*** F.A.Q. ***

[bb]

Q1: What's the difference between DV type-1 and DV type-2?

A1: AVI files can contain different video and audio streams. The native interleaved DV data produced by camcorders is stored in a type-1 AVI file with stream type ivas.
If the native data is split into separate audio (auds) and video (vids) streams, then we have a type-2 AVI. In this case the vids stream still contains the audio interleaved, thus the filesize is a little bigger.
Video-for-Windows (VfW) applications like VirtualDub don't work with type-1 files. You can either convert to type-2 or use AviSynth and its DirectShowSource command to open the video. In the latter case process the audio separately.

Q2: Progressive or interlaced?

A2: Most camcorders support interlaced only. If your target format doesn't support interlaced video (e.g. DivX, MPEG1), then you have to deinterlace, which costs time and quality. In case of MPEG2, which supports interlaced video, you can still go the progressive way, because you need lower bitrate for the same quality, or if you turn it the other way round: the quality will be better for the same filesize. The reason why people use interlaced is the smoother motion on a TV set, and they likely have enough bitrate because they create DVDs, not SVCDs...

Q3: How do I properly resize a DV video?

A3:
DV Aspect Ratio
---------------

A 720x576 (PAL) captured DV file has an aspect ratio of 1.25. You want to watch your video on a 4:3 (= 1.33) screen, so all you have to do is resize it to 640x480, which has the correct aspect ratio of 4:3 (640/480 = 1.33), right?

Wrong!

The above would be correct if the DV pixels were square, but unfortunately they are rectangular, which results in a deviation of about 9.4% according to what I have measured. To make sure I don't tell you anything wrong I filmed a square sheet of paper and measured widths and heights on my TFT screen.

The square sheet of paper was 12.6cm x 13.9cm on my screen, so the width/height ratio was 0.906. That means, if you want to get a video having a standard width of 640 pixels and a height divisible by 16, you have the choice:

1. You can crop 32 pixels, 16 from each side and resize to 640x480.
AVS sample script:
AviSource("C:\DV\testAR_DV_type2.AVI")
crop(16,0,688,576)
BilinearResize(640,480)

2. You can resize to 640x464.
AVS sample script:
AviSource("C:\DV\testAR_DV_type2.AVI")
BilinearResize(640,464)

As a result circles look like circles, not elliptic on computer monitors, because computer monitors have a pixel aspect ratio of 1:1. TV screens don't have square pixels (e.g. for PAL the ratio is about 1:1.1). I do not recommend to produce the video for pixel aspect ratios other than 1. Maybe your video card adjusts for the pixel aspect ratio, otherwise let the player do the proper resizing.

Note: If you have a camcorder which supports 16:9, then resize your 16:9 DV video to 640x352.

Q4: Where do I get technical information about DV?

A4: Look here: http://www.adamwilt.com (thanks to Ookami for providing the link).
Another interesting link: http://www.shortcourses.com/video/chapter01.htm

[bb]

Q5: How do I install the Panasonic DV codec?
A5: Windows 98/ME:
Extract the codec and copy the file pdvcodec.dll to the windows\system directory. Add the following entry in system.ini, section [drivers32]:
vidc.dvsd=pdvcodec.dll

Win2000 / WinXP
Extract the attached REG file, then double click it to install.

Or enter the following registry key manually:



[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\drivers.desc]
"pdvcodec.dll"="Panasonic DV Codec"
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Drivers32]
"vidc.dvsd"="pdvcodec.dll"
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\MediaResources\icm\vidc.dvsd]
"Driver"="pdvcodec.dll"
"Description"="Panasonic DV Codec"

[bb]

Q6: How to deal with CCE's field order bug (V2.50)

A6: CCE 2.50 always encodes with top field first, whereas DV is bottom field first. Use pulldown.exe after CCE 2.50 encoding to revert the field order (set it to even). Then your DVD will play fine on a DVD player. (Thanks to nick ctrl.)

Instead of pulldown.exe you may use ReStream as well.

