CX2388X colour settings

[Incast]

I've purchased 2 CX2388X cards over the last year from Hauppauge and both have suffered by default a contrast that seems far too high. I've tried to adjust Brightness, Huge, contrast and saturation to produce better settings, but I find this difficult due to being colourblind. If anyone has any advice on settings they use it would be greatly appreciated.

[tedkunich]

If you have a copy of Video Essentials, Dscaler has an automated calibration of brightness, contrast, hue, and saturation. Granted the calibration is corect for only that source (LD or DVD) but it is better than nothing.


T



Originally posted by Incast
I've purchased 2 CX2388X cards over the last year from Hauppauge and both have suffered by default a contrast that seems far too high. I've tried to adjust Brightness, Huge, contrast and saturation to produce better settings, but I find this difficult due to being colourblind. If anyone has any advice on settings they use it would be greatly appreciated.

[Arachnotron]

I've had problems with a CX23881 Hauppauge card delivering much to bright caps, especially from the TV tuner. The latest driver solved that for me. (wintv2k44_22107.exe)

As a side note, Hauppauge enables the peaking filter by default (= sets the sharpness slider to 1). This causes extra (over) sharpening that was not in the source. Set it to 0.

I still have problems with intermittend interference and such. The chip itself is great, but the drivers still suck. :( I'm still hoping that someone adapts the BTWincap drivers fot it, but I'm not holding my breath. I bought another card for now.

For the rest I would keep the sliders at default and correct in post. There is no universal optimal setting anyway, analogue sources vary to much, especially vhs.

If you cannot trust your eyes, you could try using a histograph/vectorscope plugin. There is some stuff about that in the latest capture guide to get you started.

[Incast]

Thanks for your help. :)

Arachnotron did you purchase a Phillips SA7134 chip card, I'm considering one of these but am worried about what appears to be excessive blurring caused by a filter in the card.

[Arachnotron]

Arachnotron did you purchase a Phillips SA7134 chip card, I'm considering one of these but am worried about what appears to be excessive blurring caused by a filter in the card.

Yes, I did. It does look a tiny bit blurrier than CX23881 based cards, but IMHO this is caused by excessive sharpening by the conexant based cards rather than by a problem with the Philips chip.
But not veryone agrees with me on that (Hi Ivo ;) )
Anyhow, the difference is very, very small and if noone told you you would probably never notice. It's more a matter of taste than anything else.
If you like I can make some screenshots from a videoessentials DVD tonight with both cards. (I still have my CX23881) using default settings, and one from the CX with pealing disabled. The SAA has a peaking filter too by the way, only it is off by default and has slightly different characteristics.

But in all fairness, there are plenty happy CX23881 user s out there, so it should be possible to get this beast to work.

[Incast]

I would greatly appreciate those screenshots, I have searched the forum and found a Japanese comparison but a more recent one would be great! Thanks a lot for your offer.

I'm trying to make the best choice before i begin my conversion of VHS to DVD, I've purchased an SVHS player with a TBC and want to make sure i've made the right capture card choice.

[Arachnotron]

I finished the caps.

You can find them here as rar files of 0.5 Mb:
http://www.arachnotron.nl/temp/CX_sharp_0.rar
http://www.arachnotron.nl/temp/CX_sharp_1.rar
http://www.arachnotron.nl/temp/PH_sharp_0.rar
http://www.arachnotron.nl/temp/ph_sharp_2.rar

I used bmp to avoid jpg compression artefacts. Use paint to compare them, not the M$ image viewer or fax viewer, since those filter and resize the pictures.

I had some serious interference problems again with the wintv2k44_22107.exe drivers, so I went back to the 295_22083 beta ones. These deliver a much to bright picture, so I used the defaults from the other driver, but at least they work on my athlon machine.

Brightness 5000
Contrast 2667
Hue 5000
Sat 6133
Sharpness 0 or 1

I capped at 684x576, since with this driver the capture window was smaller than for the Philips card. I capped that one at 704x576. By doing this, the pixels have the same size for both cards so you can compare better. (look at the triangles at the sides to see what I mean)
With the CX, I capped with sharpness 0 and 1
With the SAA, I capped with sharpness 0 and 2
A setting of 1 on the CX corresponds to +2 db at 4.43 MHz
A setting of 2 on the SAA corresponds to +2.2 db at 4.1 MHz

Note that the philips at default is darker; I believe the philips is correct in this, but I did not have the time to really fine-tune this for both cards. (By the way, a correct cap should look darker on a computer screen, which is why it is so difficult to tune by eye on a computerscreen if the target is DVD)
My Philips caps can be transferred to DVD without luma range adjustments. The CX caps usually come out to bright.

The Philips does look a bit blurrier, the CX sharpening cannot account for the difference completely. But the CX has it's own problems (look at the white letters in the colored fields and the aliassing in the various line patterns).

There is a clear difference. I cannot say which one is better though. You be the judge. :)

[fellaw]

Arachnotron takes the point.

I was lucky to borrow the Leadtek Expert from a colleague.
I came to similar results: The difference is hard to tell.

I sense the CX23881 sharper, even with 0 instead of the drivers default(3 in my case), but there was more noise.
It's also worse with letters, the edges are far more frayed than with the philips. I can post samples if you want.

As for the post processing, the philips captures give me lower bitrates after my usual chain(pixie + spatial + tmpgenc 12a).

