Markosjal
Member
I am just posting an FYI here for other who may be searching to do the same.
My goal was to find a capture solution that gives me good quality composite / s-video captures while preserving the Interlacing. I wanted the Interlacing preserved to apply the same technique that I have applied to DV video. I did not want any capture hardware or software making decisions for me. As a former Television Engineer I was always unhappy with the results when capturing with most hardware. A few years ago , I stumbled upon ffmpeg's yadif=1 option for de-interlacing.
I have been testing the Dazzle DVC100 capturing to OBS on linux. This works well with the "one click fix" that I have posted here https://obsproject.com/forum/thread...-consistently-get-dazzle-dvc100-audio.170392/
I can save those archives as-is just as seen in the setup screens on that forum post, and am very happy with the resulting Interlaced video. I demand that half the resaolution not be thrown away as done by many other capture devices or bundled software. What I like most about it is that if I use ffmpeg post processing, I can do the following in linux (probably in MacOS or WIndows as well):
ffmpeg -i input.mkv -vf yadif=1,scale=960:720 -aspect 4:3 -c:v libx264 -preset slow -crf 18 -c:a copy -pix_fmt yuv420p output.mkv
What this will do is provide an upscaled output using YADIF (=1) which is a smart de-interlacer that will double the frame rate giving you a full frame for each NTSC or PAL field. It should work on NTSC or PAL. You will be doubling the frame rate of the resulting video so if your goal is to post to Youtube or something I do not believe youtube supports 59.94FPS and 50FPS. According to my tests they only support 60FPS and this jacking with the temporal resolution becomes obvious.
to be completely fair, I can not really show the difference in a single still frame but if you are interested, download the DV sample from the link below, or use your own (MUST BE INTERLACED- so no cheap capture cards that de-interlace), and apply the command mentioned above to the sample DV file linked below from the Internet Archive page.
the direct link is here, which will give you a zip file that you must unzip first:
Have a look at the statue about 15 seconds in in both formats after you convert it. Look specifically at the belt buckle area where there is a diagonal line. Also notice there is an overall lower incidence of "dancing ants" in the grass in the same scene.
I do not use this for all archives in fact archives are stored in the original Interlaced format and they can be converted/up-scaled later as needed. Sometimes this is not the best solution, for instance when the result will be 30 or 60fps. As mentioned earlier it would not be appropriate for Youtube for instance at this point to work from YADIF and up-scaled video because youtube will jack with the frame rate from my experience.
My goal was to find a capture solution that gives me good quality composite / s-video captures while preserving the Interlacing. I wanted the Interlacing preserved to apply the same technique that I have applied to DV video. I did not want any capture hardware or software making decisions for me. As a former Television Engineer I was always unhappy with the results when capturing with most hardware. A few years ago , I stumbled upon ffmpeg's yadif=1 option for de-interlacing.
I have been testing the Dazzle DVC100 capturing to OBS on linux. This works well with the "one click fix" that I have posted here https://obsproject.com/forum/thread...-consistently-get-dazzle-dvc100-audio.170392/
I can save those archives as-is just as seen in the setup screens on that forum post, and am very happy with the resulting Interlaced video. I demand that half the resaolution not be thrown away as done by many other capture devices or bundled software. What I like most about it is that if I use ffmpeg post processing, I can do the following in linux (probably in MacOS or WIndows as well):
ffmpeg -i input.mkv -vf yadif=1,scale=960:720 -aspect 4:3 -c:v libx264 -preset slow -crf 18 -c:a copy -pix_fmt yuv420p output.mkv
What this will do is provide an upscaled output using YADIF (=1) which is a smart de-interlacer that will double the frame rate giving you a full frame for each NTSC or PAL field. It should work on NTSC or PAL. You will be doubling the frame rate of the resulting video so if your goal is to post to Youtube or something I do not believe youtube supports 59.94FPS and 50FPS. According to my tests they only support 60FPS and this jacking with the temporal resolution becomes obvious.
to be completely fair, I can not really show the difference in a single still frame but if you are interested, download the DV sample from the link below, or use your own (MUST BE INTERLACED- so no cheap capture cards that de-interlace), and apply the command mentioned above to the sample DV file linked below from the Internet Archive page.
the direct link is here, which will give you a zip file that you must unzip first:
Have a look at the statue about 15 seconds in in both formats after you convert it. Look specifically at the belt buckle area where there is a diagonal line. Also notice there is an overall lower incidence of "dancing ants" in the grass in the same scene.
I do not use this for all archives in fact archives are stored in the original Interlaced format and they can be converted/up-scaled later as needed. Sometimes this is not the best solution, for instance when the result will be 30 or 60fps. As mentioned earlier it would not be appropriate for Youtube for instance at this point to work from YADIF and up-scaled video because youtube will jack with the frame rate from my experience.