Resizing VHS captures

[brikin]

If I am encoding VHS and want to resize the image to 352,480 is it better to do this before or after I apply filters? Or does it matter at all?

Another question. If record shows on my VHS then move these to my DV camera. When I capture this it is 720, 480. Is there a point in resizing it to 352, 480? I know that VHS is only 352, 480 but the DV camera will upsize it.

Thanks for any help.

edit: changed your title a bit.

[Wilbert]

If I am encoding VHS and want to resize the image to 352,480 is it better to do this before or after I apply filters? Or does it matter at all?
In general it is better to smooth/denoise before resizing. I suggest you look at the comments in the analog capture guide.


Another question. If record shows on my VHS then move these to my DV camera. When I capture this it is 720, 480. Is there a point in resizing it to 352, 480? I know that VHS is only 352, 480 but the DV camera will upsize it.
This is a misconception, VHS is not really 352. The DV camera doesn't upsize. I suggest you look at the preface of the analog capture guide for some comments about this. In general, one should capture using the largest possible solution (720x480 in your case), apply smoother (if necessary) and downsize. The latter depends on your target of course (DivX, SVCD, etc.).

[brikin]

Thank you for all the information Wilbert. I read the capture guide and everything makes sense now.

I did run into a problem and wanted to see if anyone else has had this problem.

I recorded a show to VHS then re-recorded this on my DV camera then captured to my harddrive. I captured at 720X480 NTSC. I had some black on the side and a rainbow effect on the bottem. I read that this usually happens when capturing VHS in the capturing guide. So I used the crop command and took the black line off the side and cropped 4 lines off the bottom to get rid of the rainbow effect. I then used a LancrosResize to 720X480. I then run it through Convolusion3D using the settings in the DV to DVD guide. Everything looks fine on my computer with this MPEG.

I then put it on DVD and play on my DVD player and I get two fuzzy areas in the picture one about 1/4 down in the picture and one 3/4 down. These fuzzy areas stretch from one side of the screen to the other.

Has anyone else had this problem?

[Wilbert]

:script:

I would would replace those four lines by black lines. That way you don't need to resize it.

They are not there in the original dv?

[brikin]

Script:
################################################
AviSource("C:\MyDVFiles\DV_type2_file.avi")

Crop(8, 0, 0, -2)

LanczosResize(720, 480)

SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()

DoubleWeave.SelectOdd()
#################################################

No I don't see the fuzzy lines in the original. I also don't see them using PowerDVD. I only see them on my TV.

[brikin]

Just so people know I found a solution to this problem. I encoded the avs file with TMPGEnc and everything looks great. For some reason this file could not be encoded with CCE without looking like crap.

[scharfis_brain]

AviSource("C:\MyDVFiles\DV_type2_file.avi")


Crop(8, 0, 0, -2)

SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
LanczosResize(720, 240)
Weave()


DoubleWeave.SelectOdd()

this way, it'll look always good!

you've resized an interlaced structure, which results in a damaged image

[brikin]

Thanks for the info I changed my script

AviSource("C:\MyDVFiles\DV_type2_file.avi")

SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
LanczosResize(352, 240)
Weave()

But when I encode using CCE it produces jerky playback with the fuzzy lines in the picture still. TMPGEnc does not do this. Does CCE have a problem with producing 352X480 mpeg's?

[scharfis_brain]

try to append doubleweave().selectodd()
at the end of this script.
it will reverse the fieldorder BFF becomes TFF and vice versa.

[brikin]

Sorry about that. I already have doubleweave().selectodd() at the end of my script. I don't understand what is going on. I tried encoding at 720X480 (just removed the resize) and my mpeg still ends up jumpy. This only happens for me when I capture TV or VCR using my digital camera. My normal camera footage works (and looks) great using CCE.

For some reason I need to use TMPGEnc for this footage. I capture the footage using the same settings as my normal camera footage so I don't get why this doesn't work.

[scharfis_brain]

then, try one of this fouer ones!

assumetff.doublweave.selectodd
assumebff.doublweave.selectodd
assumetff
or
assumebff

[brikin]

scharfis_brain thanks for the help but none of these work. It isn't as jerky as selecting the wrong field it just gets jumpy on some parts. I don't know what else to try.

[vhelp]

hi brikin,

You had us fooled for a while.. and confused in other points of your
issue(s) :o

Yeah, you have:

* recorded tv shows onto your VCR (vhs tapes)
... ** though not stated, some shows are Interalced & Telecined
... .. You seem to have miss this important fact out.


You also are doing this now:

* taking your VHS and running throught your CAM (via pass-throu)
...another confusion part (we missed)

When you record a TV show (lets assume TV Movie) and you run it throu
a pass-throu process, the source type does not change, (only the method
of the "processed frame" does ..ie, field order) hence..

If your (recorded to VHS, and then pass-throu'd) TV movie is Telecined,
it needs to be IVTC.
.
.
If your (recorded to VHS, and then pass-throu'd) TV movie is Interlaced,
then you either encode it:
* w/ Interlace (see code below) or,

LoadPlugin("c:\dlls252\decomb.dll")
AviSource("C:\MyDVFiles\DV_type2_file.avi")
Telecide()
Decimate(cycle=5)

* De-Interlace (see code below)
AviSource("C:\MyDVFiles\DV_type2_file.avi")
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
LanczosResize(352, 240)
Weave()

..give or take a few correction to the source snips :)
You may just be using the wrong script on your source files.
-vhelp

[brikin]

Sorry if I caused any confusion. Here is exactly what I did.
I recorded an episode of American Chopper off of the Discovery Channel to my VHS. I then hooked my digital camera up to my VHS and recorded the show on to a miniDV tape. I then took my camera to my computer and captured the image to my hard drive. I captured this images as 720X480 interlaced. I removed some commercials then output this to a avi file using the Panasonic codec. I then wrote an avisynth script

AviSource("Old School Chopper 1.avi")

ConvertToYUY(interlaced = true)

SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
LanczosResize(352, 240)
Weave()

DoubleWeave.SelectedOdd()

I then open this in CCE 2.66. I set this up as a 4:3 movie with Alternative Frame type, 10 DC precision, I did not check the progressive frame box. GOP Sequence M=3 N/M=4. I check the "Add Sequence End Code" tab, Lum level to 16 to 235, and frame rate to 29.97. I then encode the movie and it plays fine as an mpeg on my computer. I then create a DVD with DVDLab and the output stutters a little bit every 5 to 10 seconds. This happens on PowerDVD as well as my set top player.

This does not happen if I encode the same .avs file with TMPGEnc. I have no idea why this happens with CCE.

Thanks for all the help guys.

[troy]

Just a thought. Maybe it is something to do with dvdlab. I sometimes get some strange things with dvd lab.

[brikin]

I thought the same thing but when I encode the avi using TMPGEnc it works great. So I think it must be CCE.

[dplaton]

CEE is maybe faster, output a better photography, but as far as I understand it is able to deal only with interlaced sources top field first, I don't know if CEE has an option to specify the field order like TMPEGenc.
So if you want to encode interlaced use TMPEGenc, the best.
Anyhow there is no reasonable reason :) to deinterlace, just in case if you want a longer time to encode and a poorer image quality. On a TVset interlaced is not just fine but is also proper, on a PC just separatefields and play at 50 fps or 60 fps.
I wasted a lot of time, in some cases encoding with 1fps speed, on deinterlacing.
I think that is misleading to folow the doom9 guide that describe how to deinterlace (with bad results) and who don't mention that is NO NEED to deinterlace.

[scharfis_brain]

this thread is more than one year old!