VideoHelp Forum

Our website is made possible by displaying online advertisements to our visitors. Consider supporting us by disable your adblocker or Try ConvertXtoDVD and convert all your movies to DVD. Free trial ! :)
+ Reply to Thread
Results 1 to 11 of 11
Thread

Threaded View

  1. Member
    Join Date
    Jun 2019
    Location
    Australia
    Search PM
    Hi everyone

    Currently trying to learn some skills in VHS restoration but I'm a little new to it. I am currently using Virtualdub2 using multiple filters as explained below.

    My setup is as follows.

    Panasonic NV-HS830 SVHS > S-video > Panasonic DMR-ES10 (NR Off) > Component > Topfield TRF-2400.

    I have attached the Raw capture from the Topfield and my own edit in virtualdub using the following filters:
    a. Deinterlace
    b. HSV Adjust (Saturation increase to 130%)
    c. Smoother (4 points, just a fraction to remove grain)
    d. Sharpener (Slight increase)
    e. Brightness/Contract (Contrast slight increase)

    Fault/Limitation Acknowledgments


    Listing these honestly as they may be causing quality issues too.

    a. High quality composite cables instead of proper component (Still waiting for my cables to come in mail)
    b. Topfield 2400 to record (Still deciding on best capture device for Windows 10)
    c. Mid range SVHS Player
    d. Video recorded in 1996 from free to air television.
    e. I have not cleaned the VHS tape yet.

    What I'm hoping to get

    Obviously the best quality picture possible. I'd like some constructive criticism on my noobish work and how I can improve it using virtualdub. I have no idea how to use Avisynth.

    I think my edit turned out good but I think the yellow should be a little stronger and green a little lighter but not sure how to achieve both.

    Your constructive feedback is welcome please

    Raw 'capture' - https://mega.nz/#!f5kW1AQT!7OhJ_T4PUlGU9tTf3WxcVK4j2hryKJosG4_kptTvBjg
    VD Edit - https://mega.nz/#!ugkyRQCB!ts3AaThwA61sM8jeQqBgD7qRiuUFuTk7BPn5Zd58HF8

    PS. Cricket fans may be aware this video is from the Cricket World Cup 1996.
    Last edited by geordie10; 18th Jul 2019 at 06:04. Reason: Update
    Quote Quote  
  2. You're capturing as MPG is a big mistake (loss of detail, macroblock artifacts, DCT ringing). The AVI has been poorly deinteralced and has chroma blending artifacts. Your VHS deck has created severe oversharpening halos. If it has a sharpness control -- turn it down/off. The capture device may be exacerbating the halos. Your colors are probably illegal (over saturated) too (can't really tell because you converted to RGB).

    For serious restoration you need to learn AviSynth.
    Quote Quote  
  3. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Why are you deinterlacing? Is there a reason for it?

    HD capture cards do a really poor job at SD capturing. It's like buying a car to haul a load of gravel, or a truck to haul a football team. Or screwdriver to whack a nail, or hammer to insert a screw. Wrong tool for the task. You have no idea what sort of processing is happening at the component stage.
    Quote Quote  
  4. Member
    Join Date
    Jun 2019
    Location
    Australia
    Search PM
    Ok thanks for the criticisms I guess, but is there any recommendations how I fix these problems using virtualdub? Or are there any avisynth for beginners tutorials that I can learn from?

    Acknowledge mpg - working on it.
    Chroma blending artifacts - what are they? I've been trying to find some basic examples all day on Google with no luck.

