # scene change detection, higher= more sensitive
        cutoff = 0.5  # no need to change this, but you can play with it and see what you get
        stabRef = c.crop(20,20,-20,-20).colorYUV(autogain=true).crop(20,60,-60,-60)
        mdata = DePanEstimate(stabRef, trust=trust, dxmax=maxstabH, dymax=maxstabV)
        return DePanStabilize(c, mdata, cutoff=cutoff, dxmax=maxstabH, dymax=maxstabV, method=0, mirror=15)
    }
     
    function fastDeinterlace( clip c ) {
        LoadPlugin(pd + "nnedi3\nnedi3.dll") 
        LoadPlugin(pd + "yadifmod_v1\yadifmod.dll") 
        edi=nnedi3(c, field=-2) 
        yadifmod(c, mode=1, edeint=edi) 
    }
     
    function slowDeinterlace( clip c ) {  # see http://avisynth.nl/index.php/QTGMC
        LoadPlugin(pd + "mvtools-2.6.0.5\mvtools2.dll")  # Mod16 version of MVTools 2
        LoadPlugin(pd + "masktools_VitMod\mt_masktools-26.dll")
        LoadPlugin(pd + "nnedi3_VitMod\nnedi3.dll")
        LoadPlugin(pd + "RemoveGrain_VitMod\RemoveGrainSSE2.dll")
        Import(pd + "QTGMC-3.33d\QTGMC-3.33d.avsi")
        c = QTGMC( c, Preset="Slow", EdiThreads=4 )
        return Distributor(c) # This line may or may not be necessary, try removing it and see if you get more speed
    }
     
    function deHalo( clip c ) {
        LoadPlugin(pd + "masktools_VitMod\mt_masktools-26.dll")
        Import(pd + "dehalo_alpha\DeHalo_alpha.avsi")
        return DeHalo_alpha(c)
    }
     
    function vividLessYellow( clip c ) {
        c = Tweak(c, cont=1.15, sat=1.05)
        return ColorYUV(c, off_u=3, off_v=0)
    }
     
    function autoLevels( clip c ) {
        LoadPlugin(pd + "autolevels_06\autolevels.dll")
        c = AutoLevels(c, output_low= 6, output_high= 235)
        return ColorYUV(c, autowhite=true, autogain=false)
    }
     
    function deGrain( clip c ) {
        Import(pd + "LSFmod\LSFmod.v1.9.avsi") 
        Import(pd + "GrainFactory3\GrainFactory3.avsi") 
        LoadPlugin(pd + "dgindex\DGDecode.dll")
        LoadPlugin(pd + "colormatrix\ColorMatrix.dll")
        LoadPlugin(pd + "nicaudio_206\NicAudio.dll")
        LoadPlugin(pd + "nnedi3\nnedi3.dll") 
        LoadPlugin(pd + "yadifmod_v1\yadifmod.dll") 
        LoadPlugin(pd + "removegrain-1.0b\RemoveGrainSSE2.dll") 
        LoadPlugin(pd + "removegrain-1.0b\RepairSSE2.dll") 
        LoadPlugin(pd + "mvtools-v2.5.11.3\mvtools2.dll") 
        LoadPlugin(pd + "masktools-26-for-2.6alpha4\mt_masktools-26.dll") 
        LoadPlugin(pd + "addgrainc-1.7.0\addgrainc.dll") 
        super = MSuper(c, pel=2, sharp=1)  # prepace multilevel (hierarchical scaled) frames data
        mvbw2 = super.MAnalyse(isb = true, delta = 2, overlap=4)  # generate backward motion vectors between the frame n and n-2 or n+2
        mvbw1 = super.MAnalyse(isb = true, delta = 1, overlap=4)  # generate backward motion vectors between the frame n and n-1 or n+1
        mvfw1 = super.MAnalyse(isb = false, delta = 1, overlap=4)  # generate forward motion vectors between the frame n and n-1 or n+1
        mvfw2 = super.MAnalyse(isb = false, delta = 2, overlap=4)  # generate forward motion vectors between the frame n and n-2 or n+2
        MDegrain2(c, super, mvbw1, mvfw1, mvbw2, mvfw2, thSAD=200) 
    }
     
    function upScale(clip c, int "height") {
        Spline36resize(round(height / height() * width() / 8) * 8, height)  # upscale
        LSFMod(Smode=5, strength=70)  # sharpen
        GrainFactory3(g1str=4, g2str=6, g3str=8)  # add some HD grain
    }
     
     
    # prepare source
     
    src = loadSource("DVD\video1")  # UPDATE!
     
    # start processing
     
    dst = src
    dst = false ? fastDeinterlace(dst) : slowDeinterlace(dst)
    dst = SelectEven(dst)
    dst = restoreChroma(dst)
    dst = stabSlow(dst)
    dst = deHalo(dst)
    dst = vividLessYellow(dst)
    # dst = autoLevels(dst)
    dst = deGrain(dst)
    cropLeft = 14 cropTop = 6 cropRight = 18 cropBtm = 18  # typical VHS overscan, keep the results mod8
    dst = crop(dst, cropLeft, cropTop, -cropRight, -cropBtm)  
    src = crop(src, cropLeft, cropTop, -cropRight, -cropBtm)
     
    # compare side-by-side
     
    # COMMENT OUT THE NEXT LINE FOR FINAL VERSION
    dst = stackhorizontal(subtitle(src, "original", size=28, align=8), subtitle(dst, "restored", size=28, align=8))
    SetMTMode(1) # multithreaded Avisynth only - end of script p1
    GetMTMode(false) > 0 ? distributor(dst) : dst # multithreaded Avisynth only - end of script
    Quote Quote  
  3. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    In my third restoration script iteration , I think I have been able to get an additional small incremental improvement.

