Virtualdub Interlaced Export?

[elliot_123]

I made an avisynth script to scale my 60i video from my sony dcr-pc1000 to 1080i, but when I export it I get interlaced lines on the video. The script I used is below, along with pictures and a video sample.

DirectShowSource("dvcapture 01.avi")
ffdshow("scaleto1080")

I used the codec "ffdshow video codec" and selected MPEG-2 with the following settings:

Interlaced encoding - checked off
Top field first - checked off

I even made some modifications to my script by inserting SeperateFields and Weave and AssumeFrameBased and AssumeFieldBased(I changed the script a few times). I got that info from http://avisynth.org/mediawiki/Interlaced_fieldbased

I still get interlaced lines on my video but it looks different(the lines) when I import the video directly into virtualdub and process rather than using an avisynth script, maybye avisynth is processing the video as progressive or something.

Here are some pictures of the scaling I used and a sample video of the export from Virtualdub:

SAMPLE VIDEO(MPEG IN AVI HD)

http://www.mediafire.com/?dcnbrezj8vfcjyr

Pictures of Scaling I used:

[poisondeathray]

I made an avisynth script to scale my 60i video from my sony dcr-pc1000 to 1080i

1) Why are you upscaling? It's a long way to go from 480i59.94 to 1080i59.94 . There aren't very many benefits, you end up using a lot higher bitrate for lower quality

2) You need to use an interlace aware resize method, (or bob-deinterlace, upscale, then re-interlace)

[elliot_123]

I have used Magic Bullet Instant HD in After effects and exported mpeg-2 with lower field first and it looked fine. The scaling makes the video much sharper because the camera seems to have lower resolution in interlaced mode.

[Gavino]

You need to use an interlace aware resize method, (or bob-deinterlace, upscale, then re-interlace)

He has ticked the 'interlaced' box on the resize settings.

I agree with pdr that there is little benefit in upscaling.
But if you're going to do it, use a better method in your script, eg QTGMC + nnedi3.

[poisondeathray]

If you are going to use an avs script, there are better ways to do it than using ffdshow (I guess I'm asking why are you doing it that way?)

HAHA I didn't see Gavino's post , but that's what I would use too if I had to upscale for whatever reason, plus some sharpening . I would actually stop at 720p59.94 however. Reinterlacing it would give zero benefit and when it's viewed, will just lower the quality even more when deinterlaced by the TV

[jagabo]

Your player isn't recognizing that the MPEG 2 video in the AVI file is interlaced. It's treating it as progressive.

In addition to that the upscaling has produced over-sharpening artifacts on thin/sharp horizontal lines/edges. Or maybe they were already in your source video. Do what Gavino suggested. And store your MPEG 2 video in an MPG container, not AVI. Encode with HcGUI, for example. Even better, leave your video 60p and encode with x264.

[elliot_123]

If you are going to use an avs script, there are better ways to do it than using ffdshow (I guess I'm asking why are you doing it that way?)

HAHA I didn't see Gavino's post , but that's what I would use too if I had to upscale for whatever reason, plus some sharpening . I would actually stop at 720p59.94 however. Reinterlacing it would give zero benefit and when it's viewed, will just lower the quality even more when deinterlaced by the TV

The reason I leave the video in its interlaced form is because wikipedia said that interlaced video is meant to be editod and stored in the same format. I think that deinterlacing(if not done properly, or maybye I am wrong) reduces quallity of the video. I let the TV perform the deinterlacing as if I had the camera connected to the tv and playing it back, that way the tv performas the deinterlacing and quality stays the same because the fields are left intact and unaltered.

[elliot_123]

You need to use an interlace aware resize method, (or bob-deinterlace, upscale, then re-interlace)

He has ticked the 'interlaced' box on the resize settings.

I agree with pdr that there is little benefit in upscaling.
But if you're going to do it, use a better method in your script, eg QTGMC + nnedi3.

Do you have a script that would allow for good scaling using the method you mentoid above(QTGMC+ nnedi3), I wanted to test it to see if it looks better

[jagabo]

Do you have a script that would allow for good scaling using the method you mentoid above(QTGMC+ nnedi3), I wanted to test it to see if it looks better

You'll need to download all the required filters:

QTGMC: http://avisynth.org/mediawiki/QTGMC
nnedi3: http://web.missouri.edu/~kes25c/

Assuming a 16:9 DAR source:

WhateverSource("filename.ext") 
AssumeTFF() # or AssumeBFF(), whichever is appropriate
ConvertToYV12(interlaced=true) # if necessary
QTGMC(preset="very fast") # or faster, fast, etc.  see QTGMC docs
nnedi3_rpow2(4, cshift="LanczosResize", fwidth=19200, fheight=1080)

# at this point you will have 1920x1080p60 video
# if you need interlaced output:

AssumeTFF() # or AssumeBFF(), whichever you want
SeparateFields()
SelectEvery(4,0,3)
Weave()

[Gavino]

The reason I leave the video in its interlaced form is because wikipedia said that interlaced video is meant to be editod and stored in the same format. I think that deinterlacing(if not done properly, or maybye I am wrong) reduces quallity of the video. I let the TV perform the deinterlacing as if I had the camera connected to the tv and playing it back, that way the tv performas the deinterlacing and quality stays the same because the fields are left intact and unaltered.

That's generally good advice - but because here you are resizing, the fields must be altered and deinterlacing is required. In that case, you have the choice of reinterlacing the upscaled result or leaving it as 60p.

BTW, there's a typo in jagabo's script (should of course be 1920):

nnedi3_rpow2(4, cshift="LanczosResize", fwidth=19200, fheight=1080)