VideoHelp Forum

Try DVDFab and download streaming video, copy, convert or make Blu-rays,DVDs! Download free trial !
+ Reply to Thread
Page 9 of 10
FirstFirst ... 7 8 9 10 LastLast
Results 241 to 270 of 285
Thread
  1. Originally Posted by oln View Post
    Yup, made a little progress on NTSC color today. I spent way to long not realizing that the color under frequencies of PAL and NTSC are not the same, but now that I accounted for that the phase of the color signal was as expected. Still needs way more work to look good though.
    Awesome work as usual! Hopefully someone who can code will come along to help. Your time spent on this is much appreciated by everyone here.

    Originally Posted by dellsam34 View Post
    If there is a time where this project could have a big boost to materialize it is now where everyone is indoor due this pandemic.
    Agreed, I'm working from home for the foreseeable future. I had to get my VCR and stuff off my desk so I'm glad I got some captures I can do test decodes with now.
    Quote Quote  
  2. I built the latest source and did a decode, but the output tbc file is equivalent from the previous source. Is there a flag I need to use to decode chroma?
    Quote Quote  
  3. The chroma is output separately as a .tbcc file at the moment as a workaround. Needs to be merged manually after decoding, I've been sort of doing it manually but I should add a script to automate it properly. I think zcooger made one that worked with ffmpeg and PAL at one point.
    Quote Quote  
  4. Got it. I'm happy to wait for a merge script if you plan to write one.
    Quote Quote  
  5. I renamed the .tbcc files to .tbc, saved some screenshots of the frames with good hue from ld-analyse, and composited the luma and chroma images in Gimp. All of of our samples look really good! Spider-Man and my broadcast sample look terrible until you run them though ld-chroma-decoder and deinterlace them.






    Last edited by Titan_91; 9th Apr 2020 at 07:04.
    Quote Quote  
  6. This is the script I use mentioned in the topic:

    Code:
    #!/bin/sh
    
    rm -f "/media/veracrypt1/$1_luma.avi"
    rm -f "/media/veracrypt1/$1_chroma.avi"
    rm -f "/media/veracrypt1/$1_luma.rgb"
    rm -f "/media/veracrypt1/$1_chroma.rgb"
    rm -f "/media/veracrypt1/$1_luma.avi"
    rm -f "/media/veracrypt1/$1_chroma.avi"
    rm -f "/media/veracrypt1/$1.tbc.json"
    rm -f "/media/veracrypt1/$1.tbcc.json"
    rm -f "/media/veracrypt1/$1.tbc"
    rm -f "/media/veracrypt1/$1.tbcc"
    
    /home/zcooger/Desktop/vhs-decode/vhs-decode --cxadc3 --othertrack -p "/media/veracrypt1/$1.r8" "/media/veracrypt1/$1"
    
    /home/zcooger/Desktop/vhs-decode/tools/ld-chroma-decoder/ld-chroma-decoder -f mono "/media/veracrypt1/$1.tbc" "/media/veracrypt1/$1_luma.rgb"
    cp "/media/veracrypt1/$1.tbc.json" "/media/veracrypt1/$1.tbcc.json"
    /home/zcooger/Desktop/vhs-decode/tools/ld-chroma-decoder/ld-chroma-decoder -f pal2d "/media/veracrypt1/$1.tbcc" "/media/veracrypt1/$1_chroma.rgb"
    rm -f "/media/veracrypt1/$1.tbcc.json"
    
    ffmpeg -y -f rawvideo -r 25 -pix_fmt rgb48 -s 928x576 -i "/media/veracrypt1/$1_chroma.rgb" -f rawvideo -r 25 -pix_fmt rgb48 -s 928x576 -i "/media/veracrypt1/$1_luma.rgb" -filter_complex "[0:v]format=yuv444p[0v];[1:v]format=yuv444p[1v];[0v][1v]mergeplanes=0x100102:yuv444p[v];[v]scale=768x576[u];[u]bwdif=1[o]" -map "[o]" -c ffv1 -pix_fmt yuv420p -shortest "/media/veracrypt1/$1_vhs.avi"
    Save as .sh file, pass one argument of file name you have captured the tape to without extension like
    Code:
    sudo '/home/zcooger/Desktop/vhs-decode.sh' mytape
    You have to modify it a bit for your needs - directories, frame rate, resolution and the color system.
    The script has been optimized with "rm" commands to consume as little disk space as possible during processing. If you do not wish to lose the data during command rerun remove them. You can run the script during tape capture even if it did not finish.
    Code:
    $1
    in the script is replaced with the argument of your file name.
    Last edited by Zcooger; 2nd Apr 2020 at 02:29.
    Quote Quote  
  7. Awesome, will give the script a try.
    Quote Quote  
  8. I went over the script and made all needed changes. But vhs-decode is crashing for some reason, complaining about the phase rotation argument. I'll need to figure that out. But I did try --othertrack and got color from A Bug's Life.
    Last edited by Titan_91; 9th Apr 2020 at 06:55.
    Quote Quote  
  9. Originally Posted by Zcooger View Post
    Code:
    ffmpeg -y -f rawvideo -r 25 -pix_fmt rgb48 -s 928x576 -i "/media/veracrypt1/$1_chroma.rgb" -f rawvideo -r 25 -pix_fmt rgb48 -s 928x576 -i "/media/veracrypt1/$1_luma.rgb" -filter_complex "[0:v]format=yuv444p[0v];[1:v]format=yuv444p[1v];[0v][1v]mergeplanes=0x100102:yuv444p[v];[v]scale=768x576[u];[u]bwdif=1[o]" -map "[o]" -c ffv1 -pix_fmt yuv420p -shortest "/media/veracrypt1/$1_vhs.avi"
    I have 3 issues with my ffmpeg output. The first being the resulting file is 59.94fps instead of 29.97. The other two are what appear to be deinterlacing or blending (scanlines are gone) with ghosting chroma frames. Do I need to tweak something?