    In my case captures over hdmi in Huufyuv YUY2 gave me almost all the full luma range ( 1-255) and deshake in virtualdub works on RGB32, so itīs mandatory to previous convert to avoid crushing super black and super whites that can have importance for the final quality

    Now my process is as follows:

    First step - conversion to RGB avisynth script

    Code:
    AviSource("C:\name.avi")
    ConvertToRGB32(matrix="PC.601",interlaced=true)
    VirtualDub output :uncompressed RGB32 ( temporary file) as Huufyuv is not supported .The files have a bit rate of about 335 Mbps so I used a fast and large SSD .
    It could have been done with colorYUV(levels=”PC->TV”) for instance and the output file could still remain as Huufyuv YUY2 but I think the first approach is better

    Second Step – Deshake in virtualDub
    Input – Raw uncompressed RGB file
    Output – Uncompressed YV12

    VirtualDub Filters

    Null transform ( crop black borders)
    Deshake v3.1 – tweaked it again
    Resize (Lanczos3)

    Third Step -avisynth restore script

    Code:
    AviSource("E:\Video8\deshake.avi")
    ComplementParity()
    Cnr2("xoo",4,2,64)
    ChromaShift(C=-4, L=2) # align chroma over luma
    QTGMC( Preset="Very Slow",SourceMatch=3,Sharpness=0.5,TR2=2,Lossless=2,MatchEnhance=0.75, NoiseProcess=1, NoiseRestore=0.7, Sigma=1.5 )
    santiag(strh=2,strv=2)
    autolevels(output_low=6, output_high=235)
    MergeChroma(aWarpSharp(depth=10), aWarpSharp(depth=20)) #  white balance
    Overlay(last, ColorYUV(off_y=-8, off_u=5, off_v=-2), 0, 0, GreyScale(last).ColorYUV(cont_y=30))
    #DeHalo_alpha(rx=2.0, ry=1.0)
    UnsharpHQ(THRESHOLD=20, SHARPSTR=5.0, SMOOTH=0.5, SHOW=false)
    #LanczosResize(720, 576)

    Fourth step - Conversion to H.264
    using virtualDub .It could be done in step3 without the need of another intermediary file choosing the H264 compressor and YV12

    I made some additional modifications that I think have improved a little the final result
    Transferred Cnr2 to just before QTGMC line
    Introduced a new Chroma shift line to align Chroma over luma
    Introduced an antialiasing line with santiag immediately after QTGMC
    Adjusted the autolevels()
    Retained the MergeChroma(aWarpSharp… as so far it seems the one that gives better results however as the overlay was introducing an excess of blue I tweaked a little to reduce the blue color
    Introduced a DeHalo_Alpha line to reduce Halos but with all the other modifications the halos are less visible, so I have not applied it but let the line just in case
    Tweaked the UnsharpHQ line

    I think the results are now already acceptable but as I have no experience in restoration I donīt know if these video8 results are near the best we can achieve or are still very far away.The script is small but itīs very slow even with a very fast computer is doing something like 1,5 fps .I would very much like to have your opinion about the quality of the restoration achieved .

    Related to the santiag I am not sure itīs working but it seems. I initially tried to use the santiag script from cretindesalpes in avisynth.nl and have done a text.avsi file with it and included it in the avisynth plug-in folder together with all the required plug-inīs however it gave an error , It seemed to me that the script as is could not be used directly and was necessary to select functions and other data to make a functional script as I didnīt know how to do it I searched for examples of santiag use and found smaller apparently functional scriptīs that I tried to use and gave no errors so I suppose itīs working and I think a very, very small benefit is visible but Iīm not complete sure about the benefit and if itīs working .If itīs working the benefit appears to be minimal , however I don’t know also exactly how to work with it, eventually it could be improved.

    Code:
    Function santiag (clip c, int "strh", int "strv")
    {
    	strh = Default (strh, 1)
    	strv = Default (strv, 1)
    
    	c
    	santiag_dir (strh)
    	TurnLeft ()
    	santiag_dir (strv)
    	TurnRight ()
    }
    
    Function santiag_dir (clip c, int strength)
    {
    	c
    	w = Width ()
    	h = Height ()
    	santiag_stronger (strength)
    	Spline36Resize (w, h, 0, -0.5, w, h * 2)
    }
    
    Function santiag_stronger (clip c, int strength)
    {
    	c
    	(strength <= 0)
    \	?	NNEDI3 (dh=true, field=0)
    \	:	santiag_stronger (strength - 1)
    \	.	NNEDI3 (dh=false, field=(strength % 2))
    }
    
    AviSource("E:\Video 8 Finais\Videohelp4\Deshake.avi")
    
    santiag()
    In what Chroma shitf is concerned I could adjust Y/C delay on the Hi8 Deck however it will be no easy task as the tape must be running and it must be difficult to adjust with the tape running . However the hardware correction could be eventually more effective than the software approach I donīt know ( but the software approach seems easier, I think)

    Related to sound I made a small test with isotope cleaned hum and applied a spacial denoiser and almost all background noise has disappeared, perhaps too much denoising.

    Related to a process for erasing timestamps and avoid them runing free on the screen after deshaking I tried mask in deshaker it doesnīt seem to work at all, tried logoaway it worked but without a stable background the results are not fantastic.I have to try delogo and one MSU filter but I donīt think they will work 100%.
    In another post I will put an example of the problem and results achieved so far.
    Image Attached Files
    Last edited by FLP437; 15th Nov 2016 at 19:18.
    Quote Quote  
  4. Originally Posted by FLP437 View Post
    Is it possible to toggle time/date overlays on some of the captures ? That would be "best"
    I am unaware that for video8 itīs possible to toggle time /date overlays during capture , I thought that time and date were hard recorded on tape if there is any chance do disable the overlays I was very much interested to know how to do it.
    I might have a bad memory , but there are at least a few models that could toggle on/off date time . I remember a long time ago I was able to. Maybe your model wasn't one of them, but here is a list of some Sony models
    http://www.octochron.com/faqSonyHi8RcTimeCodeDataCode.htm

    EDIT: My bad, you're using "video 8" not Hi8 !!

    I have 2 versions of avisynth the standard and MT on the same pc. I have some applications like videocleaner and others that depend on the standard avisynth so I could eventually run into problems or can they co-exist pacifically and the applications find which one they need?
    You can run "regular" scripts with avisynth MT 's .dll . They cannot co-exist at the same time (you just swap out one .dll for the other). There are very few plugins and cirucumstances that require the original "vanilla" .dll


    .
    .
    LoadPlugin(pd + "depan1101\DePan.dll")
    .
    .
    You should update your depan plugin, because fizick recently fixed the dreaded depan green frame bug. Seemingly random green tinted frames would occur with depan , and functions based on it like stab

    VirtualDub output :uncompressed RGB32 ( temporary file) as Huufyuv is not supported
    Another option to using uncompressed video (either RGB or YV12) is use a lossless codec, such as ut video codec, magic yuv etc...They are faster than huffyuv, more compressed, and have multiple colorspace optoins
    Last edited by poisondeathray; 24th Apr 2016 at 01:31.
    Quote Quote  
  5. Originally Posted by poisondeathray View Post
    fizick recently fixed the dreaded depan green frame bug.
    Hooray!
    Quote Quote  
  6. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by jagabo View Post

    Originally Posted by FLP437 View Post
    The auto white balance that I included is part of an original multiscript from Lordsmurf. However I tried to do it without the second line as Jagabo told about a blue shift to the brights and I don’t dislike the final color
    The first of those two lines

    Code:
    MergeChroma(aWarpSharp(depth=10), aWarpSharp(depth=20))
    has nothing with white balance. It sharpens the luma a little, and the chroma more. Since the chroma is so low resolution horizontally I would use something more like:

    Code:
    MergeChroma(Spline36Resize(width/4, height).aWarpSharp(depth=10).Spline36Resize(width,height))
    That keeps the luma intact but sharpens the chroma significantly. You might need a ChromaShift() after the operation to better align the sharpened chroma with the luma.

