Which DV codec does capture software select?

[hollowman]

I have a copule of capturing software utilities installed on my Win XP PC: Ulead, WMM, WinDV and Canopus Let's Edit. I also have a couple of CODECs "installed". Each capturing utility, AFAIK, "automatically" selects what they want which, based on MediaInfo data, are: Canopus Let's Edit (Canopus DV), WMM ("DV"), WinDV (Sony/dvsd), Ulead (Sony/dvsd).

I need YUV DV, but only Canopus seems to put out YUV (I don't like Canopus for a couple of reasons, so I would rather not use it).

I downloaded Panasonic DV codec, but I don't know how to make a capturing utility -- like WMM or WinDV -- use IT instead of whatever they are using by default (Sony/dvsd -- AFAIK).

How can I FORCE capturing software to use a CODEC of my choosing?

[guns1inger]

You can install and use several directX DV codecs, but you can only have one VFW DV codec in play at any given time. However you need to first note that none of these come into play at all when you are transferring the video from the camera or from an external DV device. When you are "capturing" from a DV device (ADVC/Camera etc) you are simply transferring the data that was encoded by the device hardware.

The only time the codec comes into the game is when you are actually encoding new DV in software from one of these applications.

If you want a DV codec that you can use with virtualdub then try the cenodica DV codec, which should meet your needs. But most applications, if they install their own DV codec, will use that by default. Some will let you choose other codecs (Vegas, for example, will allow you to chose the Canopus codec if installed), but many, especially budget programs, wont. Pure capture tools such as WinDV don't use the codecs installed on the system for anything other than the preview, so it doesn't matter what they point at.

[jagabo]

Typo: Cedocida

[guns1inger]

One day I will spell it correctly . . . . . . . . . .

and the universe will collapse in on itself and vanish up it's own orifice

[hollowman]

Thx for everyone's reply. Yes, "transferring data" from a DV camcorder is a more accurate description, tho' all the capturing software I know of still (erroneously) refer to it as "capturing".

First, I need to clarify that I'm specifically trying to run Yadif (deinterlacing) AVS scripts in VirtualDubMod (the script also does a few other things). Here’s the script:

loadcplugin("c:\program files\avisynth 2.5\plugins\yadif.dll")
AviSource(".06-08-10_10-52.00.avi")
crop(0,50,720,380)
Yadif()
BicubicResize(640,360)

This script will work for Canopus.

I installed Cedocida per your suggestion.

Next, I transferred footage from the same DV camcorder and tried to open the script in VDM. I got an error stating:
“Couldn’t locate decompressor for format YV12 (unknown).…”

Regular VirtualDub (not VDM) opens this script fine. But I need to works with VDM, not VD.

[hollowman]

Adding the line "converttoYUY2(interlaced=true)" to the AVS script allows VDM to work properly.

loadcplugin("c:\program files\avisynth 2.5\plugins\yadif.dll")
AviSource(".06-08-10_11-01.01.avi")
converttoYUY2(interlaced=true)
crop(0,50,720,380)
Yadif()
BicubicResize(640,360)

More here:
http://forum.doom9.org/archive/index.php/t-130990.html

I was aware of the above before installing Cedocida; so, for my particular app, installing Cedocida wasn't a solution.

[hollowman]

Pure capture tools such as WinDV don't use the codecs installed on the system for anything other than the preview, so it doesn't matter what they point at.

What codec does WinDV use by default for transferring/capturing DV from a camcorder?

[jagabo]

I have no problems importing YV12 into VirtualDubMod. You can also set Cedocida to output YUY2. In fact, that's the best way to set it up. YUY2 is much closer to the 4:1:1 DV subsampling than YV12.

WinDV simply copies compressed DV data from the firewire port to an AVI file. No codecs are involved in this transfer. I don't know what it uses to show you the video on-screen.

[edDV]

Pure capture tools such as WinDV don't use the codecs installed on the system for anything other than the preview, so it doesn't matter what they point at.

What codec does WinDV use by default for transferring/capturing DV from a camcorder?

WinDV doesn't decode the direct DV stream except for AV preview at low resolution. It uses the MS DirectShow codec as far as I can tell. Quality preview doesn't seem possible with WinDV.

[edDV]

Thx for everyone's reply. Yes, "transferring data" from a DV camcorder is a more accurate description, tho' all the capturing software I know of still (erroneously) refer to it as "capturing".

Think of it as "capturing" the DV stream to a file. The DV stream is reordered and wrapped as avi but no decoding is taking place.

[hollowman]

I have no problems importing YV12 into VirtualDubMod. You can also set Cedocida to output YUY2. In fact, that's the best way to set it up. YUY2 is much closer to the 4:1:1 DV subsampling than YV12.

Still pretty much a newbie to this stuff...IAC...

When you say "You can also set Cedocida to output YUY2.", do you mean like this (via AVS script):

converttoYUY2(interlaced=true)

My ultimate goal is to:

- deinterlace using Yadif (not a VDM plugin AFAIK, so I have to use the AVS script noted prev.)

- create 640x360 h264 avi files for YouTube

That said, I'm very much interested in the best possible results for any use, not just YT.

[jagabo]

When you say "You can also set Cedocida to output YUY2.", do you mean like this (via AVS script):

converttoYUY2(interlaced=true)

No. Start VirtualDub. Go to Video -> Compression... Select Cidocida. Press Configure. Set the output format:



The output from Cedocida will now be YUY2. No need to ConvertToYUY2(interlaced=true) in AviSynth.

