PDA

View Full Version : Yet another Capture Test ... (DV devices)


trevlac
21st July 2004, 05:38
How much do you wanna spend?

I did some testing of DV device captures, and wanted to share the results. I basicly tested for resolution, noise, and color. There are more things to consider (like ease of use, macrovision, audio sync, TBC, etc.) but I didn't look at those.

The round up:[list=1]
Laird LTM-5000FS - A professional YC DV device ~ $1800
Sony TRV900 - 3CCD minDV cam ~ $1500
Sony TRV520 - D8 cam ~ $1000
Sharp VL-NZ105 low end minDV cam ~ $300
AverMedia BT878a based PCI card ~ $30
MSI TV@nywhere Master CX23883 based card ~ $50
[/list=1]

The prices are around what the devices cost new (for comparison). You can get them about 50% cheaper used.

The Test
I used a really cheap apex pd-450 dvd player. From it (s-video out), I played test images from the Avia test DVD. The Sharp does not have s-video. So I used composite. Each device digitized the signal. For all of the DV devices, I used winDV to capture the test images. All but the TRV900 have passthru. I had to record to tape 1st for the TRV900. For the Aver and MSI, I used amcap and captured YUY2 uncompressed. BTWincap and MSI drivers respectively. I grabbed the images in virtualdub and made the compares in Photoshop. The images here are compressed, but they faithfully show the results. For the base DVD images, I grabbed the frames from the VOBs with vdubMod.

Resolution
http://trevlac.us/pics/Res.jpg

For this we want to look at the circle, the H rez chart at the top, the background, and the 6.75 text.

The Laird has the highest resolution. The detail in the circle is mostly there. It also has quite a bit of noise. I'm not sure if this is comming from the DVD player, or the Laird. Look around the 6.75 text. The grey is also a bit dark.

The D8 looks to have better rez than the TRV900. I like the D8 image the best. It has very little noise (actually, the unit has a DNR option. I had it off.)

The Sharp suffers from the composite signal. This pic is B&W for compression purposes, but the sharp produced all kinds of color noise. The other images were ony grey scale.

The BT878 has a lower rez and is a bit dark. I'd rank it above the Sharp.

The CX is very dark. I used the MSI drivers defaluts. Unlike the laird, the entire picture is dark (look a the bars below). This should be easy to correct by upping the brightness a tad. It's resolution is just below the laird. There is also not as much noise. Somehow it looks a little 'oversharp' to me. Again, this could be the drivers which is a general problem with the CX cards.


Color
http://trevlac.us/pics/Bars.jpg

These color bars are 75% on the top and 100% on the bottom. Look for noise and bleed.

The Laird is the worst here. I'd guess it needs to be serviced. The Red, Magenta, Blue appear 10-15% too low. If you zoom, you can also see a bit of noise.

The sony's are about the same. Very little noise. Maybe a bit dark on the chroma colors. The TRV900 may be better on the edges between colors.

The Sharp has poor edges between colors. Look between red and blue.

I like the BT878 the best here. Could be due to 4:2:2 vs 4:1:1.

The CX is again dark. This should be correctable with a driver setting. It looks no better than the BT to me.


An Extra test
For kicks, I also did a test with a cheapo Recoton switch I have. This switch excepts s-video and RCA inputs and sends it to both RCA and s-video out. Well, I compared straight s-video to the Laird; with s-video into the switch and s-video out of the switch to the Laird. Let me repeat, I was not trying to use composite, just s-video on the switch.

http://trevlac.us/pics/Colors.jpg

The switch does a ton of damage to the signal. Needless to say, I don't want it in my capture path.


Some conclusions?
S-video makes a big difference. The luma resolution on the DV devices is better than the BT. The CX has a higher res than everything but the laird. This of course may not matter depending on the source signal. The colors on the BT are better. On real moving images ... I'd doubt I could pick out anything but the low colors and the s-video/composite resolution.

:D

Edit Added the CX test.

jggimi
21st July 2004, 15:00
:goodpost:

Arachnotron
21st July 2004, 17:47
Hey Trev,

Interesting test! A quick question: at which resolution did you cap with the BT878a?

trevlac
21st July 2004, 17:51
Originally posted by Arachnotron
A quick question: at which resolution did you cap with the BT878a?

712x480

Is my circle round ? :D

I should also try my CX too. But I never really use that card, as you know. ;)

Arachnotron
21st July 2004, 20:00
I just vaguely remember seeing something at 6.75 MHz in my BT878 caps, while in yours it is completely grey. That would indicate the BT can do even better, and the limitation is the analogue prefilter, not the chip.

But naturally, no old testcap survived on my HD and I don't have the card in any working machine right now. :(

trevlac
21st July 2004, 22:03
Originally posted by Arachnotron
I just vaguely remember seeing something at 6.75 MHz in my BT878 caps, while in yours it is completely grey. That would indicate the BT can do even better, and the limitation is the analogue prefilter, not the chip.


