VideoHelp Forum
+ Reply to Thread
Page 7 of 7
FirstFirst ... 5 6 7
Results 181 to 197 of 197
Thread
  1. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    A few tweaks with your test32 mpg. And a sample of crushed detail. Nice work on some of those shots; as usual, VHS has those crummy level and color changes. Always happens. Pain in the neck. Here are some before & after tweaks:

    Image
    [Attachment 11927 - Click to enlarge]

    Image
    [Attachment 11928 - Click to enlarge]


    Note the lower-left image in the lower group of frames. The kids is too blue, but the major problem is that his hair is a purplish-black blob instead of a lighter color (his hair looks dark brown in other shots). His t-shirt looks blacked-out but is navy blue in other shots.Had to play with levels and color in YUV before trying to tweak these close-ups of him. Visually these shots of the kid don't look like the other shots -- very common with VHS. Dont'cha just love it?

    In the image below I purposely brightened only the color and luma below RGB 64. Note the darkest areas in his hair, darker shadows on the denim vest, and the dark t-shirt under it. These read down near RGB 0-0-0 in the original post. Colors in that area were crashing against the left-hand side of histograms, indicating crushed darks. You can raise the brightness level of those areas all you want; they will always look dark gray, no color change, and no detail. In this image the RGB 0-0-0 sections were raised to RGB 34-34-34. That will never change hue (always gray) and no detail can be retrieved.

    Image
    [Attachment 11929 - Click to enlarge]
    Image Attached Thumbnails Click image for larger version

Name:	Before_After_A.png
Views:	209
Size:	1.24 MB
ID:	11931  

    Click image for larger version

Name:	Before_After_B.png
Views:	231
Size:	1.29 MB
ID:	11932  

    Click image for larger version

Name:	f395_crush.jpg
Views:	198
Size:	38.8 KB
ID:	11933  

    Last edited by sanlyn; 13th Apr 2012 at 15:40.
    Quote Quote  
  2. sanlyn, those sample images look great. I was mainly targeting the bench scene when I made that clip. When I figure out how to cut up the different scenes and put them back together in a decent workflow, I'll be able to work with each more directly. Do you have the settings saved for the samples you posted?

    When I get my second s-video cable, I'm going to do the full capture. Is it typical to have VirtualDub automatically break-up the capture into smaller chunks? I'm thinking 5-10GB files may be easier to handle.
    Quote Quote  
  3. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Originally Posted by videeo View Post
    When I figure out how to cut up the different scenes and put them back together in a decent workflow, I'll be able to work with each more directly. Do you have the settings saved for the samples you posted?
    The settings are on another PC. Will load them later today when I get home.

    It is in no way unusual to have to have to work with separate scenes in VHS. That's the just the way VHS is; those voltage changes on tape look fairly OK on the old CRT's, but once you get into digital things have to be more precise. Digital is not forgiving of level and color shifts. At least, once you find that a series of similar shots has similar problems, the same filters can usually serve for all shots of that kind. But I can't tell you how many times I've had to work that way.

    I "cut" the scenes in Avisynth or in VirtualDub. But be careful, for each shot you need maybe 1/2 second of extra trailing frames so that when you join them you won't get "holes" in the audio at scene changes (sometimes you get them, sometimes you don't). Sometimes I join them with Trim() in Avisynth a few scenes at a time, then join those segments into biugger ones. That saves me from having to diassemble the whole video if something needs more work. After I get my final DVD, I get rid of all thjose chuncks (but save the settings and scripts you came up with. There are many ways you could do it. But no matter how you work it, it's a pain. When I cut up recordings off digital cable, I never have to do all this levels and color drudgery.

    Originally Posted by videeo View Post
    When I get my second s-video cable, I'm going to do the full capture. Is it typical to have VirtualDub automatically break-up the capture into smaller chunks? I'm thinking 5-10GB files may be easier to handle.
    VirtualDub breaks up the video by default. For a continuous file, in the capture utility top menu go to "Capture. . ." and uncheck "enable multisegment capture". I capture into one big file, then break it into 4-GB chunks for archiving to disc. With a really long video (I worked with 3.5 hours once), I stop capture early, then restart in a new file. I capture to a 2nd hard drive in my PC. Then I copy it to an external drive and pull it off in pieces for work.
    Last edited by sanlyn; 21st Mar 2014 at 20:34.
    Quote Quote  
  4. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Originally Posted by videeo View Post
    Do you have the settings saved for the samples you posted?
    Most of the corrections were done with gradation curves. Their settings are difficult to describe. Listing them would print out a full 8.5x11 page. I saved them as VirtualDub .vcf files. To use a vcf you open an AVI in VirtualDub, then click "File. . ." and "load processing settings...", then tell VirtualDub where the .vcf is located. .vcf's don't have to be in your plugins; you can keep them anywhere.

