DV vs MPEG2 SD

[dellsam34]

I see a common question in the forums which is: Which capture device should I get DV or MPEG2 ?
I could answer this question by capturing an analog video tape with both capture devices but I took a different approach for now.
Using Pinnacle Studio starting from a test pattern picture in a higher resolution than both PAL and NTSC standards and a mp3 320kbps audio file, I created several video files in the following standards:
- MPEG2 SD, NTSC, 720x480 pixels, 4:3, 29.97 fps, 9 Mbps, 4:2:0, 8 bits
- MPEG2 SD, PAL, 720x576 pixels, 4:3, 25 fps, 9 Mbps, 4:2:0, 8 bits
- DV NTSC, 720x480 pixels, 4:3, 25 fps, 30 Mbps, 4:1:1, 8 bits
- DV PAL, 720x576 pixels, 4:3, 25 fps, 30 Mbps, 4:1:1, 8 bits - I cannot create 4:2:0 which should be the standard for PAL DV

If you are ok with the difference in the digital files, you will most likely be ok if the source was to be an analog tape.

Download Link: http://www.mediafire.com/download/pfl9192ig2x6ze2/MPEG2SD_vs_DV.zip

[dellsam34]

Notice from the samples that MPEG2 SD has more color resolution horizontally, while DV has more color resolution vertically, Except for PAL DV which is normally 4:2:0 just like MPEG2 SD, I couldn't get Pinnacle to output it.

[Brad]

It's a nice idea, but the comparison isn't realistic.

MPEG-2 normally uses interframe compression, and you have done so here. But it's a static test pattern. A still frame source is completely useless for judging interframe compression quality.

Does Pinnacle Studio convert files in realtime or faster? Does this file conversion use the same settings and algorithm as a realtime capture would?

[dellsam34]

The main purpose for this test is to show the loss of chroma of the DV standard, Besides chroma I don't see much of a difference, My DV caps were more stable than MPEG2 hardware compressed ones, Edirol VMC-1 internal procamp handles frame drop and video stability better than any device that I have ever come across, As to your questions about pinnacle studio, I can't tell I was using a still frame not a video source.
I'm waiting on the BlackMagic SDI ADC and then I will do a real tape capture MPEG2 compressed vs lossless vs DV. It won't be a home video tape nor a commercial one, I will probably record it myself, If you have a suggestion on what material should I put in the tape let me know.

[Cornucopia]

Good stress test elements:
Water/waterfalls/surf
Fire
Trees/grass with fluttery leaves/blades
Pans, esp. Racing (auto, horse, dog)
Shots of sky w fade to black. (works to transition & posterization problems)
Fast cut edits

Scott

[dellsam34]

Good stress test elements:
Water/waterfalls/surf
Fire
Trees/grass with fluttery leaves/blades
Pans, esp. Racing (auto, horse, dog)
Shots of sky w fade to black. (works to transition & posterization problems)
Fast cut edits

Scott

Thanks I will see if I have a DVD with that kind of materials, otherwise I will see if I can find something in the local stores, Which is better to do a test with S-VHS tape or VHS tape ? If VHS is the case I might go to the thrift store and get one of those workout tapes.

[Cornucopia]

DVD?!!

You need an uncompressed (or extremely lightly compressed) source to start from.

If you started from a heavily compressed source such as DVD, the artifacts are already baked into the encoding. Encoding again won't generate nearly quite the same amount of artifacts (due to the asymptotic action of entropy - you've already removed the "most difficult" portions).
And this would unfairly give advantage over a re-encoding by the same codec type vs. a different codec. So if you have a DVD (encoded as MPEG2) and decode it, then re-encode it once as MPEG2 and alternately as DV, the DV will have more artifacts than the MPG2 because the "action" of compression is different (being applied in a different manner). That would be a bad test.

VHS is also not a good source, because it is already of worse resolution (luminance) and much worse resolution (color/chrominance) than either MPEG2 or DV formats (unless at rediculously reduced bitrates). In a sense, it is heavily "pre-filtered", with the only thing being a challenge to capture is the timebase, which should be corrected with a TBC anyway. Without a TBC, MPEG2 would be at a disadvantage comrared to DV, by nature of the interframe encoding (unless you're comparing MPEG2-All-Intraframe setups).

Scott

[jagabo]

Yes, a better test would be a standalone media player or old Blu-ray player playing high quality HD material, down converted to s-video or composite.