    So from this I take it stay with s-video all the way.
    Quote Quote  
  5. Originally Posted by geordie10 View Post
    Chroma blending artifacts - what are they?
    The problem is that the MPEG decoder outputs interlaced YV12 (4:2:0) and VirtualDub converts that to RGB as it it's progressive YV12. This blends the colors of the two fields together. It's a little hard to see in this video because there is a lot of motion blur:

    Image
    [Attachment 49532 - Click to enlarge]


    Notice how the yellow color doesn't line up with the player's luma. It's shifted to the right here. In the other field form the frame the color is shifted to the left. Also not the horizontal stripes (enlarged in the inset). And the colors aren't pure anymore, some of the stripes are yellow, some orange, a blend of the yellow shirt and the green grass -- Looking at the two fields of one frame side by side:

    Image
    [Attachment 49533 - Click to enlarge]


    On the top is VirtualDub with the Deinterlace -> Unfold Fields Side By Side option. On the bottom is the equivalent with AviSynth. You can see that the colors are aligned with the luma properly with AviSynth. Also, if you zoom in, you'll see comb artifacts in the colors in the VirtualDub image, not in the AviSynth image.

    Originally Posted by geordie10 View Post
    is there any recommendations how I fix these problems using virtualdub?
    Most of those issues can't be fixed in VirtualDub. Some can't really be fixed in AviSynth either.

    There are some AviSynth filters for reducing oversharpening halos. But they tend to eliminate some real detail too. It's best to avoid them in the first place. Turn off the sharpening filter in your VHS deck. If you can't then find a deck without a sharpening filter or one that can be disabled.

    Details lost via MPEG compression can't be restored. Blocky artifacts can be deblocked but you're left with smooth gradients. That looks marginally better than big blocks but there's still no detail. DCT ringing artifacts can be reduced but some real detail may disappear too.

    QTGMC in AviSynth is the best detinterlacer for most material. It's very good at eliminating aliasing artifacts and keeping real edge details. Unfortuantely, it's slow and one of the most difficult third party filters to get set up because it relies on several other third party filters.

    The blended chroma problem can be avoided in AviSynth by using ConvertToYUY2(interlaced=true) or ConvertToRGB(interlaced=true). VirtualDub can handle the results of those properly.

    The problem of illegal colors is a little complex. You've probably read that YUV (more properly called YCbCr) has valid ranges of Y between 16 and 235, and U and V between 16 and 240. But not all combinations of those values are valid. In fact, only about 1/6 of the YUV colors map to valid RGB colors:


    from: https://software.intel.com/en-us/node/503873

    The big outer cube is the 8 bit YUV colorspace. The inner cube is the valid 8 bit RGB colorspace. The RGB cube is shrunken, shifted, and rotated relative to the YUV cube. All the YUV values outside that inner cube would map to RGB values less than zero or greater than 255. How those illegal colors are displayed varies from device to device. You want to avoid them when possible. And you want to avoid converting YUV to RGB before you have legalized all the YUV values.

    Originally Posted by geordie10 View Post
    Or are there any avisynth for beginners tutorials that I can learn from?
    There are the very basic instructions at the AviSynth wiki:

    http://avisynth.nl/index.php/Main_Page
    http://avisynth.nl/index.php/First_script
    http://avisynth.nl/index.php/Getting_started

    From there it gets more complex. But you can get help here when you're ready.

    Originally Posted by geordie10 View Post
    So from this I take it stay with s-video all the way.
    Note that s-video is term for analog video -- where the luma (brightness, Y) is carried on a separate wire from the chroma (colors, U and V). Once digitized they are usually called YUV or, more properly, YCbCr. YUV usually has the chroma (U and V) channels at lower resolution than the luma. This is based on the theory that the human eye has less resolution for color than for brightness. Hence you have YUV 4:2:2 and 4:2:0. With the former the chroma has half the horizontal resolution of the luma (for example 720x576 luma and 360x576 chroma), with the latter the chroma has half the horizontal resolution and half the vertical resolution of the luma (for example 720x576 luma and 360x288 chroma).

    Too much to read and digest? And it's still just scratching the surface. Ask for clarification if you need it.
    Quote Quote  
  6. I looked at the manual for your VCR.

    1. Set the sharpening to "Soft." It will not actually soften the picture, but instead will turn off the sharpening circuits. If you really want to add sharpening, do that after the capture, using advanced sharpening software. The sharpening done digitally is massively better than the simplistic sharpening found in VCRs.