    The filters are different for each of the 4 scenes I posted. Caution: the shots of the big kid in denim have a different levels and color balance than the other scenes. To prevent crushing darks and blowing out highlights, I had to run a a script and use the vcf's ONLY for those shots. The other vcf's were used by opening the test32.mpg directly in VirtualDub (or you can use a .d2v project and open it that way). Remember, I didn't have the original test32 AVI capture, just the filtered mpg. But they will likely work OK with the AVI's as they were before you encoded them to MPG.

    The script uses SmoothLevels() and RemoveSpots. If you didn't notice, there are prominent white spots in the big kid's hair in frames 390 and 391 of test32.

    Below are two attachments. The AVS script was used to prep levels and out-of-bounds colors in test32.mpg for only the closeup shots of the kid in denim. The two vcf files named "denim" were used only or those shots.

    The .zip attachment contains:

    - All the plugins needed to run RemoveSpots. Copy all of the plugins in the zip'd "RemoveSpots" folder into Avisynth plugins.

    - The complete SmoothAdjust() package; the .dll contains the SmoothLevels() function, so just copy the .dll plugin to Avisynth. But NOTE: the SmoothAdjust folder has two sets of dll's, one set for Intel or Athlon, and 320-bit and 64-bit systems. Don't copy both sets; just copy the dll's for your system. They are described in the SmoothAdjust readme.txt file, near the top of the text.

    - A couple of VirtualDub plugins that were used, and you might not have these. Copy everything in the zip'd "VirtualDub" folder into your VirtualDub plugins, including the .txt file.

    - All the .vcf files files used, which will load the filters and their settings, in the proper order. The titles are self-explanatory, but to avoid confusion:
    boy_f.vcf has settings for shot with the small blond-haired boy.
    girl_c.vcf is for the little strawberry-blonde gal.
    group_g.vcf is for the group shots. They should work, even if both group shots really aren't alike.

    denim_p.vcf is for the big dark-haired kid's closeups, and demin_end_b.vcf is for the last shot of that kid at the end of the test32 sample. These "denim" settings work only after running the sample clip with the AVS script.

    Like I said, restoration and capture are two different worlds.
    Image Attached Files
    Last edited by sanlyn; 21st Mar 2014 at 20:34.
    Quote Quote  
  5. Thanks sanlyn, I'll be capturing later today. I think I'll start a new thread in the restore section of the forum once I get started on that part of the project.
    Quote Quote  
  6. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Good idea. You might find a different crowd (or you might run into the same old fuddy-duds, like me!). I was going to post some sample Trim() scripts, but let's save that for the new captures. No sense changing every script in sight for a new process. I think you've picked up a lot and making some better caps. Good luck.
    Last edited by sanlyn; 21st Mar 2014 at 20:34.
    Quote Quote  
  7. I think this question still pertains to capture... Before working on restoration, I'm thinking I should first IVTC the whole thing before splitting it into smaller sections for tailoring the restoration. I see the advantage here of working with fewer frames and speeding up the subsequent rendering with filters in VirtualDub. What would be the best way to do it as to minimize colorspace conversions? Just in AviSynth or AviSynth/VirtualDub? Should any other filtering be done with the IVTC step that would be common to all the footage?

    I was also wondering, is there a known method to take the redundant frames in IVTC and average them together to reduce noise in those frames?
    Quote Quote  
  8. Originally Posted by videeo View Post
    I was also wondering, is there a known method to take the redundant frames in IVTC and average them together to reduce noise in those frames?
    Are you sure you'd want to do that? Wouldn't that create a 'strobing' effect from the resulting one frame in four being 'cleaner' than the rest? Wouldn't you want to denoise the whole thing more-or-less equally?

    And yes, I'd want to IVTC the thing before doing anything else. The best (and only decent) way to do it is with one of the AviSynth IVTC filters.
    Quote Quote  
  9. Originally Posted by manono View Post
    Wouldn't that create a 'strobing' effect from the resulting one frame in four being 'cleaner' than the rest? Wouldn't you want to denoise the whole thing more-or-less equally?
    Good point, I was just wondering if it is done...I'm thinking of using Neat Video, so it would probably be disruptive anyway.

    Originally Posted by manono View Post
    And yes, I'd want to IVTC the thing before doing anything else. The best (and only decent) way to do it is with one of the AviSynth IVTC filters.
    Yes, I have been using AviSynth and it works pretty well. One thing I forgot to ask is if it's better to just to the IVTC + meta filtering in AviSynth AND also recompress with Lagarith within AviSynth as well. I have not recompressed in AviSyth before.

    Here are the steps I am thinking of doing in AviSynth before noise/color filtering in VirtualDub:

    1) IVTC
    2) Meta filter (maybe sanlyn might have advice based on the clips he has worked on)
    3) Recompress to Lagarith

    Other?: crop, borders, colorspace, etc.
    Last edited by videeo; 18th Apr 2012 at 15:15.
    Quote Quote  
  10. Originally Posted by videeo View Post
    One thing I forgot to ask is if it's better to just to the IVTC + meta filtering in AviSynth AND also recompress with Lagarith within AviSynth as well.
    I do it like that all the time if my filtering is processor intensive and/or I intend to filter in stages.
    Quote Quote  
  11. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    The IVTC/Stab scripts previously posted should work as well as before. That should be your first step.

