Dazzle DVC107 video decoder chip info wanted

[AusDaz]

Hi everyone,

Does anyone know for sure which analog video decoder chip is used in the latest Pinnacle Dazzle DVC107?

I have read that some versions of the Dazzle Video Creator Platinum use the ADV7180 chip, which has a very nice feature called "Adaptive Digital Line Length Tracking" or "mini-TBC" which can help stabilize analog video.

Pinnacle haven't been any help (no reply).

[kpmedia]

There's no way that a TBC will be inside such a small and cheap device.
The Dazzle devices have not really changed much in the past 6-7 years now. Only the Studio software really has.

[AusDaz]

Maybe you should have a read of these then...

This page mentions the Dazzle Platinum using the ADV7180: (but not the DVCxxx version)
http://www.hardwarest.com/product/Pi...num-34899.html

The ADV7180 datasheet is on this page:
http://www.analog.com/en/analog-to-d...s/product.html

This is an application note regarding the "ADDLT" "mini-TBC" feature:
http://www.analog.com/static/importe...7357AN_850.pdf

I'm trying to find someone who actually knows what chip is in the latest DVC107, not conjecture about what it could or could not contain.

If you think about how TBC works, it makes sense to put the TBC function into the capture device.
Why? Several reasons:
1. It avoids the need for an additional conversion from analog to digital and back
2. The capture device has direct access to the sync pulses
3. The capture device has control over it's own timing of when it samples the video
4. The capture device feeds it's output into a RAM buffer, isolating the output timing from the input timing
5. It's the cheapest solution!

[lordsmurf]

I reviewed one of these in 2009. It did nothing at all to help clean up VHS tapes. It looked awful.

- Do you have any idea how many devices have claimed to have "TBC functionality" in the past decade?
- Would you like to guess at how many of them actually had TBC-like performance? (Hint: It's a round number.)

Pay close attention to the exact word: "TBC functionality" -- not a TBC. This is the kind of literary trickery you find on kids cereals, where it lists out the health benefits of eating sugar for breakfast. This is going to be another case of timebase correction being used as a marketing term, instead of actually referring to a real TBC.

Note the actual description:

Quote:

The ADV7180 implements a patented ADLLT™ algorithm to track varying video line lengths from sources such as a VCR. ADLLT enables the ADV7180 to track and decode poor quality video sources such as VCRs and noisy sources from tuner outputs, VCD players, and camcorders. The ADV7180 contains a chroma transient improvement (CTI) processor that sharpens the edge rate of chroma transitions, resulting in sharper vertical transitions.

Lots of companies have made claims of patented proprietary technology that acts like a TBC -- yet it's not a TBC. Why? Well because a TBC costs money to manufacture, and takes up space. Bigger and more expensive is rarely on the wish list for consumers buying consumer devices, which is what this is.

Canopus used to insist the ADVC boxes had a "TBC", but you saw no visual proof of it. Wiggle wobbly video input came out just as wiggly wobbly as it went in. Same for chroma noise splotches. If there's a TBC inside, it must be broken or asleep.

I addressed this concept of fake TBCs in the TBC FAQ: http://www.digitalfaq.com/forum/vide...time-base.html

[AusDaz]

Oh dear.. The arguments are starting and we still haven't even found out if the Dazzle DVC107 contains the ADV7180 chip yet! (which is the reason I started this thread)

I would like to read your review of the DVC107 lordsmurf, is it available online? Are you really saying that ADLLT is some kind of useless con and doesn't offer any benefit on typical VHS tapes???

Isn't it amazing how good the TBC performance of some MiniDV cams is.. which is what got me thinking about this subject in the first place. I wonder what chipsets they use?

My point was: There is nothing you can do in a stand-alone TBC that couldn't be designed into a capture device. Basically it's resampling the video to correct the timing jitter (there is even someone working on a SOFTWARE TBC). You could look at it as a type of unwanted Frequency Modulation of the video. All that's needed is an inverse type of FM to undo the "wiggly wobbly" part. With modern ICs it doesn't have to be big or expensive to do the job. Whether a particular manufacturer of a capture device can get their hardware and software working together to make it happen is a separate issue altogether.

