digitalFAQ.com Forums [Archives] - Avisynth: Interlaced sources guide? [See pg2]

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)

- Avisynth Scripting (http://www.digitalfaq.com/archives/avisynth/)

- - Avisynth: Interlaced sources guide? [See pg2] (http://www.digitalfaq.com/archives/avisynth/9032-avisynth-interlaced-sources.html)

Hi guys, how are you!!

[quote="incredible"]

Quote:

Originally Posted by GFR

Nice thread...

All these mentioned sources to me seem to be recorded "fieldbased"! Means Real interlaced where every field (not frame) at half height gots its unique movement.

So you should try the recommendation of
Smoothdeinterlacer(doublerate=true) # or any other full fieldrate deinterlacer, also dgbob(order=x) does that well ... just try spped/quality
convert60ito24p(2,0)

I think if your source comes from a standalone VHS or Camcorder you do digitize that one using for example a HuffYUV or a mjpeg codec. Both do deliver a good quality and they come in YUY2 which got better Chroma subsampling and also there won't be a colorconversation needed BEFORE the function call as convert60ito24p needs YUY2 as input.

:wink:

Even if you choose mpeg2 as encoding format, you can gain MUCH compressibility by converting that one using convert60ito24p. As interlaced encodings do need much more bitrate if the output of quality should be the same as a progressive encoding (according to blocks etc.)

BUT keep in mind .. on the one hand you will get less blocks cause of more compression, but the "smooth" movement of your NTSC VIDEO material will be changed to that typically Film Look playback! And once convert60ito24p is proceeded on the encoding you cant restore it in the future.

Please incredible, could you clarify me. I am totally crazy with some captures of interlaced material (some weddings footages).
I am using Studio8 + DC10plus to capture some VHS tapes, interlaced material (already checked via virtualdub). When I use info() filter to check, it says that color is YV12, not YUY2 8O . So, I can't use smoothdeinterlacer neither the convert60ito24p function ( the layer needs YUY2 color) :x .
I'm trying to use CCE 2.66 to encode, only crash :? . I don't know what I messed up. I can't change the codec in Studio, so I can't figure what I made wrong.
How can I change the color to YUY2 without miss the "real color", because when I use ConverttoYuY2() in the script, all the color change (like a infrared style or something like that).

Regards,

LaBomba

labomba,
I am not an expert, but I do this to convert true 60 fields per second interlaced sources to 24 frames per second non-interlaced video and it seems to work fine when encoding with TMPGEnc.

ConvertToYUY2
SmoothDeinterlace(tff=false, doublerate=true, blend=false)
convert60ito24p(2,0) #function by Scharfi
ConvertToYV12

@ La Bomba

Strange as I also got a Pinnacle DC capture card beside a 7134chip based card. The DC series do defenitely capture in YUY2 compressed by the onboard mjpg codec! Means that Somthing happens on the way to your HD with the stream! I also did try once studio 8 but I got some issues where no correct 25.000 frames per second came out, means like 24.998 or 25.001 in regular thats normal at capturing but u should use VirtualVCR!! and its internal audio resampling engine, by this you will end up with perfect synced 29.97 fps mjpeg streams! And in YUY2!

Do you got the picVideo mjpeg codec installed to decode/view your captures?

Pleas check all that.

@ Peter 1234

Also try Dgbob(x) instead of smoothdeinterlace(xxx), and tell me your results as its also a 60fps deinterlacer and maybe its faster.

@ Jorel
Thats a NICE link!

:)

@labomba: Use ConverttoYUY2(interlaced=true) when converting interlaced material!

@Peter1234: You should also remove the ConverttoYV12 line from your script as there's really no need to feed TMPGEnc with YV12 data, YUY2 will do just fine so you'll get rid of one colorspace conversion. TMPGEnc will convert to RGB24 anyway.

incredible and Boulder,
I tried the following

ConverttoYUY2(interlaced=true)
dgbob(order=0) ### order set for bottom field first with defaults
convert60ito24p(2,0) ### function by Scharfi

and it is much faster than the script I was using. The output file is also smaller and seems
to be slightly sharper BUT there are larger residual combing artifacts. I prefer the
smoother picture generated with my above script. Field order of my DV test video is
definitely bottom field first. I also tried reducing the thresh value in DGBob and that
helped, but I still did not think it was as good as SmoothDeinterlace. However, I agree
that using DGBob is much faster and so is therefore better for most people. Thanks for
the feedback.

@incredible
Thanks, I will try with VirtualVCR. I normally use Studio8 because all other programs crashes, like IuVCR and also VirtualDub (I use WinXP and cannot use VDub to capture)
I have Morgan, PicVideo and also Huffyuv codecs installed, but I can't figure how assign one of those in Studio8, would be in the "Make AVI" section? Should I deinstall some of them?

edited:
Using GSpot, it gaves me this video info of my AVI file :
4CC : MJPG
Name: Motion JPEG including Huffman tables
Stat: CODEC Not Installed!
xy: 640x480 (1.33:1) 4:3 ---> (NTSC Land)
FPS: 29.971

I never saw any mention about Huffman tables before. Should I deinstall Huffyuv codec? How can I make it?

@Peter and Boulder
Thank you guys, I'll try to change my capture program to get YUY2 color source material. When I use ConvertToYUY2() in my script the colors changes, like a wrong chroma issue. Baddly, I undertand less what is going on, than understand English :lol:. And You can see how much i understand English ! :lol: :lol:
But I will try to find making some tests.

LaBomba

Quote:

Originally Posted by Boulder

@labomba: Use ConverttoYUY2(interlaced=true) when converting interlaced material!

Boulder,
ConverttoYUY2() is recommended to use in the end of the script!
"deinterlace" filters are recommended to use before the resizes!

then,
the better place to use
ConverttoYUY2(interlaced=true)
is the "very end" of the script or before the resizes when the source is interlaced ?
:?
my question seems stup and maybe is but i have doubts!

thanks in advance!
:wink:

Quote:

Originally Posted by jorel

If I remember correctly, the convert60ito24p() function needs YUY2 data so the conversion must be done before using it.

In "normal" cases, it's best to do the conversion last as you figured.

Yep, that was a very good reference but the workouts there cant be done understandably using avisnth like that bob&weave technique.

If you want REAL good explanantions how these interlacing techniques are done AND how you can follow/imitate them using avisynth .... incl. which avs script is needed to treat eachone individual ....

.... this one is it:
http://home.arcor.de/scharfis_brain/...esInterlacing/

use Googles Language Tools or any other online translator you like to change it to your language ;-)

Thank you Inc. I'll give a try soon.