    I don't know what you mean by "recompress". Whenever you run a script thru VirtualDub, the clip is decompressed in VirtualDub and saved by default as uncompressed RGB24, unless you specify otherwise. If you do specify a colorspace and a compressor, then in that case the frames you are viewing are decompressed and converted to RGB only for display; it will otherwise be saved in the colorspace and compressor you've chosen.

    If all you're doing in VirtualDub is running a script and viewing the results, without loading any VirtualDub filters, and you want to save the results of Avisynth processing in the original colorspace that Avisynth is using, do this: in the 'Video" menu select "color depth". Choose the color depth you want to save in. If your script was using YV12, then select YV12. Then under "Video" select "compression...". Your compressor for YV12 would usually be Lagarith. Be sure to click Lagarith's "configure" button and select the color space you want for the compression. You check on that because Lagarith's default is RGB. Then to save the results of the script without any VirtualDub processing, choose "fast recompress". Then "Save AVI...". If you use "Full processing mode", your clip will be converted to RGB whether you want it or not.

    Don't load any VirtualDub filters if you want to save back to YUV. Any loaded VirtualDub filter will first convert the clip to RGB, then your compression step will perform yet another conversion -- avoid that.

    It works this way:

    Fast recompress = save the results of the avs script in the original colorspace used by the Avisynth script.
    full processing mode = run VirtualDub filters. This will automatically convert to RGB.

    When you want to go from YUV to to RGB, convert to RGB in AviSynth this way:

    AviSource("path\filename")
    ConvertToRGB32(matrix="Rec601",interlaced=true) # <- "interlaced = false" if video is progressive)

    Don't switch colorspaces back and forth. That is, don't use YV12, then RGB, then YV12, then RGB again, etc. If you have work to do in YUV in AviSynth, complete all that work and scripts in YUV first. If you then need some RGB filters, go to RGB and stay there until you're finished with RGB. Eventually you'll go back to YV12 if you are going to encode with HCenc. Do that in Avisynth. If your work is already in YV12, you don't need to convert again.
    Last edited by sanlyn; 21st Mar 2014 at 20:34.
    Quote Quote  
  12. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Originally Posted by videeo View Post
    ...I'm thinking of using Neat Video, so it would probably be disruptive anyway.
    In the context in which you mentioned it, yes it would. From what I've seen of your clips so far, NeatVideo would be overkill. It works only in RGB.
    Last edited by sanlyn; 21st Mar 2014 at 20:35.
    Quote Quote  
  13. Here's the script I have for the IVTC. I will open it in VitualDub and save in Lagarith YV12 using "Fast Recompress" mode. Would smoothlevels be appropriate as a meta-filter?

    Code:
    #- ----------------------------------------------
    #   Source is progressive hard-telecined 3:2 YUY2
    #- ----------------------------------------------
    
    AviSource("E:\videeo\capture\full-capture-old-tape-04-17-2012.avi")
    
    #- -------------------------------------------------------
    #   Inverse telecine/decimate. Result is progressive video
    #-  in its original state as 23.97 fps movie film
    #- -------------------------------------------------------
    AssumeTFF().TFM()
    TDecimate()
    
    ConvertToYV12(interlaced=false)
    
    #SmoothLevels()
     
    Crop(0,0,0,-8)
    AddBorders(0,4,0,4)
    Quote Quote  
  14. Full processing mode in VirtualDub does not convert to RGB if you're not performing any filtering (a few years ago it did but it has gotten smarter since then). Video will stay in the colorspace it comes in as unless you force it to something else with Video -> Color Depth.

    Be careful with Lagarith and YV12 video. It always treats YV12 as progressive. After IVTC your video will be progressive so that's not a problem here. But if you're working with interlaced YV12 Lagarith will screw up the chroma channels. Of course, VirtualDub screws up the chroma of interlaced YV12 too. So it's better to convert interlaced YV12 to YUY2 or RGB before giving it to VirtualDub. Basically, never use VirtualDub with interlaced YV12.
    Last edited by jagabo; 18th Apr 2012 at 18:12.
    Quote Quote  
  15. Part of the reason I asked about direct conversion to Lagarith within AviSynth is to avoid confusion about the internal issues of VirtualDub.
    Quote Quote  
  16. AviSynth can't compress with Lagarith or any other codec.
    Quote Quote  
  17. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Originally Posted by jagabo View Post
    But if you're working with interlaced YV12 Lagarith will screw up the chroma channels. Of course, VirtualDub screws up the chroma of interlaced YV12 too. So it's better to convert interlaced YV12 to YUY2 or RGB before giving it to VirtualDub. Basically, never use VirtualDub with interlaced YV12.
    Good to know that about Lagarith. In any case, I always use Avisynth to get into RGB. regardless of where the clip is coming from.
    Last edited by sanlyn; 21st Mar 2014 at 20:35.
    Quote Quote  



Similar Threads

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