Dazzle DVC107 video decoder chip info wanted

[kpmedia]

Quote:

Originally Posted by AusDaz

Though they tell me it doesn't handle timebase errors well, and they recommend an external TBC! That's disappointing! Sounds like it's been configured for very stable professional equipment.

That's kind of what I figured.

Most of these so-called "TBCs" are really just filters with minimal abilities. If you feed a typical timing-damaged source into it, it fails, or it makes only minimal corrections. That's why you really have to be careful with anything that claims TBC functionality.

There's things like this in every industry. In the web hosting industry, it's the loosely-used term "cloud" to describe hosting or file storage lockers. While "TBC" and "cloud" were both terms initially created for a very specific type of hardware setup, others have taken extreme liberties in how it's used. People most often learn about these false claims and fudging of facts after they've parted with money (or tragedy has struck).

This site is more concerned for its readers and users, and will warn others of potential problems. We won't just reply with "Oh yeah, works great, here's a link to go buy it!" like you tend to find elsewhere online. Hence the occasional accusation of "negativity" when Site Staff try to warn others about potential issues with various hardware and software. We consider ourselves to be consumer advocates, to some degree.

Glad you found this out without having to buy something.

[Steve(MS)]

I am curious about these HD cards as I might buy one sometime in the future but would like to hear input.
Certainly the Colossus if it has a form of TBC would make it attractive for old video captures.
It seems the most popular are the Avermedia PCI version, the 1212 and perhaps the Intensity.
Anyone care to go into detail about the +s and -s of each card to provide intelligent info?
For myself I am more interested in their high def. abilities.
It appears the movie industry has killed any chance of us having a decent Blu-ray recorder (here in US).

[AusDaz]

I'm curious about them too.. Anyone know how well the Hauppauge Colossus does with timebase errors after you enable it's TBC option? (I'm getting burned out from constant searching)

kpmedia:
I'm more inclined to believe component manufacturers (who supply copious data) than some product manufacturers. Still - not many capture products I've looked at claimed to have any TBC function.

I guess I'm approaching this from an engineering point of view, rather than a user point of view... I've even considered building my own capture device (don't laugh) but high-speed data interfacing and writing a software driver is not trivial... Also the DIY approach often turns out being a lot more work than first thought.

[kpmedia]

Quote:

Originally Posted by AusDaz

I guess I'm approaching this from an engineering point of view, rather than a user point of view... I've even considered building my own capture device (don't laugh)

That actually sounds quite interesting.

I think the biggest problem is when manufacturers don't pay attention to the needs of users, or they somehow think users are stupid, and will buy into hogwash marketing. From my observations, Canopus paid dearly for blowing smoke up customer's butts (or attempting to do so), and have lost market share in a number of arenas as a result. In fact, the only time I really hear anything about Canopus is when the parent corp (Grass Valley) does a broadcast install at a college or university. Case in point: the so-called "TBC" function in DV converters boxes, which does absolutely nothing.

What's equally sad is when quality components are made, and not well utilized, or used only by inferior manufacturers. LSI Logic chipsets were underused by virtually everybody but JVC. All DVD recorders should have been based on LSI, as it was superior to pretty much every other chipset save Zoran (which lacked chroma NR, but was otherwise excellent).

[jmac698]

I've considered this as well. The main problem is the high data rate. It would require FPGA technology. It would take some time. I think the simplest approach is a decoder chip, fpga, and SD card.

But I've found that there's cheap ways to digitize raw, high speed signals that I could process in software, which is good enough for me. Soon I might hack apart a VCR and try to decode the raw signal.

But if you ever want to collaborate on an open-source hardware capture card, let me know
I can program an FPGA and am pretty good at the video standards.

[kpmedia]

If either of you want to chronicle projects, we'll open a special developer section here on this site, where you can do that easily.
I want to bring more advanced video users here to this site, as it's a shenanigans-free professional environment.

And we'll help you manage the pages, so you can spend your effort on the actual development of the tech.

[Steve(MS)]

Quote:

Originally Posted by jmac698

But I've found that there's cheap ways to digitize raw, high speed signals that I could process in software, which is good enough for me. Soon I might hack apart a VCR and try to decode the raw signal.

I have often wondered why this hasn't been done on a much wider scale, think of the opportunity not only to learn but for advancement in creating useful software.
Too bad too Manufacturers would likely not be willing to divulge old technologies.
Some TVs have very effective filters...I am surprised no one hasn't figured a way to intercept
outgoing streams then put them "inline".
I would imagine they might even frown on consumers learning to salvage.
Then what about upconversion....

[NJRoadfan]

Quote:

Originally Posted by kpmedia

What's equally sad is when quality components are made, and not well utilized, or used only by inferior manufacturers. LSI Logic chipsets were underused by virtually everybody but JVC. All DVD recorders should have been based on LSI, as it was superior to pretty much every other chipset save Zoran (which lacked chroma NR, but was otherwise excellent).