Could be, but IIRC, it might be a PAL vs NTSC thing. In the spec sheet the BT's resize filter hits at about 640 for NTSC and higher for PAL.

I really do like my BT based card. A filter can be a good thing for real source. On the other hand, a dv device is much easier to use. If I can get the Laird adjusted, I will be very pleased. No macrovision issues. It has a full frame sync (like a TBC). This should give me much more stable pics, and nice audio sync. I can also send DV back out for monitor preview (hopefully more controlable than TV out). Yada yada yada .... BTW: I found it used for ~245 Euro.

Since I'm on the topic. I noticed a sale (at a retail store here [Staples to be precise]) where a dual format 8x dvd burner could be had for about 40 Euro. :D

Arachnotron
21st July 2004, 22:19
Could be, but IIRC, it might be a PAL vs NTSC thing. In the spec sheet the BT's resize filter hits at about 640 for NTSC and higher for PAL. Aahh, that was it. How fast we forget :D

But this is a dangerous road you are traveling my friend. One day you will wake up to the sound of a fully automated black box spitting out DVD's for your children to watch. And you will find you have no reasons to visit here any more.... ;)

...To get back on topic: is there much difference in horizontal color separation? Between the CX23881 and the SAA7134 there is quite some difference.

You havent convinced me to go DV quite yet. But who knows, after a few more rounds with my Hauppauge card I might consider it..

trevlac
22nd July 2004, 04:04
Originally posted by Arachnotron

One day you will wake up to the sound of a fully automated black box spitting out DVD's for your children to watch. And you will find you have no reasons to visit here any more.... ;)


Hardly ... :) HDTV will be here soon and noone will want to watch those DVDs. Just like I can't imagine why anyone today would create a VCD. :)


...To get back on topic: is there much difference in horizontal color separation? Between the CX23881 and the SAA7134 there is quite some difference.

I'm not sure what this means. Comb filter or something else? Do you have a good test pic for this? I find that most source has such a low color res, that 4:1:1 or 4:2:0 vs 4:2:2 does not make much of a difference. I'm sure it can be done poorly though.


You havent convinced me to go DV quite yet. But who knows, after a few more rounds with my Hauppauge card I might consider it..
I havn't convinced myself. DV is easier, but not really better.

Just to get back to how good we have it in the US (with regard to electronics prices), I recently noticed that settop dvd Recorders can be had for around 150 Euro. That has got to be easier yet for some things. :D

Arachnotron
22nd July 2004, 10:51
Just to get back to how good we have it in the US (with regard to electronics prices), I recently noticed that settop dvd Recorders can be had for around 150 Euro. That has got to be easier yet for some things. El-cheapo no name recorders are still doing EUR 250 over here. Something remotely decent starts at EUR 400.

I'm not sure what this means. Comb filter or something else? Do you have a good test pic for this? I find that most source has such a low color res, that 4:1:1 or 4:2:0 vs 4:2:2 does not make much of a difference. I'm sure it can be done poorly though.
Any testpic with a sharp horizontal border between colours will do. Like horizontal colour bars. I agree that considering the amount of color info available in the average source, it probably won't make much difference.

FlimsyFeet
22nd July 2004, 14:11
trevlac, last time you tested the BT878 the results were better:
http://forum.doom9.org/showthread.php?s=&postid=518702#post518702
http://trevlac.us/pics/712Full.png
was this not a test under similar conditions?

trevlac
22nd July 2004, 15:19
Originally posted by FlimsyFeet
trevlac, last time you tested the BT878 the results were better:
http://forum.doom9.org/showthread.php?s=&postid=518702#post518702
http://trevlac.us/pics/712Full.png
was this not a test under similar conditions?

Gosh, I think you are correct. There is was probably something wrong with the BT vs DV test. I was really testing the DV devices, but I needed a reasonable set of card results for comparison.

As arachno pointed out here (and Ivo did off line), the BT should be better. Even reading the specs say I should get 480TVL or ~ 6MHz. The pic above does not even look to be 450TVL.

I will redo the test.

Thanks

trevlac
23rd July 2004, 04:05
Originally posted by FlimsyFeet
trevlac, last time you tested the BT878 the results were better:
http://forum.doom9.org/showthread.php?s=&postid=518702#post518702
http://trevlac.us/pics/712Full.png
was this not a test under similar conditions?

I added a new cap with the BT. On my 1st try, I had set the 'sharpness' option on and to 50 (the default). For the BTWincap driver, this causes the internal peaking filter to be activated and in this case, actually reduce the resolution. Some sharpness huh?

Anyway, I wrote about this at the end of the page on the following thread. I had to read the registers, look at the code, and read the specs to figure this out before. I don't have the energy for that, so I'm sticking with my story :D.

http://www.videohelp.com/forum/viewtopic.php?t=199669&start=120

BTW: The circle in that old pic does seem to have a hair more detail, than my latest test. I'm going with different DVD players as an excuse. The one I use to test now is a portable Apex I got from Wallymart for about 110 EUR. 5" LCD and even s-video out. Keeps the kids happy on long car trips. :sly: