How to Capture DV source in RGB32?

[mlong30]

Hi All,

Does anyone know of a way for me to capture a DVsd source video (Canopus ADVC110) to RGB32? Are there any DV devices that can capture in 32bit?

Thanks.

Matt

[edDV]

No, the ADVC and all other DV capture devices encode to YCrCr 4:1:1 (4:2:0 for PAL). Luminance is 8 bits (with gamma applied). Cb and Cr are nominally 8 bit but processed to DV levels.

You could call this compressed 24bit but that would be a bit of a stretch. A second DV luminance channel can be added to get to "32 bit" the Alpha is needed.

See: http://en.wikipedia.org/wiki/DV "Technical standards" for a good description of how DV is processed through a DV camcorder.

Explain better what you are trying to accomplish. When you decode DV format to RGB with say the Panasonic DV codec in Virtualdub, you are left with 24bit RGB (8bits per component) but that isn't the same as capturing to 24bit RGB.

[mlong30]

Explain better what you are trying to accomplish. When you decode DV format to RGB with say the Panasonic DV codec in Virtualdub, you are left with 24bit RGB (8bits per component) but that isn't the same as capturing to 24bit RGB.

I'm trying to create a workflow that will not loose much picture quality, because of switching colorspace.

1. Capture from ADVC110 Uncompressed AVI to (24bit RGB)
2. Import AVI to VirtualDub(RGB??) using NeatVideo Filter in (YCrCb).
3. Save Virtual Dub file to Lagarith Codec RGB24
4. Import into Sony Vegas, which I believe converts to RGB24.
5. Render out to Mainceipt, or AVI file using Lagarith Code RGB24 for CCE.

How can I, if possible, reduce loosing picture quality?
What colorspace does VirtualDub use?
NeatVideo has a RGB colorspace, but I was told by NeatVideo not to use it.

Thanks.

Matt

[edDV]

The ADVC will not capture to anything but DV format (YCbCr) which is the same colorspace as DVD MPeg2.

Also, although Vegas is RGB internally, colorspace is preserved in DV project settings. I'll let others comment on Avisynth technique to maintain colorspace.

[mlong30]

The ADVC will not capture to anything but DV format (YCbCr) which is the same colorspace as DVD MPeg2.

When I run a avs script using Info(), its says the DV format file is RGB24. Is this the same as (YCbCR)?

Thanks.

Matt

[JohnnyMalaria]

The ADVC will not capture to anything but DV format (YCbCr) which is the same colorspace as DVD MPeg2.

When I run a avs script using Info(), its says the DV format file is RGB24. Is this the same as (YCbCR)?

Thanks.

Matt

A DV format file cannot be RGB24. A decoder can generate an RGB24 image from a DV file. I don't know what the Info() function does (never used avs). If it is telling you what the decoder is generating, it's believable. If it is supposed to tell you the native format of the information in the file, it's wrong.

DV video data are encoded and compressed as Y, Cb and Cr (often termed YUV or similar).

When decompressed and rearranged to create an image, the native DV colorspace is also a YCbCr/YUV variant.

The DV specification doesn't dictate how the decompressed image is arranged in memory (there are so-called packed and planar ways) but, the nearest planar match to DV is YV12 (for PAL). For NTSC, YV12 isn't strictly correct.

For packed formats, the nearest match to DV is Y411 for NTSC. No direct equivalent for PAL(!).

Then there are packed formats that expand the native 4:1:1 or 4:2:0 to 4:2:2 such as UYVY, YUY2 and more ( http://www.fourcc.org/yuv.php#Packed%20YUV%20Formats )

Yes, it's a ridiculously confusing situation!

(That's why our software does even bother to rearrange the data into a traditional frame. It works directly on the decompressed data in the same way they are arranged in the DV file).

[mlong30]

A DV format file cannot be RGB24.

After reviewing Gpsot, I agree with you, even thou Gspot has the field order wrong.

A decoder can generate an RGB24 image from a DV file. I don't know what the Info() function does (never used avs). If it is telling you what the decoder is generating, it's believable. If it is supposed to tell you the native format of the information in the file, it's wrong.

The following is the script that I run, and since I'm not using any decoder, because I'm just display it on the screen, I would assume it's native.

#Load Video Source
AviSource("D:\VideoCapture\Nicks70thBirthdayBash20 06_Type2.avi")
# Retrieve Video File Info
Info ()

This is the feed back, I'm getting.

That's why our software does even bother to rearrange the data into a traditional frame. It works directly on the decompressed data in the same way they are arranged in the DV file).

Which software is this?


Thanks.

Matt

[edDV]

VirtualDub will use a Video For Windows DV codec like the Panasonic to decode to RGB24. All frames are decoded.

Vegas in DV project setting maintains native DV format except for frames that are filtered. Those frames are decoded, filtered and recoded to DV format.

[JohnnyMalaria]

FYI, VirtualDub uses RGB32 with our decoder (see the tell-tale 'RGB32v': http://www.enosoft.net/products/enodvdec/How-To/VDub/gif/12%20-%20In%20and%20out%20windows.gif )

RGB32 is often preferable because each pixel is stored with 32-bits - a very convenient size for use with modern CPUs. RGB24 uses 24-bits and is quite clunky - it doesn't suit the CPU's architecture very well. Hence, processing RGB24 frames is often slower.

Re the Info(), it must be telling you about the decoded frame. It lists a "video pitch" - which is basically the number of bytes from the start of one line to the next. The concept of "video pitch" is meaningless in the native DV world.