    Quote Quote  
  10. Get rid of deinterlacing filter:
    Code:
    [v]scale=760x488[u];[u]bwdif=1[o]" -map "[o]"
    >
    Code:
    [v]scale=760x488[u]" -map "[u]"
    Change chroma format at output:
    Code:
    -pix_fmt yuv420p
    >
    Code:
    -pix_fmt yuv422p
    or
    Code:
    -pix_fmt yuv444p
    I would change first the chroma output and leave the deinterlacing filter to see if the chroma delay is still present.
    Last edited by Zcooger; 8th Apr 2020 at 09:19.
    Quote Quote  
  11. Thanks, I'm not familiar with ffmpeg. I changed the color format to yuv422p and yuv444p but am still seeing the ghosting. I disabled the deinterlace filter but the chroma delay is still present. In this screenshot it's the pink pencil outline on the right.

    Last edited by Titan_91; 9th Apr 2020 at 11:45.
    Quote Quote  
  12. Is there a flag I can set to skip the tbcc file generation? For example, I have some sources with extra grayscale data in the blanking intervals I plan to try and decode and don't need chroma in this case for time and disk space reasons.
    Quote Quote  
  13. You can remove this line from the script - this way you run only the luminance decoder:
    Code:
    cp "/media/veracrypt1/$1.tbc.json" "/media/veracrypt1/$1.tbcc.json"
    /home/zcooger/Desktop/vhs-decode/tools/ld-chroma-decoder/ld-chroma-decoder -f pal2d "/media/veracrypt1/$1.tbcc" "/media/veracrypt1/$1_chroma.rgb"
    rm -f "/media/veracrypt1/$1.tbcc.json"
    And change the ffmpeg command to:
    Code:
    ffmpeg -y -f rawvideo -r 25 -pix_fmt rgb48 -s 928x576-i "/media/veracrypt1/$1_luma.rgb" -filter_complex "[0:v]format=yuv444p[v];[v]scale=768x576[o];" -map "[o]" -c ffv1 -pix_fmt yuv400p "/media/veracrypt1/$1_vhs.avi"
    Last edited by Zcooger; 15th Apr 2020 at 07:15.
    Quote Quote  
  14. Thanks, but I'm wondering if I can skip the chroma decoding and tbcc file in vhs-decode.
    Quote Quote  
  15. Originally Posted by oln View Post
    Originally Posted by Titan_91 View Post
    I tried to load in a short clip of Arthur to compare 28.6 and 35.8, but a lot of frames were dropped and ld-analyze shows a wrong output. The others I have decoded fine for how the software is right now. Oln do you want the Arthur samples to help troubleshoot?
    Latest git decodes them fine now, needed to increase the maximum offset the code that looks for the hsync line starts allowed (there won't be a lot of deviation on that on laserdiscs obviously). The head switch being in the vsync area made the difference between one hsync position and the next compared to the normal line length very large and messed up he rest of the field.
    I hate to bring this up again but the issue has returned for me, but only for a specific set of videos. I'm trying to help someone decode the speech audio from various TV Teddy (Teddy Ruxpin) tapes. Some fields are dropped when decoding and the some of the surviving ones look like this:



    My understanding is the head switch is happening in the vertical blanking interval. Hopefully this isn't a side effect of my machine, I would guess it's more along the lines of how the source was recorded. However, keep in mind my VCR has a slight DC offset at the head switching points if you think that could be it. Maybe there is a way to remove this bias from the FM signal during the first step of decoding?

    I can also provide a sample of this if needed.
    Last edited by Titan_91; 20th Apr 2020 at 12:41.
    Quote Quote  
  16. Member
    Join Date
    May 2001
    Location
    Bolton, UK
    Search Comp PM
    I won’t be surprised if things are getting confused because of macrovision copy protection within the signal
    Quote Quote  
  17. I don't see any Macrovision in these videos, though A Bug's Life does have it but decodes mostly fine now. The shaded bar on the left side is the Teddy toy's audio subcarrier. I initially thought this was causing it, but the skew also occurs in scenes without the subcarrier (bar) being present. For example, the first minute of the video doesn't have the bar and Teddy isn't talking.
    Quote Quote  
  18. Member
    Join Date
    Mar 2019
    Location
    Henderson, KY, USA
    Search PM
    Loving the work everyone's putting into this. I got some questions about the project.