I am still looking for the answer to my original question (in case anyone reading this knows): Is the ADV7180 video decoder chip used in the latest Dazzle (Platinum HD) DVC107? At this rate it looks like I will have to actually buy one before I can find out...

[lordsmurf]

I don't know if the Dazzle DVC107 contains this specific chipset...

... but I do know that the Dazzle unit has no TBC functionality whatsoever. "Enhancing" video was hyped on the box, but it was a dud in reality.
The full review is no longer online, but I do have an editor draft of it. Here's the important excerpt:

Quote:

No Way to “Enhance” Videos: Reading the box, I was actually hoping there would be a way to improve and restore the quality of my VHS transfers, but no such filters exist. This appears to be a marketing word used to describe the YouTube/iPod/etc. encoding feature.

The software TBC is being worked on by a member from this site (jmac). He's on this site plus at least two others. I've submitted research samples to him, so he can try to perfect the software TBC. It's a nice start, but it still has a long way to go still. And it may never work out properly, though I'd really like to be wrong here. The major problem is that the analog signal timing is lost by the time you get to software, so you're having to guess at timing based on visual content errors. And that gets tricky.

Another site online claims that the DVC107 contains the Empia chipset, which is also found in the EasyCap devices. Now that would be more believable, as the quality from both of those units matches up pretty well.

Looking back at the spec sheet one more time, the ADV7180 appears to be a television decoder, not chipset used in capture cards or recording devices. Though I still have doubts about the qualified term "TBC functionality" instead of simply stating that it has time base correction.

Yes, some DV cameras do have very impressive TBCs. Some do not. The same can be said about S-VHS and Hi8 video cameras, too. However the presence and quality of TBCs in cameras varies quite a bit. Much like the suggested VCR list maintained on this site, there are definitely lists of preferred cameras for this same reason. (Those lists area also floating around this forum, thanks to NJroadfan and some others who post here.)

I also don't think anybody is arguing. Just giving input.

[sChen77]

Quote:

Originally Posted by lordsmurf

Another site online claims that the DVC107 contains the Empia chipset, which is also found in the EasyCap devices. Now that would be more believable, as the quality from both of those units matches up pretty well.

Hi AusDaz,

I believe lordsmurf is right on this one, and that the Dazzle DVC107 uses the Empia 2861 chipset.

ref.

• VideoGlide's software page -- software only works for Empia 2861 products (which includes the EasyCap DC60+ and Dazzle DVC107
  http://www.echofx.com/videoglide.html
  
  
• Dazzle DVC 107 Question on Se7ensins Forum -- pls browse the whole thread
  http://www.se7ensins.com/forums/thre...estion.447613/

Hi lordsmurf,


I also saw this 2004 factsheet on the Empia 2860 chipset which describes a clock, sync and field ID generator for the Empia 2860 chipset. Would this sort of qualify as some kind of TBC (in the sense of synchronising video and audio out-of-sync issues), but not necessarily that of fixing "video tearing" issues like a Panasonic DVD recorder might? (ref. page 3 of 2004 factsheet, http://wenku.baidu.com/view/0ce94f1a...1b90dd8e3.html)

An updated factsheet and Windows 7 drivers can be found here. (ref. http://home.eeti.com.tw/web20/pdfs/2...TI_2008_VA.pdf + http://home.eeti.com.tw/web20/eg/IC_support.htm)


Cheers,
Stephen
Singapore

[AusDaz]

Hi Stephen,

That chip performs a different function.

The Empia 2861 chipset is a USB 2.0 interface chip used by many capture devices. It's the chip that handles the data flow from the capture device to the PC.

The chip I've been talking about (ADV7180) is a multistandard video decoder + analog to digital converter (ADC). It decodes the analog PAL/NTSC/SECAM colour signal and digitizes it. There are also other chips in a typical capture device for audio ADC and hardware MPEG encoding (and possibly others still).

Daz

[sChen77]

Hi AusDaz,

Sorry, this is beyond me then.

I had the impression that the price difference between an EasyCap and the Dazzle is mainly due to branding and software bundle, not to additional or better hardware.