    The second line, Overlay(...) applies a color shift to the brights but not the darks, but the shift is in appropriate for the video. Unless those white walls are supposed to be blue. In my opinion the white balance of the original clip is pretty close and it doesn't need any adjustments.
    Keep in mind that the script is beta. I may have mistakes in notes, and some lines are not refined (or duplicated for more options) yet. The current version has many corrections and additions, but is too messy to share yet.

    When I'm finished with it, I also plan to have a few people like jagabo audit it for mistakes or refinements.

    And of course, it will come paired with exhaustive guides on restoration.
    Want my help? Ask here! (not via PM!)
    FAQs: Best Blank Discs • Best TBCs • Best VCRs for capture • Restore VHS
    Quote Quote  
  7. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    I made initially my video8 captures with a D8 camcorder, however some time ago I bought a Hi8 Deck Sony EV-S9000E in order to see if I could get slightly better captures. Now I have almost all my video8 tapes recaptured with the HI8 deck and I am with a dilemma as I`m unable to decide which one from the two devices provide better captures.

    As longer as I can evaluate ( but Iīm no expert ) the deck does a better job with vertical lines, even without TBC ( both captures have been made without TBC and DNR) , the D8 captures seems to have a little more noise, they have the color edge issue on the right, the sound seems better from the deck ( and the voice booster works well) however the picture from the Hi8 deck seems more blurred (smear ), it seems that the D8 does provides better picture detail. I thought initially that the D8 was doing some sort of artificially over sharpening but after applying my “restore” script it seems that it still provide better detail. And detail could be everything in this low resolution format, so Iīm in a dead end as Iīm unable to make my mind, .I am in need of expert advice, Iīm hoping you can help me identify which one provides the best capture .Iīm posting two captures sets one with the D8 Cam and the other with the HI8 deck both raw and “ restored” using my actual script (identical for both).

    Related to my avisynth script I would be interested if possible to do only one procedure with only one script to simplify the overall process, that is, as it is now , somewhat complicated. Basically the problem is the deshaker used in virtualdub that prevents having only one script . One possibility is to use Depan as proposed by jagabo ( now without the bug ) , however is there a possibility to call deshaker from inside an avisynth script ( using the saved processing settings) ?.If there is such a possibility could anyone explain me how to do it ?

    Even if my script gives, I think acceptable results for footage not very problematic it could still easily incorporate incorrect approaches like incorrect order for some plug-inīs , incorrect settings , not best in class plug-inīs and versions, etc .Also some sort of light denoizing I think could be useful but I have not been able so far to include MDegrain or similar approach . So any help in this context that would allow optimizing the script is welcomed.

    Related to Defliker that johnmeyer as proposed I have not yet found the avisynth version it seems the Russian site is not active anymore , only found virtual dub versions from Donald Graft or the MSU filter, will this be ok in virtualdub just before deskaking or it was better to use the avisynth version Iīm unsure if itīs a script or plug-in and where to get it.

    For very low light shootings with grain I found that use of Neat video could bring some benefits, even if the result may look sometimes a little over processed but even so it does provide a better result.

    However can I call neat video from inside an avisynth script? If I could it was the right approach I think. If not where should I position it in virtualdub just before deshaken or at the end of the “restore” script
    Image Attached Files
    Quote Quote  
  8. D8_raw.avi has worse oversharpening halos but has a lot more detail and less horizontal time base jitter. Of course, the differences here aren't just a matter of raw YUV vs. DV capture. All the other hardware is different.
    Quote Quote  
  9. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Thank you for giving me your opinion jagabo. Iīm tempted to stay with the D8 captures as masters for my restoration project because of the additional detail even if the D8 doesnt do a reasonable job with vertical lines and captures seems to have a little more grain. However there is another thing that bothers me, I frequently get captures with more frames from the same tape from the deck HI8 than from the D8 Cam , also the deck gets always some extra frames from the beginning of the tapes. I have to verify and confirm but it seems to me that the dvd recorder as ADC lost some frames from the D8 captures and probably none or less from the deck as it has a dropout compensation circuit. Oblivious Virtual dub reports nothings as if such a problem arises is upstream ( player->dvd recorder ) not on the hdmi stream. However I donīt get a video audio desynchronization
    ps- related to the previous captures all the workflow is identical only changes the player D8 Cam or Deck Hi8
    Last edited by FLP437; 28th Apr 2016 at 17:46.
    Quote Quote  
  10. The D8 source is better on almost every level, especially the detail captured in the shadow. If you go to the last second of the clips and look at the big doors on the side of the temple, you can barely see anything in the shadow on the old deck, but the D8 capture shows all sorts of shadow detail. I don't think this is the result of any sharpening being applied by the deck.

    As for the sharpening, I did not find it objectionable. I paid a lot of attention to the fence at the bottom of the frame and other places where I would have expected halos. Yes, some halos exist, but I wouldn't classify them as objectionable, although that is obviously a matter of persona preference.

    The one place where I thought the D8 was far worse was time base errors. So I am seeing the opposite of what jagabo is reporting. I've gone back a couple of times to check my results, but I keep seeing the same thing.

    Here is the result:

    D8 capture (note squiggly vertical lines)



    Deck capture (straighter lines)


    I see no downside to using D8 as your source.

    As far as your restoration, I only looked at the D8 restoration clip and while I like the improvement from doing motion stabilization, you increased the contrast, which was already pretty high, and absolutely crushed the blacks. I'd recommend that you back off a lot on those settings.
    Quote Quote  
  11. I took another look at the oversharpening halos and I was mistaken earlier. The Hi8 video has worse oversharpening light halos, but less dark halos. Look at the dark window frame on the white wall at the top center. The D8 video has a more pronounced bright halo there. But the vertical pillar at the center bottom of the frame has a very pronounced dark halo just to it's left in the Hi8 video.

    Regarding the horizontal jitter, it depends on what part of the picture you look at. Yes, the Hi8 is better near the bottom of the frame where the fence is, but over most of the rest of the frame the D8 is better -- like the light post in the bottom center of the frame.

