Am I doing anything wrong with this avisynth script to crop & de-interlace?

[bergqvistjl]

I'm capturing off of a PAL VHS, and i've written the following script that should trim the overscan area, stretch what's left to 768x576 (I'm playing back the resulting video on a player that will not honour Pixel Aspect Ratio, so square pixels it is), and de-interlace to 50p. Is this script OK, or have I done anything wrong?

I have an Intel 10900 CPU which has 10 cores & 20 threads.

 SetFilterMTMode("QTGMC", 2)
 FFmpegSource2("CapturedVideo.avi", atrack=1)
 AssumeTFF()
 QTGMC(Preset="Very Slow", FPSDivisor=1, Edithreads=5)
 Crop(8,0,-8,-0)
 Lanczos4Resize(768,576)
 Prefetch(18)

I'm wondering about colour space, I'm capturing from the VHS using YUY2 (this is simply the default my capture card - Magewell Pro HDMI - has chosen), is this right? And do I need to convert it to anything else in the AviSynth script? I've done a test run and it all looks fine to me, however.

[jagabo]

That looks ok for 4:3 SD video. You may need to convert to YV12 at the end depending on what your encoder expects. Specify rec.601 (aka smpte170m) when you encode. If you're going to upload to youtube you'll want to convert to rec.709 with ColorMatrix(mode="rec.601->rec.709").

[Selur]

Also, I recommend to test on a sample whether 'Edithreads=5' really is the best choice, it seems rather high and higher isn't always better. Lower value might provide more speed.

[bergqvistjl]

That looks ok for 4:3 SD video. You may need to convert to YV12 at the end depending on what your encoder expects. Specify rec.601 (aka smpte170m) when you encode. If you're going to upload to youtube you'll want to convert to rec.709 with ColorMatrix(mode="rec.601->rec.709").

I'm encoding using x264 (8 bit) using YUV422 (in VirtualDub) right now, if that helps.