Zoran had its on fair share of issues related to their MJPEG codec chip and capture devices. namely, they refused to support them at all and didn't bother to assist manufacturers that wanted to write drivers for newer OSes for their products.

[AusDaz]

Honestly, I just wanted to digitize some old tapes, not start a revolution! hehe

jmac & kpmedia:
Thanks for your generous offers

Steve(MS):
Video tape recording technology is well known - The main problem with the software VCR demodulation approach is it uses a huge amount of processing power.

jmac:
Not sure if I really want to take on so much work, but it sounds interesting. I was thinking of a PC interface, first FireWire (then I read it's "complex and convoluted"), then USB2 (maybe a bit too slow, but development boards are available), now PCIe... I've never worked with FPGAs but they are very powerful devices. I figure the data rate is just over 31 megabytes per second, for RGB24 video (there might be better choices).

As for digitizing from a VCR head amp, there are actually 4 outputs.. 2 from the video heads, and 2 from the hifi audio heads. You could use the headswitching in the VCR and halve the number of signals, but then you lose the option of overlapping DSP demodulation and glitchless baseband switching. Either way it's a lot of data. I'd guess about 12 Meg sampling for the video channels and about 4 or 5 Meg each for the sound might be just enough. The frequency response of the heads is not flat, it may need EQ before you digitize.

In DSP many deluxe features become available, such as glitchless headswitching for video and sound, dropout concealment, better linearity, better filtering, and demodulation that can reject noise. Oh and yes - software TBC! Or should it be called "Picture realignment" when it's no longer analog?

[lordsmurf]

PCI Express is long in the tooth, and was really nothing more than a video gamer's expansion of AGP. It didn't do a whole lot outside of graphics cards used for playing games. I'd instead look at USB3 or straight ethernet appliances. (Thunderbolt is more proprietary Apple hardware, and that means 3 things: expensive, complicated, future unknown. Otherwise I'd have mentioned it, too.) Remember that most broadcast hardware works over ethernet now, not inside computer towers. SO much, in fact, that broadcasting subjects are now dominated by networking details.

I actually liked your assessment of failed TBCs over at VH: "it won't be able to follow the rapid timebase errors from typical VCR video. So people may think the TBC is useless when in fact it is just too slow for their purposes." ... which comes from an engineering outlook, of course.

Though to repeat what I wrote there, "I will accept this theory, but reject the conclusion. If it doesn't perform to the expected norms, I don't think it should be considered a TBC. I've long held this belief with hardware from Sima and others. Too much confusion is caused by people looking for effective TBCs. I often think it's intentional deceptive marketing."

You also make a good point on the "software TBC". Since timing does get lost on typical digital files (excluding jmac's driver hack), it very much is a picture realignment process. But I'd still give some leeway on the name, since it provides the desired TBC functionality.

I think we've had a great conversation thus far.

[AusDaz]

lordsmurf:
I've enjoyed all the discussions - they've been very informative. Great forum you guys have here.
I'm glad you've enjoyed the conversations also - it didn't seem that way at first!


Well, I've ordered a Hauppauge Colossus.

I was able to order the card locally online, but the retailer doesn't stock the s-video adaptor cables for the Colossus. They are a special input adaptor, Hauppauge model number 6021291

Hopefully an s-video to 2 x RCA patch lead (Y/C) will work with the existing adaptor cables. Otherwise I'll have to order the Hauppauge s-video adaptor from somewhere.

[jmac698]

Do let us know how it works out.

[sChen77]

Quote:

Originally Posted by AusDaz

I was able to order the card locally online, but the retailer doesn't stock the s-video adaptor cables for the Colossus. They are a special input adaptor, Hauppauge model number 6021291

Hopefully an s-video to 2 x RCA patch lead (Y/C) will work with the existing adaptor cables. Otherwise I'll have to order the Hauppauge s-video adaptor from somewhere.

Hi AusDaz,

I took a look at the picture of the special input adaptor and find it interesting that the cables leading to the adaptor look really skinny.

I mean, we can use Monster-quality RCA or S-video cables out from the tape deck or TBC, and then for the "last-mile" before the recording chip, we have to run the signal through a skinny little and probably unshielded cable?

Or does the quality of the cable not matter that much for such a short distance?

(The same thing can be said about the AV-cable adaptors for the year 2010 models of Samsung LED TV's. The skinny breakout cables just don't seem very appropriate.)


Cheers,
Stephen
Singapore

[jmac698]

I've actually tested cables. First, I've seen some tests of cables with high-end equipment, and yes there's measureable differences. In fact there's some very special bridge type procedures for measuring impedance. However, over short lengths, I've found only one slight difference.
I used a sensitive measurement program http://audio.rightmark.org/index_new.shtml
And various cables, some expensive and some cheap. The worst condition was many $1 cables put together with adapter plugs, and loops around a power bar, along with a dozen other cables behind my desk.
I tested with an XFi external usb soundcard in 24bit.
There was no measureable difference, except in the distortion test if I coiled the cable.
That's it.
No hum was picked up.