    And about the noise: I think the noise of the D8 cap is what's on the tape. The Hi8 cap has reduced the noise and obliterated a lot of detail along with it. You can reduce the noise of the D8 cap in software and retain most of the detail, but you can't get back the detail that was lost in the Hi8 cap.

    Overall, I agree, the D8 is better.
    Last edited by jagabo; 28th Apr 2016 at 20:37.
    Quote Quote  
  12. You're right. I missed those TB issues on the pillar: there are some nasty 1-2 line TB offsets. I have never been an expert on TBC circuits, but I guess this shows how the simple ones found in consumer decks only address some of the issues.
    Quote Quote  
  13. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Thank you johnmeyer.With already two similar opinions in favour of D8 capture (my own opinion goes also in the same sense) I have no doubts now to use the D8 captures as my masters for restoration.I have only to confirm if really I have lost frames in these D8 captures but if they don't mess with audio synchronization and the number of eventually lost frames are limited perhaps it's not a big deal.
    I will be out during this week-end and will have no possibility to test the script with new settings as per your recommendations.
    I will try to see if I can get settings that could be used for the majority of tapes/ scenes but that seems a difficult task as most of them seems to need individual adjustments .
    I still have to test some of the proposed ideas given in this thread.Related to deshaker I think I have already discovered how to use it from inside avisynth and that will simplify my procedure a lot but I can't test now.I want also to try reversing deshaker ana Qtgmc order and see if I will get better results.

    Related to tbc both D8 an Hi8 deck were off in these captures.However the D8 even if on almost doesn't work.The deck TBC is better if on it works reasonably well.
    Last edited by FLP437; 28th Apr 2016 at 21:05.
    Quote Quote  
  14. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Due to professional matters I have been obliged to be out of this forum for some weeks as I had not much time to devote to my video activities and it may happen that I will continue to have less time available for my video activities during the coming times.

    Although Iīm still very much interested to develop my know how in terms of avisynth and avisynth scripts and trying to develop my own approach, itīs however a time consuming activity and even if I reach a higher level of knowledge it will be in practice almost impossible to do better than what has been already done by the developers in these areas, so trying to discoverer the wheel again could be a relative loss of time namely now that I will have less time to dedicate to these matters .

    I think that probably a good option for me now could be to try to use scripts already developed and made available for public use.
    In this context probably two types of scripts could be quite useful for me

    . videoFred restoration script , not the original from videoFred as I think is it applies to progressive media but I think I have saw some adapted versions for analog VHS (and probably video8) interlaced media ( I think that johnmeyer as made a version of this type I not sure)

    . Simple and efficient analog denoising scripts namely for use in conjunction with a NLE

    I will be grateful if you could recommend the most appropriate and recent versions in your opinion ( for video8 and VHS) and where they can be found. Also if there are dependencies from other avisynth versions and / or specific plug-inīs ( in this moment I use avisynth 2.5.8) , in case they are not documented in the scripts themselves.

    If necessary I can afterwards try to adapt (if I can achieve this goal) to any specific needs I may have
    Quote Quote  
  15. You definitely do not want to use any variation of VideoFred's script on VHS material. It would do really bad things.
    Quote Quote  
  16. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    And do you think there may exist a generic restoration script for use with VHS and video8 media that I can can use as a base and try afterwards to adapt to my needs if necessary , if such a script exist it it could eventually speed-up my process if not it still will go on but slowly.
    Quote Quote  
  17. Rancid User ron spencer's Avatar
    Join Date
    May 2005
    Location
    Ish-ka-bibble
    Search Comp PM
    nice scripts..do you have a final workflow?
    'Do I look absolutely divine and regal, and yet at the same time very pretty and rather accessible?' - Queenie
    Quote Quote  
  18. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    very interesting thread ! I too have been spoiled for choice between a high end 'consumer-pro' HI-8 deck versus a D8 camcorderfor for
    my 8/Hi8 transfers. I finally opted for D8 especially for ...price and workflow convenience.

    Indeed I would think at first glance that a high end deck solution could better grab the very last half MHz or so luma bandwidth or
    possibly provide with better chroma combing , etc... not to forget what a high end external TBC could also provide for.

    But in the end ...I was worried that all this would STILL happen in composite signal form . Weak by design. Only very few high end
    HI8 Sony pro-deck would provide with an yuv output signal and even up to SDI for one or two top notch unit if my memory is good .

    The remaining consumer semi-pro decks ( while yet quite costly ) only provide with a composite output and more annoying
    with an inner composite signal path all the way under the hood.

    having said that why would the D8 could be different here ? that is provide with an inner signal transfer in component signal to the
    built-in hardware DV encoder chip . Well ....possibly simply because it requires it (yuv) ! Indeed I may be wrong and only a detailed
    schematic diagram would help but I dont have any .

    Now if this is correct then the inner signal conversion (Hi-8 to DV) should gain some extra quality as compared to the usual path via
    a composite output signal to some outboard gear.

    Which also lets little hope that the choice of any (good) digitizing card will make any significant difference . Then I would
    probably capture as interlaced YUV 4:2:2 (miniYUV codec) to avoid a DV step (not the best option as "MJPEG" 25Mbits ) ;

    Still I don't know if that would make a significant visible difference in the end ... I would gladly read from somenone having made
    such exhaustive comparison

    I do have some very good cards , Osprey and Truevision for instance , but the latter captures as MJPEG anyway while up to 50Mbps...
    with a good driver !! ; now it remains that DV (as some 'Mjpeg like' version) is quite clever in its refinements even as 25Mbps data
    rate . But I don't own any Hi8 pro Deck , too bad


    One word on John's D8 vs Deck pictures : this is a 'good' example about the difficulty of conducting a fully 'objective' comparison
    which remains a demanding practice. Obviously here constrast is almost the basic simple difference . And gamma as well .

    Adding a difference in color saturation and you get the picture , I mean after having fixed/realigned all these points (quite simple at
    first glance ) there should be left very little difference .

    So yes that does leave D8 as a very simple straithforward option ...unfortunately all PAL D8 camcorders seem plagued by the right
    sided color column aberration ! really too bad since sometimes they can be pretty large up to 16 pixels . NTSC D8 camcorders seem
    free of these color artifacts , which is quite disapointing from Sony

    Regarding the earlier point about to deinterlace or not ( my kingdom for a Deck ! ) , but my 2cts first :


    since I will never watch again on a CRT interlaced display the choice remains very simple : the video WILL BE deinterlaced in the end
    . So it remains to be proved that say QTGMC does an inferior deinterlace job (as used in the first place) when compared to
    a TV buil-in deinterlacer. Absolutely not my experience so far. So both pratcices are okay to me . While deinterlace in the fisrt place seems somewhat more adequate , imho.
    Quote Quote  
  19. Originally Posted by kerryann View Post
    the video WILL BE deinterlaced in the end
    . So it remains to be proved that say QTGMC does an inferior deinterlace job (as used in the first place) when compared to
    a TV buil-in deinterlacer.
    Two flaws in your thinking.

