Capture Test result on downloadable DVD

[maa]

I made a downloadable test DVD 5.7mb comparing two analog capture cards and capture via Firewire with a Canon Camcorder.
This was as a result of this thread:
http://forum.doom9.org/showthread.php?s=&postid=450326#post450326
The test shows the final product including encoding to mpeg with CCE. No filters were used. The source was a DVD player.

The DVD contains four 2 second clips running in a loop
ATI Rage 128 Pro ViVo
Phillips TV Tuner card
Canon MV450i
Original VOB


I thought I'd repost it as it got buried in the other thread.
Download the test DVD here - its 5.7 mb
CapTest (http://www.photosyndikat.de/guides/more/CapTest.zip)

[Arachnotron]

What may be an issue here is that the Philips, the camcoder/DV and DVD all are 27 MHz clock (ITU?) devices. The ATI is the odd one out. I don't know what clock it runs under, but it is something else.

When I did some comparisons (SAA7134, SAA7113, CX23881 and BT878) using a DVD as source, the Philips cards came out on top. But when capping VHS tape, the differences were not so clear. I cleaned up the files, so I cannot post them :(

My theory at the time was that the identical clock combined with the accurate sync timing you get from a DVD -no horizontal jitter- each sample will be perfectly aligned with the original samples from the DVD source. This is not however an accurate representation of how a device performes with a real analogue source.

Anyone cares to comment?

[Inc]

Thats exactly the issue I discussed here:

http://forum.doom9.org/showthread.php?s=&threadid=67560&highlight=bt878

As you see the Philips/Zoran DC30 and the BT878, both did result in nice images if captured a DVD Source, means a clear signal.

But if capturing a TV Tuner signal via a sVHS VCR or a VHS Tape, the BT878a (well, the model which is tested there) did result in these horizontal artifacts as in the thread there people do say that the Philips chip also got an "Antialaising" filter.

[Arachnotron]

as in the thread there people do say that the Philips chip also got an "Antialaising" filter

All digitizing equipment has an anti-aliassing filter. They are essential to their functioning. There are some differences in implementation, but all capturing devices will have either an (external) analogue anti-aliassing filter, or one build-in into the chip (Philips). Leaving such an component out would result in aliassing of the analogue signal. For instance, both my BT878 and CX23881 card have external low-pass filters on them.

The internal anti-aliassiing filter in the SAA7134 seems to be programmed at a cut-off of 6.75 MHz, which is also the maximum frequency a DVD player can deliver and way above the maximum for VHS.

As to other filters, you will have noise filters and sharpening filters (peaking) which can be switched on or off by default by the drivers and also can make a difference from device to device.
So far I have found the default Phillips driver for the SAA7134 switches the peaking filter off by default, as does the BTWincap driver for the BT878.

All in all, at least for the Philips I don't think the explanation is in these filters.

Something else I just made up :D

Analogue sources like VHS, s-VHS and broadcast all have a very low bandwidth available for chroma (typically in the 0.5-1 MHz range) while with DVD the bandwidth for chroma is much higher, up to half that for Luma (6.75 MHz/2)

So, if I designed a capture chip for capping TV and tape, I would put in a low-pass filter for chroma to get rid of everything above 1 MHz, since that would be noise. With a standard source, you would not lose any detail and the cap would look cleaner. But with DVD, which has higher frequencies in the chroma, the result would be less good.

Mind you, I have no idea if this is indeed the case for the ATI chip. Just some theorizing on the side.

But one thing I do believe: you should be very carefull with drawing conclusions about the quality of devices for analogue capping when using a DVD player as a test device.