How to extract text VBI channels with CC Decoder?

[boomereboot]

I find it funny how in one of my first threads I talked about wanting to get some additional data from the VBI, and I wondered about Teletext and also the TEXT based service that was short-lived in the 80s on ABC (Program Listings Update Services) and certain PBS affiliates.

It has come to my attention that a recent fork on the GitHub cc_decoder that was made years ago, has been updated to support these TEXT channels!

Here's how to do it:

1.) You need a capture card that would be able to capture the VBI at the top. The datalines MUST be visible and interlaced for this to work, it can be any video file, as long as it's interlaced with line 21 datalines at the top.

I have found that a few cards, like those Dazzle DVC100 models and Domesday Duplicator cards seem to output it well. I believe there are other cards that display the VBI at the top, but these two are the ones that come to mind. Even if here they're considered not the best for Video capture, for VBI support like this I guess you could consider them as useful?

2.) Install the latest fork of cc_decoder on github: https://github.com/eshaz/cc_decoder/

3.) Install Python and modules like Pillow (I use a Linux computer to run Python scripts.

4.) Once installed, find your video file with the interlaced signal and line 21 datalines at the top.

Run this command:

python cc_decoder.py --ccformat text --bitlevel 60 PATH/TO/VIDEO_FILE.mpg -o PATH/TO/TXT_FILE

I have provided my two sample text documents from T1/T2 that came from one of my Saturday Morning Cartoon VHS tapes attached below.

Hope this helps for those who want to extract an extra bit of metadata from VHS tapes. The extraction time for these were very quick too.

Also if there's any other cards out there that capture the VBI by putting it at the top of the screen in the video signal, PLEASE let me know I have so far tested this method on NTSC VHS tapes.

[Gary34]

Here’s the other thread he had in case anyone else is looking for how to get CC information.
https://www.digitalfaq.com/forum/vid...zle-model.html

[Aya_Rei]

I've tried messing around trying to capture VBI data. I can use a Sony DCR-SR100 HDD camcorder to record a tape onto it then extract the close captioning data from it's mpg file on the hard drive.

In terms of cards, I've had partly success using a Hauppauge WinTV-HVR-950. Tried using an IO-Data USB2 and while it does have a VBI pin, connecting it the same was as the WinTV does not dump the VBI data using GraphStudioNext.

I can use the WinTV card to successfully dump the data as a raw file and convert it to SCC subtitles, I just can't get this avisynth script to work where it reads that raw data as a VBI signal that can placed on top of the video. I always get "file size is too small error"

Would love to get that to work as I tend to always edit down any TV recording tapes I get.

Quote:

Originally Posted by boomereboot

Here's how to do it:

1.) You need a capture card that would be able to capture the VBI at the top. The datalines MUST be visible and interlaced for this to work, it can be any video file, as long as it's interlaced with line 21 datalines at the top.

May we see a raw screenshot of a tape of yours with the VBI data on top of the active image? I'm wondering if it needs to be the full VBI signal or just like two pixels of it.

[boomereboot]

I've tried messing around trying to capture VBI data. I can use a Sony DCR-SR100 HDD camcorder to record a tape onto it then extract the close captioning data from it's mpg file on the hard drive.

Quote:

Originally Posted by Aya_Rei

In terms of cards, I've had partly success using a Hauppauge WinTV-HVR-950. Tried using an IO-Data USB2 and while it does have a VBI pin, connecting it the same was as the WinTV does not dump the VBI data using GraphStudioNext.

I can use the WinTV card to successfully dump the data as a raw file and convert it to SCC subtitles, I just can't get this avisynth script to work where it reads that raw data as a VBI signal that can placed on top of the video. I always get "file size is too small error"

I think the proper solution for the WinTV method would be for CC Extractor to have it implemented in it. You can somewhat get the TEXT data when importing it in VideoReDo, but it's wonky on how it displays it really. CC Decoder for what I've seen just reads the VBI at the top and uses FFMPEG and Pillow to decode the VBI signals, frame by frame, hence why I also think the reason for it being interlaced is important. If anything I've noticed CC Decoder supports other ways of decoding the VBI either way, since Dazzles, vhs-decode and I think if the video file has a VBI on the top, would work here.

Quote:

Originally Posted by Aya_Rei

May we see a raw screenshot of a tape of yours with the VBI data on top of the active image? I'm wondering if it needs to be the full VBI signal or just like two pixels of it.

Sure. My apologies for leaving that out. The attached image file below I have given an example from just a DTV converter box, just to show the VBI at the top. I think certain VCR's seem to strip this until they play the tape back. However I have ran my Panasonic VCR into my DVD recorder and they seem to always display the VBI. If anything the example of the VBI at the top is mainly what's important there.

[Aya_Rei]

I do get tapes that have 2 pixels of VBI data on top but I guess that is not enough, hope I can rip the vbi data from the DCR SR100 and use Avisynth to display it on top of the video

So both the footage and vbi data get cut at the same time. Really only keep the commercials in which not all of them even have close captioning, nor the network bumpers

Quote:

Originally Posted by boomereboot

Sure. My apologies for leaving that out. The attached image file below I have given an example from just a DTV converter box, just to show the VBI at the top. I think certain VCR's seem to strip this until they play the tape back. However I have ran my Panasonic VCR into my DVD recorder and they seem to always display the VBI. If anything the example of the VBI at the top is mainly what's important there.

So what do you use to capture the whole frame including the vbi data? A dazzle and it's built in software or the decode method?

[latreche34]

Technically all capture cards capture the entire 525/625 scan lines because all data is needed, but they only output the active video lines by driver limitation, I would assume that a special program can make the card spits out the entire frame, I haven't delved into this due to interest in it, There was a member here or at VH who once posted a full frame capture, here is a frame grab:

[boomereboot]

I really need to update this post, as now a new fork of cc_decoder has added support for Colored Text via an HTML file export. I've been having some issues updating Python on my Linux computer. The newer version requires Python 3.12+ and also Numpy as the new module.

[Aya_Rei]

So what device do you use to capture VBI data? A Dazzle? Was thinking on grabbing one myself to use just for VBI Data extraction.