Favourite methods for playing VHS captures on modern TV?

[superduck]

Hi, I'd like to watch some of my VHS captures on my plex server, which so far are captured through AmaRec at 25i Lagaryth. If I do a simple recompress to a more manageable 25i mp4 my TV does an ok job at deinterlacing but there's often judder and combing artifacts.

I've also tried yadif 25p and 50p deinterlacing through ffmpeg. My favourite result I think is 50p interpolation as there's no artifacts and gives that 50fps illusion that you get through a CRT tv but it seems a very inefficent way to go about simply playing vhs captures on a modern TV. The files are also much bigger.

So I was wondering what your favourite methods are for playing back VHS captures on a modern TV without artifacts? Thanks!

[Skiller]

My favourite result I think is 50p interpolation as there's no artifacts and gives that 50fps illusion that you get through a CRT tv but it seems a very inefficent way to go about simply playing vhs captures on a modern TV. The files are also much bigger.

Hi,

couple of things I want to mention:

1) Put simply, bob-deinterlacing is merely an interpolation on the spatial dimension of video – not time. It turns those 50 fields per second into 50 frames per second by filling the gaps (missing scanlines). Hence the motion of bob-deinterlaced 50p video is not an "illusion", it is indeed proper 50p like from a native 50p source.

2) Encoding 50p is not particularly inefficient. Assuming you are encoding h264 (such as x264), there is not a big difference in encoding efficiency when encoding 50p versus 25i.

3) In case of CRF encoding in x264, file sizes probably differ because of the way x264 distributes bitrate/quality in interlaced vs. progressive encoding.
If you were to tune CRF so that both encodings (25i and 50p) end up with pretty much the same file size, they are going to look both pretty similar in terms of quality (compression artefacts).


So my advice is: bob-deinterlace to 50p with whatever deinterlacer suits your expectations (or CPU time^^) and encode 50p at the videos native resolution (no upscaling, use SAR).

[dannyboy48888]

Now and days I deinterlace with qtgmc on my captures. The motion is always correct and I quit having loss of detail. If it's truly 60 fields the file size reflects that ( I encode with x264 in qp not crf mode). If it's only telecined I noticed a 10% increase in size at the upper end which is worth it to me to not have to worry about the motion and detail per capture.