Perhaps you can try corresponding with the developer of the VideoGlide software to see if he has any knowledge of the presence of the Analog Devices chip, since I would imagine the presence of the chip in the video processing workflow would affect the video that his software is able to capture (especially if it is written specifically to address the Empia 2861 chip).

You might also like to note that Avid has just announced that they will be selling their consumer business (including Dazzle) to Corel. (ref. http://www.avid.com/US/press-room/Av...nes-Operations + http://www.photographyblog.com/news/...ting_products/)


Cheers,
Stephen
Singapore

[jmac698]

I agree with AusDaz - in fact, the chips in "real TBC's" are just the usual video deccoder chips that are also found in other devices like tv's, camcorders, digital recorders, and yes, capture cards*. Maybe some of these chips work better than others in their TBC aspects. I also have no doubt that a capture card can do the "line TBC" function as defined by Mr. Smurf.

I've created a test video that allows me to independently re-align video via a mathematical equation, when I captured it, I could measure the resulting line jitter. I got +-.5 pixels on the left side and +-.25 pixels on the right side; in other words the capture card functioned as a line TBC to an accuracy of less than 1 pixel.

Mr. Smurf seems concerned with other important problems such as vertical jitter and flagging. I take his word for it that a capture card doesn't fix these problems. However it's true that I've made a software "line TBC", and also I've been able to capture the synchronization signals with an ordinary capture card, so the horizontal timing information is NOT lost. I've just turned this into a plugin if you want to test. The vertical jumping I believe I can fix in software too. The flagging, and improved quality of line TBC will be much further in the future, but I have a proven algorithm for it (just need to write the plugin).

There's also capture cards that allow full, raw capturing, including all timing information, so if you have a supported card, it would be possible to make as "real" a TBC, even a BBC PAL decoder or PalPlus decoder, as you like in software. There is a lot of research into these cards because of the software project GNURadio, which is looking for cheap hardware to digitize radio signals (and a tv capture card is perfect for that!). They found that some chips have a test mode which captures raw signals, like a Digital Storage Oscillscope.

There is also a question of semantics in this disussion, in agreeing on what a "real TBC" does, but I agree with Mr. Smurf that the capture card won't fix *all* your possible problems in capturing.

Software TBC beta
http://forum.doom9.org/showthread.php?t=162726

3MHz Digital Storage Oscilloscope or Software Defined Radio for $20!
http://sdr.osmocom.org/trac/wiki/rtl-sdr

*This is off the top of my head; I just recall reading about the chips in a TBC device and recognizing the chip, and a few other instances of trivia I've come across.

[jmac698]

ps if you do a patent search, you'll eventually find this "patented ADLLT algorithm", and you can read a full disclosure of how it works. If you find it, I can probably tell you what's innovative about it. However, most chips have something like this anyhow. They sample the line more than they have to, find the two endpoints, then resize the pixels in-between to a standard length, just like I do in software. No more than a bicubic resizer. The innovation mgiht come from missing sync pulses, or severely distorted ones, maybe they make a good guess about these.

[lordsmurf]

Quote:

Originally Posted by jmac698

also I've been able to capture the synchronization signals with an ordinary capture card

I didn't think that had yet been agreed upon. The visual sync lines are still wibbly-wobbly even after correction of the image. This has been a concern of mine since the start of the project, and both johnmeyer and sven_x mentioned it early into one of the doom9 threads. Ghitulescu also pointed out the physical tape warping effects that a hardware TBC has to overcome, and that does require more than realigning the lines -- now digitized into pixels, when a video file is the source -- though I think you have addressed that aspect.

(Ghitulescu and johnmeyer are two people that I've been in contact with multiple times in the past decade, often with discussions that headed into advanced and theoretical video topics. Ghitulescu is also a member here. Definitely not dummies.)

In terms of real-world application, I'm not yet sure how your software TBC is going to work. So far, it seems that the filter requires atypical data be captured (sync), using an outdated card chipset. But if somebody still has the tape, running it through hardware TBC would be ideal. The software is also not realtime, like hardware-based solutions.

The primary demand for a software TBC is for situations where the initial capture was lousy, and the tape is now gone or unavailable -- Youtube clips, bad DVD conversions, nth generation tapes where sync errors are now embedded. Such a software TBC would have to rely on image information, likely a retained overscan that shows image content borders.

