A question about A/D conversion methods/formats?

[DNICE_ONE]

Hi!

Even though I have been consistantly reading this forum about capture cards and devices which can be used to convert analogue to digital, I am still confused about which way to go. Some use internal capture cards and others use DV capture devices. Can someone please clarify which format exactly the Canopus ADVC-110 caps in (DV AVI or Raw AVI)? I read somewhere LS said that colourspace conversions deteriorate the video quality so which method does not require colourspace conversion?

Also I do not intend to do major editing on my converted videos just cutting and splicing (parts showing my feet etc.). So is it possible to do this kind of editing with MPEG2, with which software and will it affect the video quality?

Hopefully after I am clear about these things I should be able to make a better choice. Any feedback will be appreciated. Thanks.

[jimmalenko]

Can someone please clarify which format exactly the Canopus ADVC-110 caps in (DV AVI or Raw AVI)?

25Mbps DV-AVI and 1536kbps WAV audio - 13.2GB per hour or thereabouts.

[junkmalle]

The ADVC-110 converts analog video to a DV stream (about 5:1 compression compared to raw YUY2) and then streams it to the computer. The computer captures the DV stream and puts it in a AVI container -- about 13 GB/hr.

DV is a good editing (cut/paste) format because each frame of video is self contained. You can cut the video anywhere you want without uncompressing and recompressing any of it.

MPEG is generally not as good for cut/paste editing because most frames only encode the difference between frames. You typically have 1 key frame (self contained image) followed by 14 frames that only contain the changes from frame to frame (called a "group of frames" or GOP). If you cut in the middle of the GOP, at least that GOP has to be reencoded. So there will be some loss of quality and the possibility of A/V sync problems. Some less smart editing program will simply reencode the entire video if you make any changes.

The whole colorspace conversion thing is very contentious. In my opinion, although there are some cases where the conversion of DV (4:1:1 subsampling) to MPEG (4:2:0 subsampling) can lead to visible artifacts (smearing of colors), and you can certainly create test images that show the problems, most real world video does OK going through a single 4:1:1 to 4:2:0 conversion.

[davideck]

The whole colorspace conversion thing is very contentious. In my opinion, although there are some cases where the conversion of DV (4:1:1 subsampling) to MPEG (4:2:0 subsampling) can lead to visible artifacts (smearing of colors), and you can certainly create test images that show the problems, most real world video does OK going through a single 4:1:1 to 4:2:0 conversion.

Do you have any test captures showing this "smearing" that you could post? 4:1:1 sampling should be more than adequate for any consumer tape source. Perhaps the quality of the sampling/reconstruction hardware is an issue.

Some broadcast quality professional gear utilized 4:1:1 back when memory was expensive. My FOR-A TBC is 4:1:1 and is noticeably more transparent than my 4:2:2 Datavideo TBC.

Dropping half of the chroma vertical resolution by sampling any source at 4:2:0 seems like the more detrimental operation...

[DNICE_ONE]

DV is a good editing (cut/paste) format because each frame of video is self contained. You can cut the video anywhere you want without uncompressing and recompressing any of it.

The whole colorspace conversion thing is very contentious. In my opinion, although there are some cases where the conversion of DV (4:1:1 subsampling) to MPEG (4:2:0 subsampling) can lead to visible artifacts (smearing of colors), and you can certainly create test images that show the problems, most real world video does OK going through a single 4:1:1 to 4:2:0 conversion.

Thanks Junkmalle for your very informative reply it helped me a lot. But is it the same case with raw AVI when it comes to cutting and pasting?

Surely the results cannot be that bad when converting 4:1:1 to 4:2:0 otherwise a lot of people who convert to DV and then to MPEG2 for DVD must have stopped using the ADVC altogether.

I am a bit curious to know if the raw AVI can be edited as easily and without a lot of fuss as the DV.

Thanks for the feedback.

[junkmalle]

I am a bit curious to know if the raw AVI can be edited as easily and without a lot of fuss as the DV.

Raw AVI capture cards usually capture in the YUV colorspace with 4:2:2 subsampling. This is very close to how video is broadcast. You can save that uncompressed (about 70 GB/hr), losslessly compressed with something like HuffYUV (about 30 GB/hr), or with any lossy compression you choose (DV, MPG, Divx, etc).

Editability will vary depending on the codec. Uncompressed is easily edited but the files are very large. Codecs like HuffYUV (lossless)and MJPEG (Motion JPEG, lossy but can be very good) create self contained frames which can be cleanly edited. The high compression codecs like Divx and Xvid use very large GOP sizes like 300 -- making them difficult to edit. You're also at the mercy of your editing software -- it may not like certain codecs.

The main down side of raw YUV capture is the greater system resourced needed. With DV capture the computer is receiving 13 GB/hr from the firewire port, doing very little processing, and saving that much to disk. With raw uncompressed YUY2 capture you're receiving 70 GB/hr, doing little processing, and saving 70 GB/hr to disk. This makes it much more succeptible to dropped frames. If you're compressing the raw YUY2 data then the CPU has to do a lot of processing but you are saving fewer GB/hr to disk. You're still more likely to drop frames than with a DV capture.

Oh, another issue with raw YUV capture -- some cards only capture video and require you to use your sound card to capture the audio. This can easily lead to A/V sync problems.

Regarding the ATI AIW cards: There are lots of people making great captures with them. On the other hand I see a lot of messages here about problems with drivers. Only certain versions seem to work properly. I wouldn't use an AIW card on a computer that is used for say, playing games (where you need the latest drivers and are updating them all the time), as well as video capture.

One last thing you should consider: if you're planning on mostly cut/paste editing (no fancy transitions, overlays, etc) then you might consider a hardware MPEG capture card like the Hauppauge PVR-250. The results aren't as good as say a raw YUV capture followed by a well done, time consuming, 2-pass software MPEG compression. But the capturing is nearly foolproof (you can continue using your computer while capturing), and the MPEG2 files can be complaint with DVD specs. Combine that with a smart editor like Womble's and you have fairly clean, quick and easy process.

[Cornucopia]

You keep saying "RAW AVI" like it's so separate from other forms of AVI. Unless you mean "uncompressed 4:2:2 YUV" (which would be ~20MB/sec) or RGB or a losslessly compressed version of that, all other caps will have some form of compression (with quality loss) and colorspace contraction (from 4:2:2 to 4:1:1 or 4:2:0).
Most consumer cap cards give you either straight DV-encoded AVI, straight MJPEG-encoded AVI, MPEG or allow you to choose the codec, colorspace, framerate and resolution downstream from the hardware A-to-D conversion (which is almost always ITU-601 based--4:2:2 YUV 720x480/576@29.97/25fps internally).
DV and MJPEG are Intraframe-only compressors, so they and uncompressed YUV are ALL good for editing. MPEG can be Intraframe-only or (usually) both Intraframe and Interframe compressing. If I-frame only, it is still good at editing, if not, it's much more complicated and is best avoided by neophytes.

Scott

[DNICE_ONE]

Thanks junkmalle and cornucopia your feedback was very elaborate and most helpfull. Now I understand the concept and will decide which way to go with much ease. Thanks again.

[junkmalle]

Thanks junkmalle and cornucopia your feedback was very elaborate and most helpfull. Now I understand the concept and will decide which way to go with much ease. Thanks again.

Let us know which way you decide to go. And your reasoning behind the decision.

[DNICE_ONE]

Let us know which way you decide to go. And your reasoning behind the decision.

Will do surely.

While we were discussing colourspace I found a great link about colourspace. Tell me what you think of it?

http://www.animemusicvideos.org/guides/avtech/colorspace.html

[junkmalle]

While we were discussing colourspace I found a great link about colourspace. Tell me what you think of it?

http://www.animemusicvideos.org/guides/avtech/colorspace.html

It has some minor details wrong (or at least you should take them figuratively not literally) but overall it's fair.

Some more resources:

http://davis.wpi.edu/~matt/courses/color/
http://en.wikipedia.org/wiki/YUV
http://en.wikipedia.org/wiki/Chroma_subsampling

For excruciating detail (but assumes you understand the basics of colorspaces):

http://www.fourcc.org/

[DNICE_ONE]

It has some minor details wrong (or at least you should take them figuratively not literally) but overall it's fair.

Thanks for the links junkmalle I will check them out.

[lordsmurf]

While we were discussing colourspace I found a great link about colourspace. Tell me what you think of it?

http://www.animemusicvideos.org/guides/avtech/colorspace.html

It has some minor details wrong (or at least you should take them figuratively not literally) but overall it's fair.

Yeah, I think so too.

[payton34]

Hi everyone,

While on the subject of formats and codecs I have a question regarding HUFFyuv.

As of today I am the proud owner of an ATI All In Wonder X800XT. This it is a graphic/TV tuner card. I down loaded and installed the HUFFyuv codec version 2.1.1. While on the digitalfaq site I noticed in his ATI AVI capture guide he showed a menu in the ATI software where his version of HUFFyuv had another option where you could set "field threshhold."

The menu showed his version to be "2.1.1 CCESP Patch v0.2...."
I went to the download site for HUFFyuv off this site and only found a beta version "220". I went and installed it and I now also have that option of setting "field threshhold." Everything has been capturing great thus far.

My only question was why is my version only a beta described as "220" and his seems to be something different, described as above?

Is there a difference between the two?

Where can I get this other version?

Should I get this other version?

Thanks

[lordsmurf]

Whatever works is fine.