Doom9's Forum - View Single Post

mariush · 20th January 2011, 02:26

The s-video and composite signals are all analogic, color. The USB capture card or the USB stick converts this analogic signal in a digital one and sends it to applications encoded in a specific color space.

Color spaces are types of encoding of the color information for each pixel. The most basic one is RGB where each pixel is represented by 3 bytes, red , green and blue. YUY2 and YV12 are two types of pixel encodings in which the capture device takes advantage of the fact that humans notice variations of luminosity and color differently so they use less bytes to store one of these components so for example YUY2 encodes 2 pixels in 6 bytes and in the case of YV12 it encodes 4 pixels in less space than what YUY2 would use to encode 2 pairs of pixels.

See this for a better explanation: http://www.animemusicvideos.org/guid...olorspace.html

Now normally, you would think ... hey I want RGB to get the best quality, but the fact is that DVDs, x264, xvid, all lossy codecs use YV12 (and very few also allow the slightly better YUY2) to encode the content, and the difference between RGB and YUY2 or even YV12 is so small when capturing from a VHS, it's not worth capturing in RGB if it's possible to record in YUY2 or YV12.

When you go in Virtualdub to Video > Set Custom Format and set there YUY2 or YV12, Virtualdub simply tells the usb capture device driver to switch from RGB to one of those color spaces - being Chinese cheap device, the driver may simply not support these color spaces properly and output black and white image when it should output color image in the color space you selected. So in this case you'll just have to keep using RGB, which is OK but your computer will work harder and save more data which will be lost anyway in the end.

So the Video > Custom format tells the driver to send images to Virtualdub in that colorspace. If you don't save raw data to disk (and you shouldn't, it doesn't make sense), you can tell the codec you select to compress the video part to convert from RGB to YUY2 or YV12 and save disk space. You just go to Video > Compression and select Lagarith or Huffyuv (you have to install them first though, before you start Virtualdub) and you click on the Configure button and select the color space you want. I'd suggest using YUY2 if possible,

Lagarith and Hufyuv will use about 30% of a quad core processor to encode a 720x576 25 fps video. On an older dual core processor, it uses about 70-80% of the processor but the usb capture device requires much more processor compared to a TV tuner card like the one I mentioned - the USB device may use up to 10-20% of the CPU when running.

So the reason why I'm not recommending to compress the audio during capture is because the lossless compression of the video and the usb capture device already use plenty of processing power so the 5-10% the audio encoding uses, could cause Virtualdub to skip video frames. The audio data is about 170-200 KB/s of disk space, so it's really not a problem compared to 4-5 MB/s of data compressed with Lagarith/Huffyuv or raw data that's about 40 MB/s of raw video.

So anyway... play with audio sampling rate, 44100, 48000 Hz, see what works best. Also try using Video > Overlay or Video > Preview coupled with Video > Preview (both fields) to see if changing this reduces processor usage - sometimes Virtualdub uses a bit of processing power to display the images it captures - setting to Overlay should usually work best but on some video card you have to use Preview with acceleration set to Both fields.

You may also want to try to capture on 24 fps or 29.97 even though your VHS is PAL therefore it should be 25 fps. With cheap Chinese USB devices, who knows...

20th January 2011, 02:26	#5 \| Link
mariush Registered User Join Date: Dec 2008 Posts: 589	The s-video and composite signals are all analogic, color. The USB capture card or the USB stick converts this analogic signal in a digital one and sends it to applications encoded in a specific color space. Color spaces are types of encoding of the color information for each pixel. The most basic one is RGB where each pixel is represented by 3 bytes, red , green and blue. YUY2 and YV12 are two types of pixel encodings in which the capture device takes advantage of the fact that humans notice variations of luminosity and color differently so they use less bytes to store one of these components so for example YUY2 encodes 2 pixels in 6 bytes and in the case of YV12 it encodes 4 pixels in less space than what YUY2 would use to encode 2 pairs of pixels. See this for a better explanation: http://www.animemusicvideos.org/guid...olorspace.html Now normally, you would think ... hey I want RGB to get the best quality, but the fact is that DVDs, x264, xvid, all lossy codecs use YV12 (and very few also allow the slightly better YUY2) to encode the content, and the difference between RGB and YUY2 or even YV12 is so small when capturing from a VHS, it's not worth capturing in RGB if it's possible to record in YUY2 or YV12. When you go in Virtualdub to Video > Set Custom Format and set there YUY2 or YV12, Virtualdub simply tells the usb capture device driver to switch from RGB to one of those color spaces - being Chinese cheap device, the driver may simply not support these color spaces properly and output black and white image when it should output color image in the color space you selected. So in this case you'll just have to keep using RGB, which is OK but your computer will work harder and save more data which will be lost anyway in the end. So the Video > Custom format tells the driver to send images to Virtualdub in that colorspace. If you don't save raw data to disk (and you shouldn't, it doesn't make sense), you can tell the codec you select to compress the video part to convert from RGB to YUY2 or YV12 and save disk space. You just go to Video > Compression and select Lagarith or Huffyuv (you have to install them first though, before you start Virtualdub) and you click on the Configure button and select the color space you want. I'd suggest using YUY2 if possible, Lagarith and Hufyuv will use about 30% of a quad core processor to encode a 720x576 25 fps video. On an older dual core processor, it uses about 70-80% of the processor but the usb capture device requires much more processor compared to a TV tuner card like the one I mentioned - the USB device may use up to 10-20% of the CPU when running. So the reason why I'm not recommending to compress the audio during capture is because the lossless compression of the video and the usb capture device already use plenty of processing power so the 5-10% the audio encoding uses, could cause Virtualdub to skip video frames. The audio data is about 170-200 KB/s of disk space, so it's really not a problem compared to 4-5 MB/s of data compressed with Lagarith/Huffyuv or raw data that's about 40 MB/s of raw video. So anyway... play with audio sampling rate, 44100, 48000 Hz, see what works best. Also try using Video > Overlay or Video > Preview coupled with Video > Preview (both fields) to see if changing this reduces processor usage - sometimes Virtualdub uses a bit of processing power to display the images it captures - setting to Overlay should usually work best but on some video card you have to use Preview with acceleration set to Both fields. You may also want to try to capture on 24 fps or 29.97 even though your VHS is PAL therefore it should be 25 fps. With cheap Chinese USB devices, who knows...