My guess is that the main use for good cables is over long runs, like on stage for audio or to a ceiling home video projector for video. I've also seen sparkle coming from a cheap HDMI cable.
There's other factors which are important though; the plating material used makes a difference, because of the effect of two metals bonding together, and because of corrosion. There's the possibility of breakage when repeatedly bent. The firmness of the plugs. Being properly stored without a twist in them, and so on.
But for me, $1 cables work perfectly fine and out of some 30, I've only got 3 bad ones over the years.

Note: your skin has acids and salts which can corrode.

[AusDaz]

Uhh.. What he said!

Such a small length of cable won't make much difference, unless there is something seriously wrong with it. The frequencies and impedances used in SD video cabling are low enough that they don't make much difference for short runs (HD uses would be more critical, but I have never used HD analog). But the cables should be real coaxial cables (75 ohms), not just shielded cable (though you can get away with using that if you're stuck).

Expensive "super" cables are mostly just a gimmick to part you from your money.

The most likely problem with such a thin cable is that it will eventually break (especially if it has a heavy cable hanging off it).

[kpmedia]

While thin cables should not be an issue according to theory, they often are in practice.
The longer the run, yes, the more important the quality. But that doesn't mean it's unimportant on short runs.

We have to toss about a dozen cables in the trash every year, and most of them are thinner ones that came with hardware. The thin cables are not just more fragile, but have a habit of adding (picking up) noise in the signal path. It doesn't matter what kind of wire it is, ethier -- HDMI, VGA, s-video, component, composite, coaxial (RG59 and RG6).

Monster is overpriced, yet convenient. Some of the Philips wires sold at Lowe's and Walmart are pretty good. The better cabling is often sold from specialty stores, online only, or in bulk only. Last time I wanted to wire a house with quality RG6, I had to buy 200 feet of it on a small spool, and then add heads myself. Best RG6 wire ever. Not so much as a hiccup in 6+ years.

Thin wimpy-shielded cables are cheap to produce, and are given away in consumer devices because it's expected. But serious users will almost always need/want to replace them with better grade cables. We want actual quality, not "good enough" (a sorry excuse for accepting lousy quality). The hard-to-accept rub is that a small box of good wires can cost just as much as a hardware component (VCR, small TV, DVD recorder, Blu-ray player, etc), and it's one of those hidden costs of running a serious video business, or having a serious video hobby.

DataVideo, Nintendo (1980s), and ATI (pre-AMD) are companies that gave out quality wires with their hardware.

[lordsmurf]

Quote:

Originally Posted by AusDaz

lordsmurf: I've enjoyed all the discussions - they've been very informative. Great forum you guys have here.
I'm glad you've enjoyed the conversations also

What makes it interesting is when you get engineers/developers together with full-time hardware/software users. Too often hardware and software seemed to be developed from a very narrow set of sample parameters, and then when released, it consistently fails to live up to promises made by the lab. When end users get direct access to the devs, magic happens. Management, marketing and bean counters need to sit in the back seat and STFU sometimes.

I'll be the first to admit that I don't completely understand everything, in terms of how the drivers work, how the hardware chips work, etc. But at the same time, I probably do more video work in a single week than most people do in a whole year. And my specialty is handling signal-damaged consumer formats, mostly analog videotapes, and especially VHS. I have a drawer full of bad digital video samples (DVDs on spindles), and a large box full of bad analog tapes (mostly VHS), that should make any forensic/restoration tool developer drool.

That's why I like interacting with "the other half" of the equation. You make it, I use it. But I'd also like to help you make it, because long-term it's going to be that much better of a tool that I can use.

Part of the problem, however, is making the engineers/devs understand what doesn't currently work. There's a big issue -- and has been for as long as I've discussed video online -- in discerning theory from reality (practical application). Too often, devs get defensive, and start to spout math, ignoring the issue. In the context of TBCs, we get a lot of promises from ultimately non-functional/non-effective hardware. And your reasoning is sound as to why that happens (not enough RAM, too slow, etc), though it doesn't change anything for the end users. To us, it's broken crap and lies, and I think that's where you observed animosity on the topic.

[jmac698]

LS,
I totally know what you mean. I think there's a personality thing that goes along with the analytical mind, you can get idealistic and theoretical, but I totally understand. I can appreciate when something doesn't practically work, but in order to fix it, I need to define what exactly is happening.
As for TBC, you're right, but it took a while for me to understand what you expect from a TBC in terms of the practical visual effects.
I'll get around to the vertical jumping sometime, that should work out better than the line jitter problem I've been on.

[lordsmurf]

Quote:

Originally Posted by jmac698

I'll get around to the vertical jumping sometime, that should work out better than the line jitter problem I've been on.

I have at least 2-3 varied samples that I want to submit to you, for any vertical-only anti-jitter filter.
Probably 2-3 more beyond that, if I get some time to think about it.
Current methods all rely on demotion/deshaker, but it's far from usable on vertical-only jitter.

This is probably best left to another thread -- or the special dev site we've talked about creating.

[jmac698]

I have one sample to try, you posted a faq about tbc. It was some cartoon.