    2. Turn off the "3D DNR" function. Page 9 of the manual states: "However, when performing editing,
    press the [3D DNR] (40) to turn this function off ..."

    You sound as though you are quite serious about getting good results from your captures and restoration and you appear to be plenty sharp enough to get AVISynth up and running (it seems intimidating, but actually isn't all that hard). There really isn't any comparison between what you can do with VirtualDub filters and what you can do with AVISynth. This is due to the fact that AVISynth makes it much easier to perform functions which require input from multiple frames (like temporal noise reduction).
    Last edited by johnmeyer; 9th Jul 2019 at 17:41. Reason: added last paragraph; later, typo
    Quote Quote  
  7. Member
    Join Date
    Jun 2019
    Location
    Australia
    Search PM
    Thanks Jagabo and Johnmeyer. That's is really good advice and what I am look for so thankyou.

    I do appreciate input from professional eyes to give guidance as opposed to some questionable YouTube tutorials.

    Yes I have done some reading on Avisynth and it doesn't seem too overwhelming if you start learning small and progressing. So from here I'll spend some time learning Avisynth and hopefully get back with and update on my capture. Obviously once I get a proper capture card capable of capturing lossless AVI.

    Thanks again for the replies !
    Quote Quote  
  8. By the way, you'll want a way to view the results of your AviSynth scripts. I use VirtualDub(2) to open AVS scripts with File -> Open Video File...

    You'll need 32 bit VirtualDub to work with 32 bit AviSynth, or 64 bit VirtualDub to work with 64 bit AviSynth. 64 bit AviSynth+ is stable enough now and there are enough third party 64 bit filters to work with it -- so I'd recommend 64 bit VirtualDub2 and 64 bit Avisynth+. http://avs-plus.net/

    Some things are hard to do without a GUI (like color correction). So you may sometimes want to use VirtualDub for further processing.
    Quote Quote  
  9. Member
    Join Date
    Jun 2019
    Location
    Australia
    Search PM
    Hi again

    I've spent a bit of time doing some more research and tracking some some new equipment for my captures. As a result I have produced the following sample for review. I do hope to get some constructive feedback on how I could do some post processing on this file in Avisynth. I am a newb with this so some examples of script I should use would be excellent.
    See sample link below.
    https://mega.nz/#!7h1mWQiK!hRJrYpWgdXrxbqbQ017__VOMjQEU7_TATUKnqm_a6VY

    The setup I am using;
    JVC HR-S5700AM SVHS Player > S-Video > Panasonic DMR-ES10 > S-Video > Diamond VC500

    Thanks for any replies.
    Quote Quote  
  10. I suggest you do this:
    1- deinterlace the images (with qtgmc in lossless mode with light denoising or TFM or else) / Objective: getting full frames as sharp as possible
    2- sharpen (with filters like finesharp, msusharpen or else)/ you don't want to sharpen blurrier images than they already are, do you ?
    3- denoise as much as needed / not too much careful....
    4- Color correction if needed (but it would be wiser to do this as a second step imo)
    Quote Quote  
  11. First, you should capture as YUV with 4:2:2 chroma subsampling, not RGB. The latter risks losing the darkest blacks and brightest whites (though it was ok in this cap). The video isn't really interlaced but you could use QTGMC for other edge cleanup and light denoising.

    Code:
    AviSource("warriors vc500 jvc test.avi") 
    ConvertToYV12(interlaced=true)
    vInverse() # reduce combing from time base errors
    Crop(4,0,-12,-0) # maybe the bottom too, to get rid of the head switching noise
    FineDehalo(rx=3.0, ry=1.0) # reduce over sharpening halos
    QTGMC(InputType=1, sharpness=1.0, EZDenoise=1.0, DenoiseMC=true) # clean up edges, light denoising
    MergeChroma(last, aWarpSharp(depth=15)) # sharpen chroma
    Brightness and colors may need adjusting but I hesitate to make suggestions based on this short sample where it's not clear how they should look.
    Image Attached Files
    Quote Quote  



Similar Threads