    Could I get this working with any VCR?
    Will this work on Windows 10? I don't have a means of using Linux at the moment.
    Quote Quote  
  19. Originally Posted by ENunn View Post
    Loving the work everyone's putting into this. I got some questions about the project.

    Could I get this working with any VCR?
    Will this work on Windows 10? I don't have a means of using Linux at the moment.
    Alright to answer your questions:

    1. Yes, but results vary depending on how your model works and if it amplifies before or after the RF tap point

    2. No. As of right now, vhs-decode only works on Linux distributions. I recall hearing something about the Domesday Duplicator software having an experimental Windows branch, but that's of course only if you're using the DdD hardware, and even then you'd still need to decode it.
    Quote Quote  
  20. Something is causing issues with hsync detection in the teddy sample judging by the image. It may be as simple as just having to up the hsync tolerance a bit more, or it may be something more causing issues, still needs work. There is also something not quite right with the NTSC luma foroutput in high brigthness areas (as you can see on the audio bit on the left), so that may also be causing issues still.

    Haven't had the opportunity to tinker with vhs-decode recently, but will look into this when I've had time.

    It should in theory be possible to compile/run the software decoding part on windows, though I haven't tested it myself. The capture end would be more tricky as one would need to develop windows drivers for it which is a bit of a mess. Driver development on linux is much more straight forward.
    Quote Quote  
  21. As always we appreciate your continued work on this. As for the luma issue, I did examine some tbc files via Audacity using 16-bit raw data mode. I'm seeing the luma amplitude is clipping and rolling back over to zero in some areas.


    Last edited by Titan_91; 22nd Apr 2020 at 18:12.
    Quote Quote  
  22. Member dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Paris Ca, 92345 Mexico
    Search PM
    I just remembered that some of the U-Matic decks have a BNC connector on the back labeled "Tape out" which is the RF signal read by the video drum (not to be confused by DUB out), Wouldn't be easier to use one of those decks if access to a deck is possible to test with a good legal RF signal? Note that Y/C frequencies are little higher than of a VHS and also no Hi-Fi track just video.
    Quote Quote  
  23. I think you have the wrong thread.
    Quote Quote  
  24. Member dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Paris Ca, 92345 Mexico
    Search PM
    Originally Posted by Titan_91 View Post
    I think you have the wrong thread.
    Either he was drunk or he just woke up from a dream and posted it right away, So he might never come back here since he has no recollection of posting such a thing.
    Quote Quote  
  25. Originally Posted by Titan_91 View Post
    As always we appreciate your continued work on this. As for the luma issue, I did examine some tbc files via Audacity using 16-bit raw data mode. I'm seeing the luma amplitude is clipping and rolling back over to zero in some areas.
    Yeah that will probably confuse sync detection as well. I think the luma spikes is probably caused in the demodulation process, they shouldn't be there. Need to play around with the filters a bit.
    EDIT: Current settings seems to cut of a bit on the top (brightest frequencies) yea
    Last edited by oln; 15th May 2020 at 12:54.
    Quote Quote  
  26. Are you just needing to adjust the bandwidth or centering of your luma filter so the demodulated white levels don't peak out? That would also result in slightly more dynamic range/contrast right?
    Quote Quote  
  27. Well there is multiple stages to it, filtering on the rf signal to get rid of signal that is outside the frequency band of the rf-modulated luma, equalisation of the luma signal after demodulation to undo the emphasis that was added to high frequencies before recording (deemphasis). Finally there's the Y Cb Cr to RGB conversion which can also end up clipping if parts are too bright, that part is pretty simple to adjust though.

    The rf filtering is the most tricky part, you want to get rid of any signal outside of the modulated luma signal but not any more. I think VCRs typically do some equalization and automatic gain control on the rf before demodulating but the exact details are not very clear. Black and white streaks is one symptom of the demodulation process picking up noise that shouldn't be there, or not picking up the right frequency.
    Quote Quote  
  28. I did some filter tweaking recently which made the teddy capture decode a lot better. It's on a separate branch in the repository right now as an upstream altered something related to chroma decoding and broke NTSC color for vhs-decode which I didn't fix yet.
    Image Attached Thumbnails Click image for larger version

Name:	teddy_better.png
Views:	99
Size:	258.0 KB
ID:	53820  

    Quote Quote  
  29. Great work as usual! Is there an option to decode to full 525 line frames without cropping? I can decode the TV Teddy video material but the audio subcarrier may be cut off. If I can extract some audio I'll share it here.
    Quote Quote  



Similar Threads