[bb]

Q7: Which camcorder formats exist?
A7: Currently the following camcorder formats are available:

1. DV
The DV format is similar to MJPEG (motion JPEG). Every frame is a keyframe which makes editing an easy task. The compression ratio is 5:1 at a data rate of ca. 3.6 MB/s. This results in an excellent quality. You will encounter mainly two cassette formats, either miniDV or Sony's Digital8. The data is stored the same way on either cassette type, so there's no difference in the DV stream.

2. MicroMV
Sony's proprietary MicroMV format allows for cassettes even smaller than miniDV. It is a keyframe-only MPEG-2 format, and not many editing programs support it. The data rate is 12 MBit/s, which is one of the reasons that make a direct transfer to DVD impossible, because DVD is limited to 9.8 MBit/s.

3. DVD-RAM
Hitachi recently created a camcorder using DVD-RAM as storage medium. The video is stored in MPEG-2 format. A 1.4 GB DVD-RAM is sufficient for 30 minutes recording time. MPEG-2 is not recommended for editing, though.

4. "Hard storage"
Panasonic announced a camcorder for professional use for 2004 which uses SD cards as storage medium. Data is recorded in the DVCPRO format (a professional DV flavour).

[bb]

Q8: What's color subsampling, e.g. 4:2:2, 4:2:0, 4:1:1?
A8: Color subsampling defines the way color (= chroma) information is stored in digital formats. Here's a good visual explanation:
http://www.adamwilt.com/pix-sampling.html

[bb]

Q9: Which DV codecs use which four character codes (4CC)?
A9: Most DV codecs use the standard, which is DVSD.
Canopus is an exception; the codec uses CDVC instead.

You can easily change the 4CC in your AVI file using the "FourCC changer": http://www.doom9.org/Soft21/Editing/AviC.zip

[bb]

Q10: 4:3 interlaced to 16:9 anamorphic

Often people ask themselves how to convert a 4:3 interlaced shot into a 16:9 anamorphic video (interlaced, too). To achieve this, you crop away everything outside the 16:9 window of your 4:3 frame; typically you keep the center part of the frame. Then you upsize the cropped video to get back a full sized DV frame. But how much to crop? How to resize correctly to keep the aspect ratio, thus avoiding "egg heads"? How to deal with the interlacing?

The answers differ depending on the TV standard. Although the formulas are the same, the crop and resize values differ because of the different frame sizes. Here are the answers.

NTSC:

frame........720x480
field........720x240
4:3 AR.......1.500
4:3 DAR......1.333
PAR..........0.889
16:9 AR......2.000
16:9 DAR.....1.778
16:9 frame...720x360
crop total...120 pixels

NTSC example script:

AviSource("C:\DV\test.avi")
# alternatively: DirectShowSource("C:\DV\test.avi")
Crop(0,60,0,-60)
SeparateFields()
BicubicResize(720,240)
ConverttoYUY2()
ReInterpolate411()
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()
# BFF to TFF: DoubleWeave.SelectOdd()

PAL:

frame........720x576
field........720x288
4:3 AR.......1.250
4:3 DAR......1.333
PAR..........1.067
16:9 AR......1.667
16:9 DAR.....1.778
16:9 frame...720x432
crop total...144 pixels

PAL example script:

AviSource("C:\DV\test.avi")
# alternatively: DirectShowSource("C:\DV\test.avi")
Crop(0,72,0,-72)
SeparateFields()
BicubicResize(720,288)
ConverttoYUY2()
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()
# BFF to TFF: DoubleWeave.SelectOdd()

Feed the script into your favourite MPEG-2 encoder, and remember to switch the 16:9 flag on. This will tell your player that it has to deal with anamorphic content, i.e. it has to resize to 16:9 display.

Now that you know how to turn your 4:3 footage into 16:9 anamorphic, let me add a note which you might want to consider:

During the abovementioned process you crop your video, then you upsize it to full frame (field) size. You don't gain information in this process, you merely "blow up" the video. That means you'll need more bitrate to encode it later.

An alternative is to encode 4:3 letterboxed and use your TV set's 16:9 zoom. This way you save bitrate while preserving the same picture quality. All you have to do is to crop and add back black borders of the same size, which can be done in a single Letterbox command in AviSynth.