Deinterlace first or multiple times? Encoding source to multiple formats.

[greyfalcon]

I typically have a 1080i Transport Stream that I want to encode to multiple sizes and or formats, I'm curious about de-interlacing in the process.

I typically use MeGUI and create an index of the MPG stream, load up a .avs to de-interlace and resize my video during encoding, for an hour show doing a 2 pass XviD for example takes about 90 mins per pass, so a 3 hour job.

LoadPlugin("D:\MeGUI\tools\dgindex\DGDecode.dll")
DGDecode_mpeg2source("G:\Recorded TV\Show1_.d2v", info=3)
LoadPlugin("D:\MeGUI\tools\avisynth_plugin\ColorMatrix.dll")
ColorMatrix(hints=true, interlaced=true, threads=0)
Load_Stdcall_Plugin("D:\MeGUI\tools\yadif\yadif.dll")
Yadif(order=1)
#crop
LanczosResize(720,400) # Lanczos (Sharp)
#denoise

Now lets say I also want a lower quality copy of the same video for portable use, I could use the same script just change the LanczosResize values and run it again, but you have another 3 hour job.

Is there other options than just doing the encode over? Like could one de-interlace to a interim file then run off the 2 resizing jobs after? Would this actually save any time or just degrade quality by doing it separately?

[manono]

Most TV shows are shot on film these days so deinterlacing might not even be the right procedure.

But yes, I suppose you could save out to an intermediate lossless 720x400 AVI and then use that as a basis for all the encodes. It'll go slowly for the lossless file (90 minutes because of only a single pass), but all the encodes after that should be quite fast.

In addition, you could save loads of time by running one-pass encodes for quant 3 or whatever quality you want. The only reason for running two-pass encodes is for specific sizes (like a single CD). And you don't need the ColorMatrix filter at all.