[hollowman]

jagabo, thx for your USEFUL reply.

Questions:
- Is this better than x264, IF I ultimately want to YouTube the video?
- Will Cedocida work (or work well) with Yadif?
- The camcorder DV-AVI files I have are letterboxed, so I have to crop. Anything I need to be aware of as far as Cedocida AND cropping simultaneously?

(Sorry... I'm away from my editing PC currently and can't experiment to answer the questions myself!)

[hollowman]

I did the test with Cedocida and got this error.

clipboard01.jpg

[jagabo]

VirtualDub has some kind of bug that makes it unable to open DV AVI files directly when the decoder is set to output YUY2. This is weird because it can open any other kind of file that outputs YUY2. (You can get around this problem by also enabling RGB output.) If you are using AviSynth this won't be an issue because AviSynth will be opening the AVI file, not VirtualDub.

[hollowman]

Tried RGB out (both 24 and 32). But now get VDM "Output compressor error: The source image format is not acceptable (error code -2)".

The source was "captured" with WinDV. And I opened an AVS file with VDM. The AVS script is:

loadcplugin("c:\program files\avisynth 2.5\plugins\yadif.dll")
AviSource(".06-08-10_11-01.01.avi")
crop(0,50,720,380)
Yadif()
BicubicResize(640,360)

[jagabo]

The AviSynth script looks fine. What compression codec are you using? If you are trying to compress to DV you can't use 640x360. It has to be 720x480.

By the way, you can use Info() in your AVS script to view the current video properties.

[hollowman]

The AviSynth script looks fine. What compression codec are you using? If you are trying to compress to DV you can't use 640x360. It has to be 720x480.

"What compression codec are you using?" I'm not sure I understood your question: Cedocida, right??
In the past I was using x264, which works fine.

I ditched the BicubicResize line, so now I use VDM to open an AVS file with the followng:

loadcplugin("c:\program files\avisynth 2.5\plugins\yadif.dll")
AviSource(".06-08-10_11-01.01.avi")
crop(0,50,720,380)
Yadif()
Info()

Same error message appears.

Bottom line: I can run the script in VDM using x264, but not Cedocida. BTW, in VDM/AVS, I haven't tried other compression codes yet, like Canopus or Panasonic.

Thx for helping me work thru this

[jagabo]

You have a codec to decompress your source. Then you have have a codec to compress your output. AviSynth is using Cedocida to decompress the source DV AVI file. Then, I guess, you want to use Cedocida to compress back to a DV AVI file. NTSC DV AVI supports only one resolution and one frame rate: 720x480 at 29.97 fps. So you can't feed it a 720x380 or 720x360 frame size.

Note, earlier when I suggested using VirtualDub to configure Cedocida I wasn't suggesting you save your results with it. It was just the easiest way to get Cedocida to output YUY2. If you want x264 video you should be feeding your AVS script to x264, not VirtualDub.

[hollowman]

You have a codec to decompress your source. Then you have have a codec to compress your output. AviSynth is using Cedocida to decompress the source DV AVI file. Then, I guess, you want to use Cedocida to compress back to a DV AVI file. NTSC DV AVI supports only one resolution and one frame rate: 720x480 at 29.97 fps. So you can't feed it a 720x380 or 720x360 frame size.

Note, earlier when I suggested using VirtualDub to configure Cedocida I wasn't suggesting you save your results with it. It was just the easiest way to get Cedocida to output YUY2. If you want x264 video you should be feeding your AVS script to x264, not VirtualDub.

My only real experience (or "education"), thus far, with AviSynth and VDM are based on these tutorials:
http://www.youtube.com/watch?v=hLtTNE61iik
http://www.youtube.com/watch?v=RXByaCFgIHc
(and some brief readings and forum exchanges, like this thread)

So, I guess, I'm still struggling with the learning curve.

I'm not sure why I should feed AVS script to x264 (and not simply via VDM): I have x264 available as a compression option in VDM (Video -> Compression... Select x264, then Configure -- similar to the way Cedocida is config'd in the screen shot you posted earlier).

The crop script "crop(0,50,720,380)" or "crop(0,50,0,-50)" is used to chop off the black bars the NTSC DV camcorder file spat out (dunno why it isnt full-screen anamorphic instead of letterbox). I have to do this because 720x480 letterboxed images get their geometry a bit distorted by Flash-based upload "engines" (e.g. YT).

Although originally videographed with an NTSC miniDV camcorder, the footage I'm using will not be used to create NTSC DV, NTSC DVD (or BluRay for that matter). The primary use for my video projects are PC and web-based (especially "High-Def" or "HQ" video-sharing sites like Vimeo and newer YT). I think it does matter these files look as good as possible because many video-sharing sites, like Vimeo, produce excellent results (and will only improve in the near future).

With these requirements in mind, can you/anyone suggest how I can better tweak my final output files?

Thx again!

[jagabo]

If you want to put h.264 encoded video in an AVI file then do what you have been doing (open your AVS script in VirtualDub) and select x264 as the output codec (Video -> Compression, x264vfw). Since you are doing all your filtering in AviSynth you can use VirtualDub in Video -> Fast Recompress mode. x264 requires the height and width of the frame be multiples of 2. Multiples of 16 are best.

[hollowman]

"x264 requires the height and width of the frame be multiples of 2. Multiples of 16 are best. "
This is new to me, and USEFUL.

Are there any other tweaks (e.g., AviSynth denoise filters, etc.) you can suggest?