    1. Future displays might be able to display interlaced material without deinterlacing. There is absolutely nothing that says this can't or won't happen. Therefore, if you permanently store your video as deinterlaced, you will have destroyed half of your video's original quality for nothing (remember, deinterlacing inevitably and unavoidably destroys quality). There is no going back.

    2. You assume that QTGMC (a great script, and one of the best free deinterlacers) is the "ultimate" deinterlacer and that it will do a better job than hardware deinterlacers that are built into not only your current set, but future sets (which might be even better). Since the built-in deinterlacer in a good, high-end display can be "tuned" to the specifics of how that display works, it is my belief that even at this moment in time, most people will get as good, or better, quality leaving the video as interlaced and then letting the display handle that task.

    YMMV, obviously, depending on your TV set.

    Every person sees things differently and has different thresholds at which things annoy them, but for me (a professional video restorer), I have never once been conscious of interlaced artifacts when viewing interlaced material on my TV. This includes hundreds of hours watching 1080i sports, and also watching interlaced HD video from my Sony FX-1, a semi-pro camera which records only interlaced video.



    On your other point about which deck will provide a better result, some of your thoughts include trying to avoid the composite output that most decks provide. I won't argue about the issues resulting from combining color and luma in the NTSC composite output, but in my experience these issues (dot crawl, bad red reproduction, etc.) pale by comparison to the fact that 8mm, Hi-8, Beta, VHS, and S-VHS degrade the NTSC signal so much that the additional violence done to the signal by the composite output is probably pretty minor by comparison to the ugly video created by early consumer video format. If you were recording composite output from a really pristine OTA source, then I'd sing a different tune.
    Quote Quote  
  20. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    well I never advised to throw away the interlaced sources ! Indeed keep them in the hope of some hypothetical better solution to come..

    But otherwise yes I do think interlace video will vanish and eventually disapear. Simply as a past solution even if it still does a good job ! why would the industry maintain and support several standards at a time when there is no benefit for it , simple as that

    Then what do you precisely call "hardware deinterlacers"? I mean where are they different from software solutions ? Do you think a TV deinterlacer provides with some breaktrough powerful motion adaptive advanced solution (for instance) that no software could match or reproduce ? If so QTGMC would sometimes fail hard in comparison ; AFAIK and from what I see from usual TV sets it does not happen at all . I still even wait a side by side comparison with heavy weight dedicated gear such as Snell , ForA ,etc..)

    Can you precise what the hardware deinterlacer of your TV do that is so different from QTGMC ? where or how does it improve on QTGMC ?
    does it yet -at first glance-only provide with a better Vertical resolution since you mention it is the main evil point? even simply as a personal subjective feeling

    now if nothing like this happen ... then deinterlacing in the first place provides with a much more 'ubiquitous' solution (imho) since you then no more have to bother about your viewers deinterlacing solution ! not to mention it is so much more practical as a workflow

    On your last point , I am quite surprised . I am no professional of restoration . I only take interest in my family tapes some being 30 years old . And I can spend numberless hours to try and squeeze out the very last tiny detail I can !! indeed I can understand it is unacceptable for a business

    my earlier tapes were shot on a Canon VME2 ( PAL 8mm) and sometimes I am almost shocked from what remains coming from an half-inch 240k photosites sensor , and less than 40dB video S/N overall !! and that's my incentive to look for the last lost detail ...and also where the 'amateur' shines I guess

    I have only two tapes (among 50) which turned bad and these were the only BASF I ever bought ! damn bad chemists . Japanese oxyds have travelled trough time in some unbelievable manner that I did not expect at all
    Quote Quote  
  21. Can you precise what the hardware deinterlacer of your TV do that is so different from QTGMC ?
    Every TV set uses different technology. Even within a particular category of TV technology, such as LCD, there are big differences in implementation. A deinterlacer that is built into the hardware of the TV set could take advantage of specific hardware characteristics that a general-purpose software deinterlacer would not know about and, even if it did, it could not control anything because it doesn't have access to the hardware. I haven't been able to find any description of how hardware deinterlacing is actually done in a TV set, so I cannot say with 100% certainty that this is done. However, being an engineer, that is certainly what I'd do.

    As one example, my Samsung LED LCD has a feature which was new at the time (four years ago) called "local contrast enhancement." Since the TV uses LEDs for the backlighting, these LEDs can be instantaneously turned on and off. To increase contrast, when a pixel is supposed to be really dark, the LED lighting is turned off for that one frame, and then turned on again when brighter light is required. This makes the blacks blacker, and increases contrast. A deinterlacing algorithm which also had direct access to this pixel-by-pixel backlighting could use tricks, similar to anti-aliasing, to produce better results from a deinterlacing operation.

    Again, I cannot find any technical papers on this, so I can't prove that this happens. All I know is that on my particular TV set, I've never once been tempted to spend the time running any of my interlaced videos through QTGMC.
    Quote Quote  
  22. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    4 th iteration of my script

    I introduced additional denoise plug-inīs (useful ) and DeFlicker ( not so much)

    .DeGrainMedian and TTempSmooth
    .DeFlicker

    Using both DeGrainMedian and TTempSmooth seems more effective to reduce grain and noise without reducing significant detail .However I could use only TTempSmooth as I think DeGrainMedian as a minor impact on detail and I donīt see any impact on detail with TTempSmooth.

    I thought DeFlicker could have a positive impact in sudden luma changes due the camcorder AGC but I didn'tīt see any results perhaps Iīm not using the plug-in correctly , I have to see syntax in detail.

    I tried autolevels=true and also autogamma=true . They seem to have a more positive impact on scenes with over or under exposure and even if they donīt solve the situation they seem to mitigate it.

    As I have said this script is extremely slow ( about 1,5 fps in a extremely fast computer, this is due to QTGMC and the aggressive settings I have choose , if for instances QTGMC is not needed as the objective is to maintain interlaced media to do a DVD the speed goes up to about 65-70 fps. Also changing QTGMC settings speed-up significantly without significant visible problems.I have not tried for now avisynth MT but perhaps it will also speed-up the script.

    I have tried to integrate the different scripts and steps in only one script for all the process from the raw YUV2 to the final H264. The only problem to do so is deshaker. In virtualdub ( using deskaher filter with the correct settings ) and doing File -> Save Processing Settings we can access a list of all Deshaker arguments (opening the .vcf settings file in Notepad). I run previously pass one in VirtualDub filter .

