ajmoss
16th October 2003, 18:14
My new TV capture card forces me to capture in MPEG-2 format at 352x576 pixels or higher. Normally I capture at a high bit rate MPEG at 352x288, and use TMPGEnc Plus to convert it to VCD-compliant format. Now I need to deinterlace the video becore I can convert it to VCD.
The conventional wisdom is to use Avisynth to deinterlace, as it gives much better results than simply removing odd or even lines from the picture. But this is unnecessary unless you want to go back up to ???x576 pixels non-interlaced. For conversion to [SAME]x288 pixels, the simple deinterlace method of TMPG is good enough.
Which TMPG deinterlace method gives the best results for a signal captured from an analogue PAL broadcast (with all the imperfections that implies)? Presumably it makes no difference whether I use the odd or the even frames, but I'm considering using the Double setting to reduce errors. I don't mind if this makes the deinterlacing step slower.
According to the help file, the Double setting doubles the size of still frames. Surely this won't happen if I use VCD format, or indeed any CBR format that limits the size of the frames?
What does the adaptation setting do? It obviously makes things slower, but does it give better results? If the image in one frame is damaged by a spike in the input signal, will it do a better job of repairing it, or will it just spread the error out to adjacent frames?
The conventional wisdom is to use Avisynth to deinterlace, as it gives much better results than simply removing odd or even lines from the picture. But this is unnecessary unless you want to go back up to ???x576 pixels non-interlaced. For conversion to [SAME]x288 pixels, the simple deinterlace method of TMPG is good enough.
Which TMPG deinterlace method gives the best results for a signal captured from an analogue PAL broadcast (with all the imperfections that implies)? Presumably it makes no difference whether I use the odd or the even frames, but I'm considering using the Double setting to reduce errors. I don't mind if this makes the deinterlacing step slower.
According to the help file, the Double setting doubles the size of still frames. Surely this won't happen if I use VCD format, or indeed any CBR format that limits the size of the frames?
What does the adaptation setting do? It obviously makes things slower, but does it give better results? If the image in one frame is damaged by a spike in the input signal, will it do a better job of repairing it, or will it just spread the error out to adjacent frames?