On Saturday 26 Jan 2002 1:59 am, you wrote: > David Pratt (dp@xxxxxxxxxxx): > > >> Note that MPEG1 doesn't have an interlaced mode, so if you record a > > >> 640x480 stream from TV input, you'll see artifacts which bleed > > >> between the fields. Make sure you're using an encoder which can > > >> encode in an interlaced mode (like MPEG2) if you want to capture > > >> both fields in a frame. > > > > To avoid those atrifacts, why not seperate the frame into two fields > > to give twice the fps and then compress that video stream, eg pal > > 720x576x25fps > 720x288x50fps. On playback every two frames can be re- > > interlaced to get the original res/fps. > > Because each field has a parity. That is, the first 720x288 image is > all the even scanlines, and the next 720x288 image is all the odd > scanlines. If you interpolate each one up to frame size without taking > this into account, the image will jump up and down really horribly and > look terrible. Yeah, it definately does, put I was suggesting that more as a way of storing the video between capture and playback. > I can add this 'functionality' (playback at 50hz without taking into > account field parity) into tvtime and then you can see what it looks > like. :) > > > I am involved a project at the moment which has similar requirements > > and problems - i need to get dvd quality tv output and the system > > wasnt able to report which field it was displaying when i started > > coding. I have found that I can get pretty close (although the > > interlacing isnt perfect) by using video compressed as described > > above, leaving the flicker filter on, and blit-stretching each video > > frame to the full display buffer. It looks pretty close although on > > some sources the interlacing can look a little stronger than it > > should. > > Are you blitting at the field rate or frame rate? Field rate (50 or 60hz).