    The script must include something like the next lines

    Code:
    LoadVirtualDubPlugin("C:\Program Files\Virtualdub 1.9.11\Plugins\Deshaker.vdf","Deshaker")
    Deshaker("19|1|30|5|1.09259|1|1|0|640|480|0|1|2000|2000|2000|2000|4|2|1|2|50|40|300|4|C:\\Users\\owner\\AppData\\Local\\Deshaker\\Deshaker.log|0|0|0|0|0|0|0|0|0|0|0|1|1|1|35|40|10|25|1|1|30|30|0|25|0|0|1|1|1|10|1000|1|88|1|1|20|4000|100|20|1|0|ff00ff") #deskaher arguments from the vcf file
    I attempted to run the script but even if it gave me no errors the resulting video was unstable , I will have to check the script to see the origin of the problem.If I canīt solve it I will try DePan but for now i will stick with deshaker.

    Last script version

    Code:
    AviSource("E:\Video8\deshake.avi")
    loadplugin("C:\Program Files (x86)\AviSynth 2.5\plugins\DeGrainMedian.dll")
    #loadplugin("C:\Program Files (x86)\AviSynth 2.5\plugins\deflicker.dll")
    AssumeTFF
    #DeFlicker(lag=-25, scene=30) # scene=30
    Cnr2("xoo",4,2,64)
    ChromaShift(C=-4, L=-6) # align chroma over luma
    QTGMC( Preset="Very Slow",SourceMatch=3,Sharpness=0.5,TR2=2,Lossless=2,MatchEnhance=0.75, NoiseProcess=1, NoiseRestore=0.7, Sigma=1.5 )
    santiag(strh=2,strv=2)
    autolevels(autolevel=true, autogamma=true, output_low=6, output_high=235)
    #ColorYUV(autogain=false, autowhite=true)
    #MergeChroma(Spline36Resize(width/4, height).aWarpSharp(depth=10).Spline36Resize(width,height))
    MergeChroma(aWarpSharp(depth=10), aWarpSharp(depth=20))  
    Overlay(last, ColorYUV(off_y=-8, off_u=3, off_v=0), 0, 0, GreyScale(last).ColorYUV(cont_y=30))
    #DeHalo_alpha(rx=2.0, ry=1.0)
    DeGrainMedian(limitY=4,limitUV=6,mode=1, interlaced=false, norow=true)
    TTempSmooth(maxr=5, strength=3, lthresh=5, cthresh=25, lmdiff=10, cmdiff=15) # (maxr=4, strength=4
    UnsharpHQ(THRESHOLD=20, SHARPSTR=5.0, SMOOTH=0.5, SHOW=false)
    #LanczosResize(720, 576)
    Related to Hi8/video8 decks in terms of picture quality, the EV-S9000 certainly is one of the best available VCR loaded with some of the best existing technology from professional Sony Hi8 machines Just to mention a few, the EV-S9000 has picture noise suppressing system a digital comb filter (DCF) a drop out compensation , time base corrector (TBC), Picture control and a lot of other goodies. I have also an SLV-T2000 a double deck Hi8/video8 - VHS and even if not so good as the EV-S9000e it was also a high end deck and in this case it has been with me for 18 years and despite being in excellent conditions the image presents also smears / blur actually more than the EV-S9000.So it seems the decks donīt give so much picture detail as a good D8 camcorder, however only if I had the right equipment to optimize the tracking of the Video8 decks I could be certain that the D8 camcorders are better.

    There is one particularity that I have noticed with the EV-S9000E , I got between 50 and 80 more frames from the tape beginning that with a D8 camcorder, I not sure why, but it happened with all the tapes.

    Related to QTGMC for me the matter is settled. As I donīt use tvīs only computers to see the files and the results are better with QTGMC than with the on the fly PC deinterlacers the problem is solved as i donīt see any better solution. In the future if there are better hardware or software deinterlacers I will see to it when the times comes.However I think I have one or two capture cards that do deinterlacing by hardware if I have time I will try them however I think they donīt have the option to retain both fields only blend or discard one or another and so I donīt think they will provide better results ,however they will do deinterlacing on the fly perhaps with reasonable results.

    restore 4- autolevels=true, autogamma=false both denoise plug-inīs
    restore 5-only TTempSmooth and autolevels=true and autogamma=true

    I think version 5 is perhaps better however as I I have a tendency to perceive more contrast ans saturation as better sometimes it seems that version 4 looks best

    Any suggestions related to better plug-inīs , settings or relative better plug-inīs positioning on the script are welcomed
    Image Attached Files
    Last edited by FLP437; 19th May 2016 at 19:27.
    Quote Quote  
  23. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    John, it's not about endlessly argueing etc...

    But you said my thinking had flaws . Quite possible but it still remains to demonstrate so far. Let's see yours if you don't mind and
    nothing personal indeed ( I highly appreciate your most valuable posts )

    Considering you strongly advocate keeping interlace all the way "for the benefit of keeping FULL vertical resolution" , then lets
    take the best case there is : display interlace footage on an interlaced display ( it has to be a CRT here since you make some
    confusion on the way a LCD could also operate as interlace , but lets forget that )

    Then go close to the interlace CRT display ( imagine some ultra crisp pro grade PVM Sony CRT monitor , I think it was called "PVM xx"
    series..) and watch what happens to some round object moving slowly sideway ; what do you see .... terrible *vertical* resolution
    picture !! ( jaggies that is)

    Do the same experience with pictures featuring Horizontal fine detailed thin lines ; what do you see again ? terrible *vertical*
    resolution (flicker) picture . And this happens on the supposedly 'adhoc' interlaced display monitor ! heck

    You only get a relief of this messed vertical resolution ...from a distance , where your eye is no more accurate enough to notice
    the trouble.

    that -IS- a flaw ,by design, and why interlace should eventually vanish as it brings only possible trouble. imho

    Regarding your elusive description of what the deinterlacer inside your TV might be , you admit to have no clue and I will
    certainly take no advantage from this since yes you (one) can only imagine what it may be .

    My guess is that it relies on no ultra brilliant solution ( super motion adaptive compensation and the likes you think there is) but
    more simply on first yet addressing the mentionned design flaws of interlaced video via a palette of common filters for
    fighting cross color artifacts, cross luminace as well , that is say a variety of the best comb filters there exist for such kind of
    processing. To begin with

    But this does not address the hard task of mixing 2 possibly quite different pictures (fields) . So it could simply be that either
    this 'smart' TV deinterlacer simply throws away one field (..) or goes the Bob way as QTGMC !