As Arachnotron already stated, the 23881's drivers deliver a different active window(don't know how much this depends on the driver version). The philips give me an exact 704x576.

I made two encode short dvd-clips and played them on my standalone(Optim 8302) connected to a 32" Sony Trinitron. None of my dependants could differ them. That's the main reason why I'll stick to the Terratec.

There where a few reasons besides the picture why I stay at my current config:

1. The Leadtek imho is more uncomfortable with the cables:
It has FM in, RF in, IR in and some proprietary 8pin mini-Din for anything else. You get an adapter with composite, s-video and stereo chinch. The loop to your sound card can only be done inside your pc via cd-audio cable.

(I'll be happy on the day iuVCR supports the WDM audio of capture card as source ;) ).

2. The Terratec has gold leads, those could be the main reason for the fewer noise.

3. Never change a running system :-)

However, those 3 depend on one's liking.

[Incast]

I can't thank you enough for those they've allowed me to do the best comparison between the two cards yet.

Looking up close as the Japanese review i read did the Philips card produces a much smoother image which would aid compressibility.

The Philips card is definitely more blurred but by a smaller margin than i expected.

One strange thing i've noticed on the Philips caps is this
http://www.btinternet.com/~jconlin/grey.PNG
Now due to the high contrast of the Conexant chip i can't tell if that's normal but those greyer areas on either side don't look right to me.

I see what you mean about the Conexant white on the colour the effect is quite nasty.

Anyway thank you once again taking the time to do that, I think will buy a Philips card now.

[Arachnotron]

One strange thing i've noticed on the Philips caps is this I noticed that one too. (Also note that the way they deal with chroma differs a lot).

It may be intentional. This test sequence is full of hidden things to test out decoders. I have been hunting for the specs of this Will & Snelcox test sequence (there is no description on the DVD) but so far no luck. Also notice the coloured 1.5 0.5 1.0 vertical bars at the bottom. The Philips clearly has problems with those.

You have to the following in mind while judging this: a DVD has more than twice the chroma resolution from any analogue source you will ever encounter (like TV or (S)-VHS, so it is very difficult to judge from a testsource like this how the performance will be in real life. You may see huge differences, while for tape the caps will look the same as far as colour is concerned.

A final remark on the noise issue: Conexant cards use external anti-alias filters, which may differ between manufacturers. The SAA7134 has those on chip.

[Incast]

Could you possibly post a link to that test card you use i've had a look but cannot find it.

[Arachnotron]

Its a track from the digital video essentials DVD ( http://www.digitalvideoessentials.com ). It is named "Snell& Wilcox Zone Plate Test Pattern, field rate"

Their page is at http://www.snellwilcox.com , but I could not find any discription of the pattern there.

I know it exists, but I cannot find the link again :angry:

[leonid_makarovsky]

Originally posted by Arachnotron
I finished the caps.

You can find them here as rar files of 0.5 Mb:
http://www.arachnotron.nl/temp/CX_sharp_0.rar
http://www.arachnotron.nl/temp/CX_sharp_1.rar
http://www.arachnotron.nl/temp/PH_sharp_0.rar
http://www.arachnotron.nl/temp/ph_sharp_2.rar

Thanks for posting those. You know I didn't find that Conexant chip was sharper. It was hard to say 'cause CX had much brighter image. But it looked that Philips picture was a bit sharper.

I've heard that some think that 8 bit BT878 chip produces sharper image than Philips while Philips has a better picture. I did a test capture and noticed that Philips has a far sharper image than BT878. BT878 doesn't even come close.

Now what about Zoran based cards? Has anyone tried those?

--Leonid

[fisix]

Originally posted by Arachnotron
Yes, I did. It does look a tiny bit blurrier than CX23881 based cards, but IMHO this is caused by excessive sharpening by the conexant based cards rather than by a problem with the Philips chip. i'd have to agree. the CX chips seem to have a higher level of sharpness, but as far as i can tell all it does is accentuate any noise in the picture, including dot crawl. the phillips chips seem to produce a much more accurate color saturation/contrast, and either the noise is less (because some of the filtering doesn't have to go off chip?) or it just isn't as noticeable as on the CX-[pump the sharpness all the way] chips.

the only thing that the newer CX chips might have going for them is the 10bit ADC vs. the 9bit ADC in the phillips, but so far that spec hasn't translated into any kind of discernible (by me) quality difference, and so is a red herring.

Arachnotron, may i ask which phillips chip based card you bought?

[Arachnotron]

Arachnotron, may i ask which phillips chip based card you bought?

The Terratec Cinergy 400.

I recently heard there is a NTSC card somewhere that uses a NEC 3D combfilter in front of a SAA713x chip. For capping NTSC composite, that might be superior.

[fisix]

Originally posted by Arachnotron
The Terratec Cinergy 400.

I recently heard there is a NTSC card somewhere that uses a NEC 3D combfilter in front of a SAA713x chip. For capping NTSC composite, that might be superior.

ahh. if you find out which card that is, i'd be very thankfull for a heads up.

thanks

[Arachnotron]

See this (http://forums.virtualdub.org/index.php?act=ST&f=6&t=8471&) thread and this (http://episteme.arstechnica.com/eve/ubb.x?a=tpc&s=50009562&f=67909965&m=855008076631&r=202009227631#202009227631) thread.

It is about a card by Compro

http://www.comprousa.com/products/vmtvultra.htm