[Brad]

I see a common question in the forums which is: Which capture device should I get DV or MPEG2 ?

VHS is also not a good source, because it is already of worse resolution (luminance) and much worse resolution (color/chrominance) than either MPEG2 or DV formats (unless at rediculously reduced bitrates).

I think VHS is the best source since that's what most users asking about DV or MPEG-2 devices these days are capturing (and to a lesser degree, other consumer tape formats). I don't see many posts these days asking about archiving current SD TV broadcasts.

[Cornucopia]

@vaporeon800, those 2 are 2 of the most likely candidates (short list would probably also include uncompressed/lossless/near-lossless, and direct-to-h264), but if you are doing a "shootout" you want to start with a pristine source to compare against, not the usual crappy VHS, even though that will be the standard day-to-day source format. With VHS, so many things are already smudged, you won't be able to tell which problems are due to the capture/compression and which are due to the poor source. Not so with pristine digital HD.
Then you would want to use individual stressors on that pristine beginning, to see how they (DV, MPEG2, etc) react, without other confounding/extraneous/complicating variables.

Of course, both should be easily able to handle VHS material, given proper bitrate & NR/TBC/ProAmp pre-processing. IMO, from years of doing these very caps, the choice is down to:
Bitrate (and thus filesize and thus disk capacity), Types/Amount of Artifacts, Editability/Archivability/Robustness
And the priorities of those is determined by the end user's needs & expectations with workflow and how valuable he/she sees the assets. Plus, that ephemeral "the eye of the beholder" variable.

Scott

[dellsam34]

I have a little problem with HD materials, First I don't have a media player that outputs S-Video, only composite. Second, My HDMI to S-video converter generates frame tearing in fast motion scenes and don't ask me to find a better one because I tried few already and this one was the best (haven't tried $$$$ ones though, it's not worth it).
Now if I have to do this digitally using pinnacle I can do it but the Analog to digital part will be missing and that's what capturing is all about, It will be done only to demonstrate the encoding differences, But once again I will be testing Pinnacle encoding capabilities in different formats not the capture devices, So I think the test will be useless, I did the test card here just to show the chroma differences that's all.
So if I have to do devices, I need an analogue source to begin with, I feed it to the S-Video input of my VCR and either output the S-Video out to the devices live or do S-VHS/VHS tape recording and output using S-Video, maybe I'll do all, So what analog source would be considering that I can't use my HDMI downscaler ? I have a home theatre receiver I may give it a go, It will be a PITA to pull it out and do the connections.
There is another issue, Should I use TBC/frame sync or leave it off on the VMC-1 ? Because on the MPEG2 part (Blackmagic ADC Mini converter) will be no TBC available.
I will be using firewire for the VMC-1 and USB 3.0 for BlackMagic/Magewell (Analog-SDI-USB3.0) combination.
I will let you know when I get the SDI box.

[Cornucopia]

So take good HD material and downconvert to very high quality (lossless) SD material. It's only for a test, so it doesn't need to be very long, as long as your subsystem can handle the bitrate.

Can't do D->A->D? (even composite from gen. media player?) No extra PCs around?

Like I said, you can do S-VHS/VHS, etc., but they won't be as conclusive tests as you'd like.

...Good tests are often PITA.

Scott

[dellsam34]

My Blu-ray player has composite output, I will see if I can try my audio receiver first after I get the blackmagic box.

[dellsam34]

Ok I think I managed to find a source file for the tests, I used one of the HDV files and I was able to get rid of the frame tearing I guess due to the fact that HDV is 1080i and not 1080p, I posted a sample DV cap here, If you guys think it's a good source material for the tests I will use it, HDV file 4:2:0 played via TVIX player from HDMI output, Used pan & scan option in the player to chop off the sides of the 16/9 frame and set resolution to 1080i, From HDMI to S-Video using a converter and then fed to S-video input of the JVC S-VHS VCR in NTSC mode (PAL gave me some artifacts maybe because the source material is 29.97 fps and not 25 fps), From VCR to VMC-1 via S-video (TBC on, When off gave an artifact, see small attached sample with video source paused to pinpoint the artifact, it's there when video played as well, and watch when I turn the TBC on) and than to computer via firewire captured with WinDV as DV 4:1:1.
I got the BlackMagic box today updated the firmware already but the BNC to S-video adapter didn't make it to the mail box so I have to wait on that to start the tests.