    Then comes the image processing you mention and sure modern TV excel at this. With fanatics of "x" brand image processing versus "y" brand . Just too normal.

    Now I will focus back on the inner aspect of the OT thread . I jumped in since he did what I intended to do but would be reluctant to
    pay for a try

    I would in the first place consider purchase either a EVS1000b or a 9000 keeping D8 as a second solution. I simply would have liked
    to escape the DV conversion step . So I am quite pleased that FLP437 has explored both solutions since the outcome is 50/50 at best.

    In the end D8 is so simple here that I can even forgive the lost pixels on the right side

    I will now carefully read what he has has undertake after capture . Indeed i am fully aware that this huge (too 'rich' ?) script
    addresses his own video shot(s) problems and cant be used as is outside this workflow.
    Quote Quote  
  24. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Originally Posted by FLP437 View Post

    Sorry for the long text, Iīm looking forward to get your feedback and expert advice about my restoration approach, in order to fine tune sequence, script and parameters as I feel Iīm still far from the optimal and unsure related to some options and parameters. Iīm sure I will cut my learning curve dramatically and will be able to get strong results quickly with some help from the forum.

    Code:
    AviSource("E:\Video8\k15_tape.avi")
    ComplementParity()
    QTGMC( Preset="Slower",SourceMatch=3,Lossless=2,MatchEnhance=0.75, NoiseProcess=1, NoiseRestore=0.7, Sigma=1.5 )
    autolevels()
    Cnr2("xoo",4,2,64)
    ConvertToYV12(interlaced=true)
    MergeChroma(aWarpSharp(depth=10), aWarpSharp(depth=20)) #  white balance
    Overlay(last, ColorYUV(off_y=-8, off_u=9, off_v=-2), 0, 0, GreyScale(last).ColorYUV(cont_y=30))
    UnsharpHQ(THRESHOLD=20, SHARPSTR=4.0, SMOOTH=0.5, SHOW=false)
    #crop(2, 16, -22, -36)
    #LanczosResize(720, 576)
    regarding your very first post , this is quite good as deshaking, since the task was ...l immense to say the least ! kudos

    Otherwise this is quite correct for video8 , I mean as long as you stay rather in mid-shot , the lack of details of 8video remains quite acceptable. what camcorder ?
    Quote Quote  
  25. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Originally Posted by FLP437 View Post
    Related to Hi8/video8 decks in terms of picture quality, the EV-S9000 certainly is one of the best available VCR loaded with some of the best existing technology from professional Sony Hi8 machines Just to mention a few, the EV-S9000 has picture noise suppressing system a digital comb filter (DCF) a drop out compensation , time base corrector (TBC), Picture control and a lot of other goodies.
    In fact I did not consider purchasing an EV-S9000 precisely because -all this- is present on a V6000 for instance ; too bad I had one which I sold > 15 years ago !

    the 9000 deck was more specifically doted with quite extended editing facilities , but nothing more on the player side . So far I can remember the best bench figures that came out from a 'Prosumer' Sony Hi-8 came from a V6000 camcorder ! some 6MHz luma at -3dB I would say and excellent chroma S/N as well ( 45dB ?)

    Even the VX1 (3ccd) which I owned too did not quite match the V6000 figures ; I still have this one

    still the TBC remained rather minimal , but with memory enough for some DOC as well ( if not too important) , and a quite effective chroma filter . it is quite possible that Sony reintroduced all these in the better D8 camcorders , almost sure they did
    Quote Quote  
  26. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    @kerryann thanks for yours comments

    I am no professional of restoration . I only take interest in my family tapes some being 30 years old . And I can spend numberless hours to try and squeeze out the very last tiny detail I can !! indeed I can understand it is unacceptable for a business[
    Also being an amateur this as been exactly my point in these last months and to get to it I spend a lot of time and also money pursuing this objective “ squeeze out the very last tiny detail I can” .

    Now I think I should have invested in a prosumer HI8 or D8 camcorder not a deck or a standard D8 camcorder, but I began with very simple and cheap solutions and as I was not satisfied with the results I gradually evolved to more robust and expensive solutions and now itīs a little late I have already invested to much time and money and have amassed a relative large collection of players, capture cards and cables.

    However the EV_S9000 as allowed me to salvage some tapes that I could not read with the D8 camcorder and I want also to try to salvage others adjusting the tape guides to be able to read tapes that have been original recorded with loose tape guides in the original camcorder .

    So far the solution to use the Panasonic DVD recorder ( DMR-EH65) as an intermediary ADC and TBC has proven for me as the best solution however is a pity that the D8 player in not a little better but I have to stop buying equipment if not my spending's will go up exponential to get minimal or no additional results at all.

    There is only one thing that I still want to try , to use other hdmi capture card and see if it can make any difference. I bought recently an almost unknown chinese brand Magewell pro Capture hdmi pcie 2.0 card basically because I need an internal hdmi card for an HDMI capture project and because I got curious when I remarked the chips used ADV7842 and XilinxŪ 7 series FPGAs , if the implementation around these chips is good it could be eventually an interesting card without having to go to the more expensive and renamed brands, I have not acquired it for this video8 and VHS capture project but I will give it a try and compare with the startech usb3hdcap transfer / transcoding or will try directly svideo captures. If I can see any advantage I will report it or open a new thread in capture section , however I don’t have much free time now so it could take some time.

    The camcorder used for the present captures has been the Sony DCR-TRV238E ( the original that recorded the tapes was the Sony CCD-F500E ), I have the service manual that have the block diagrams and schematic’s in case you want to try to confirm your theory. I’m also including ADV7842 specs.
    Image Attached Files
    Last edited by FLP437; 19th May 2016 at 21:13.
    Quote Quote  
  27. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    well, I would not consider purchasing any "better" Hi-8 player than your EV9000 unless it has some problems , say you bought it on second hand and the former owner would have made extensive editing usage, up to raise possible tape transport and/or head drum alignment or wear for instance...

    If it works okay there is no "better" player that would be of any use to purchase , imho

    your DCR 238 is almost a twin of my DCR 230 , and both feature the mini-TBC and CNR filter , so here again no better player if you only consider straight Hi8=>DV conversion.

    ( a detailed list for'valid' D8s => http://madcat.zero.free.fr/html/sony%20DCR-TRV%20database/sony%20DCR-TRV%20database.htm )

    I was lucky to buy one in immaculate condition from a lady who did not use it for transcoding any thousand tapes library (!) , but only for shooting a couple of family stuff.

