capturing to huffyuv

[nagihcim1]

I read on http://www.100fps.com/file_sizes_of_lossless_compression.htm that capturing to huffyuv was lossless but capturing to huffyuv+yuy2 was lossy. My options for colorspaces are rgb16, rgb24, yuy2, yvu9 & I420. Should be using another colorspace to get lossless compression when capturing? Or am I totally reading this wrong?

[jagabo]

Most capture cards capture as YUY2 or a similar 4:2:2 subsampling. This is closest to the actual analog signal. Compressing that with HuffYUV or Lagarith (better compression, slower execution) is lossless.

I'm not aware of any capture card that actually captures in RGB. But if have such a card and use HuffYUV's RGB -> YUY2 conversion you will lose a little precision in that step. Not enough to worry about in my opinion. The one place where this might be a problem is with computer generated RGB animations that are not destined for viewing on TV.

[nagihcim1]

I have a ATI TV Wonder Pro and I'm using Ulead 11 to capture

[jagabo]

That card surely captures in YUY2. If you wish to save that losslessly you can use HuffYUV or Lagarith. Lagarith may not be fast enough for capture -- you'll have to test it.

[nagihcim1]

I know my card can capture yuy2. I was just confused because the site I referenced says that using huffyuv+yuy2 will be lossly, is this incorrect?

[jagabo]

If your source is RGB, HuffYUV in YUY2 mode is not lossless because the conversion from RGB to YUY2 will lose some detail. If your source is YUY2 then HuffYUV in YUY2 mode is lossless.

[nagihcim1]

I'm not sure what my source is. I'm recording off of Dish Network through a regular coax cable

[edDV]

I'm not sure what my source is. I'm recording
off of Dish Network through a regular coax cable

Dish Network sends normal SD programming at about 3.5Mb/s VBR
524x480 interlace MPeg2. Your main control of quality is to use S-Video
instead of composite or RF connection to the AIW card. The
AIW Pro is an older card with a poor tuner and poor Y/C separator
by today's standards. S-Video is your best quality capture input.

S-Video is analog Y (luminance) and C (NTSC Modulated). The AIW will
capture to Y and Cb Cr (AKA U and V) and if you capture that at YUY2
you are getting the best the card can output.

Don't kid yourself that this is high quality input but it is the best you have.
Do you need to capture "lossless" to huffyuv at ~75 times the bandwidth
of the original? That depends what you want to do next. Most will edit then
compress to MPeg2 or MPeg4.

You can also capture directly to MPeg2 with the AIW.

[jagabo]

Short answer: capture YUY2, compress with HuffYUV. The resulting AVI file will contain the best your card can capture and every bit of it will be saved losslessly.

If your source is RGB, HuffYUV in YUY2 mode is not lossless because the conversion from RGB to YUY2 will lose some detail. If your source is YUY2 then HuffYUV in YUY2 mode is lossless.

I'm not sure what my source is. I'm recording off of Dish Network through a regular coax cable

What I meant by "source" is what your capture card captures as. Your card captures as YUY2 (or some similar 4:2:2 YUV format). If you use any of the RGB options the card still captures as YUY2 but it converts to RGB before saving to a file.

[nagihcim1]

sry I guess it will probably be best for me to try out a bunch of different things and see what works best. Thanks for getting me pointed in the right direction.