    Then regarding the capture card as long as you consider collecting the composite Y/C output of either your EV9000 or your D8 , I don't see any significant improvement you can get here between say a 50€ second hand Osprey versus any fancy 1000€ exotic (named) card ; since it remains a trivial signal conversion .

    I noticed you use of Huffy or UTVideo , there again nothing to gain in terms of quality .

    I made a different choice here . I choosed Cineform which is obviously visually lossless in the 70Mbps (overkill for D1 8/Hi8) profile I use even after many generation copies ; and scrolling a multi tracks timeline remains a breeze with Cineform.

    The reason here was I wanted to use Cineform RAW in some way to 'redistribute the cards' in a wavelet form and I found it interesting ; providing some very subtle picture conversion

    On this aspect I mentionned earlier that in the actual state I would (could) not make a choice between the two pictures ( D8 vs EV9000) just a few posts ahead. Simply because there is an obvious difference in brightness and contrast !

    once you have matched or so the two pictures , I could prefer the D8 . I say I could prefer since it is very easy to make corrections on the EV9000 version so that both pictures do look alike

    And there is no "squiggly vertical lines" default per se in the D8 picture . It is quite simply that the D8 operates a full V size digitzing with the usual 8 or 10 scrambled bottom lines when the EV9000 eludes partly here !
    Quote Quote  
  28. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Having reviewed your denoising I do not see any use of MCTD, why ?

    You could consider use MCTemporalDenoise which proves to be quite good. It is some kind of an AIO/combo "swiss knife" script featuring many things from the best plugs there exist for it.

    So far I have only take interest in the sole denoise function of it which seems to use a lot of previous script works mainly from Didée to say it the short way . I guess that's possibly also why it did not receive any "friendly slap" on his writing from La Statue du Commandeur ; the author was also clever enough not to comment his script , a wise guy ! well for all these reasons, I take it for granted it is then quite good indeed

    In your evaluation you might also want to stay at the very sole denoise level in the first place . the documentation (script reading that is) will help trace what belongs to denoise only and what is PP

    Here is one (among others) version of settings I recently used for still subtle denoise (no heavy denoise at all) ; some parameters are even redundant from default settings=low but never mind its for my sake of reminding

    MCTemporalDenoise(settings="low", thSAD=256, thSCD1=256, radius=2 , sigma=5 , bwbh=32, owoh=4, blksize=8, overlap=4, maxr=4 ,TTstr=3 , adapt=true, protect=true, enhance=false, truemotion=true, stabilize=true, chroma=true, gpu=false)

    I could explain why some parameters may look unusual even weird (!) , mentioning possibly here fft3d/mvtools blockSizes, overlap, threshold of thsCD1=thSAD, etc... but it would make an even longer post , ouch ; so use it if you want and simply see if that matches any adequate 'simple' denoise with no objectionable added blur ; while it does use of some mild PP aspects; caution = MCTD is quite slow too

    I feel ready to receive some good beating, never mind , lol

    also coming back on your preset for QTGMC you may gain quite a lot of time with staying away from SourceMatch=3 (almost useless but time and ressource greedy ) and well even MatchEnhance=0.75, NoiseProcess=1, NoiseRestore=0.7, Sigma=1.5

    I did not find MatchEnhance to provide any true benefit ; and then while interesting at first glance the "two way" idea for managing noise/grain all the way through does not really work (imho) ; so you might want to preserve picture integrity -with noise- via some other settings. The more sensible and difficult to deal with is the removegrain radius and yet it is a huge time/ressource point as well, I mean TR2 ; either keep it as 2 or go down to 1 ; I do the latter with no objectionable harm on final picture quality. In the end I find myself to almost use more the very source TGMC more than QTGMC . Still I do use SourceMatch=2 , MatchPreset="slow" , and lossless=2 which are from VIT's (qtgmc) and quite valuable addings

    Summing up , the difficult task using all these Avisynth scripts remains which one to possibly use first , after deinterlacing.

    I would say get rid of unwanted remaining noise first ; and it is a very daunting task not to do this inside a timeline ! Since you will possibly have to select each take which requires this and apply the 'adequate' script value for each one ! if you are lucky you may have little to do but otherwise it really gets cumbersome and eventually you might prefer looking for NeatVideo (for instance)

    I have not yet made my choice at the moment . I simply found I could get good result on a take by take basis using MCTD but I fear I might get bored in the long time considering >50 hours tapes...too bad

    one last word you will definitely want to denoise, and get rid of possible chroma artifacts, indeed *before* applying any color correction ! that is invert #7 and #4 such as stated in a very early post

    ps : I got (regsiter) problems with UT on my day to day machine (!?) , indeed I can look for but being lazy if you stick (here) to Huffy I would appreciate
    Quote Quote  
  29. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    you forgot to tell which camcorder (video8 ?) was used for the first sequence with Deshake use ( at Thai temple) since the soundtrack is mono.

    Could it be some VHS camcorder then , since even the earlier Video8 Sony V100, V200 , or Canon like , would feature a stereo m/s mic. Quite nice sound overall

    or maybe you simply exported stereo audio as mono... in a stereo track ! indeed


    late edit
    :
    otherwise to prevent from spending too much time at thinking which is best either D8 or Hi8 player+capture card , the subject has been treated many times with almost always the same mixed down feeling of 50/50 . Overall with a preference for D8 anyway

    It is quite possibly chimeric to expect significant improvement from the capture itself . In the end it always get to 99% transfer of what is really left on the tape . Some will say it is 'poor' , I don't agree just like I would not say that 8mm argentic is also poor. recently I was impressed by a guy ( Mattias something on Vimeo) who made a double 8mm transfer with two different apertures which brings a very impressive sensation like the contrast masking 'trick' ( it works BTW you should also consider it when it will come to editing time ; but again it has to be applied sequence by sequence not once for all )

    so here it was : https://forum.videohelp.com/threads/354425-Hi8-capture-using-Digital8-camcorder-Edge-color-issues

    it mentions the main issue of the random color inversion on the right side (only on PAL D8 versions it seems) . I had (in some rare case) up to >20 pixel wide that were plagued by this. as I refuse to crop too much , when some sequence is concerned, I accept some narrow residual color aberration up to 6 or 8 pixels as some kind of 'typical datation' of the movie a bit like 8mm argentic film transfers get recognized at first glance from their colors ....with pleasure
    Last edited by kerryann; 20th May 2016 at 12:09.
    Quote Quote  
  30. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Thank you so much kerryann for the huge amount of information you have provided .I have to read it carefully and try all yours proposals after what I will report back.The original camcorder was a Sony CCD-F500E I think it was stereo I have to see what is causing the audio to be mono.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!