Wihle a good software TBC could complement a hardware TBC, I'm not yet convinced software could replace hardware, nor that it does just as good and thorough of a clean-up job.

Anyway, you know I support you on this project, wherever it ends up.


Also PM me whenever you're ready for that thing we've talked about.

[AusDaz]

Thanks for your efforts Steven.

Yes, I was aware of the company changing hands.. What bad timing! (is that a TBC joke?) Maybe that's why I'm having so much trouble getting an answer from them..


Hello jmac,

I have been following your software TBC project with great interest. Well done!

I will check out your TBC plugin, but I will need to find a way to capture with syncs using my existing card (HVR-2200).

Sounds like you already have (or are developing) a good knowledge of DSP. Nice to see you measuring your results also! Great stuff!

Software Defined Radio (SDR) is a future project of mine.. I'm also wondering how hard it would be to use SDR techniques to demodulate signals from a VCR head amp like juhok mentioned to you about 2 years back.. I can just imagine finally getting some good video capture hardware working, then immediately it's made obsolete by "software capture"..

There is an example in the ADI Application Note AN850, page 5 (the link is in my 2nd post) of the ADLLT process totally fixing flag waving on a video frame. Vertical jitter is probably caused by loss of vertical sync, or loss of horizontal sync over a longer period. An intelligent capture device (or software TBC) could just ignore the corrupted sync pulses and keep capturing using it's current timing until the sync returns. This takes decision making logic to accomplish. ADI claim their process is tolerant of corrupted syncs - I am looking forward to seeing it for myself (if I can ever find a capture device with their chip in!!!).

If there is disagreement on where the Hsync is:
The Hsync pulse looks like a blacker-than-black vertical bar. Even though it denotes the beginning of a line, it's actually closer to the right side of the picture than the left, and the sync reference point is ONLY taken from the leading edge of the pulse (the left side, going negative).

You can see it in the AN850 document mentioned above.

Daz

[jmac698]

You're right on, I agree. It has one main use case, where you can't get the original tape anymore. Otherwise, buy some good equipment and do it properly. The technique with the obsolete card was useful for me for testing and education. There are in fact modern versions though, and I'll be ordering one to test with. What this would mean, ultimately, is a hardware/software combination that costs $20 but can theoretically do anything a "real" TBC can, depending of course on my software skills - and I'm not promising anything here!. So I am convinced it is *theoretically* possible.

There is one part that you are out of date upon. In my initial experiment I found this light green line that I tried to line-up with. It turns out, this wasn't HSYNC at all, but an artifact of the AGC controls. I did eventually find the real HSYNC signal, and by capturing to NTSC-J (the Japanese version, which has black level setup), the start of the video was unmistakably clearly marked, even if it was pure black.

I have some more ideas I won't get into, but there's definitely hope for a quality improvement, and a wider range of use cases in the future.

If my skills improve to the point where I can hack Linux drivers, I could probably make any modern/popular card work on the capturing end, as well. I could put the whole setup on a USB stick and you would have to boot with it, but it would work.

Oh - as for real-time, that can be done too, there is a way to capture with my TBC in real-time but I have to write up instructions for it. This is a new ability, since I just made a plugin version which is fast enough.

[jmac698]

AusDaz,

Yes I got the sync, you can see it here
http://screenshotcomparison.com/comparison/88810
It's NOT the green line, but the black bar. I don't have the left edge, but I'm assuming the right edge is fairly consistent. I forced the video to be LIGHTER than black by using setup, with NTSC-J. I increased the brightness to clearly distinguish blacker-than-black, and captured values above 235 so there's no loss of dynamic range.

Yes that Swedish company's results were amazing, I wonder if that's just from a straight digital demodulation? You'd need about 10Ms/s, the FM luma varies from 3.4 to 4.4 MHz. You could downconvert that to about 1MHz of course, but then there's color at 740KHz and audio at 1.6MHz or so. You could do much better tracking and know precisely where dropouts are.

As for flagging, I agree it's probably missing sync, guessing where it was could be done in software with motion estimation, but also I've found that jitter followed a close normal distribution so you could guess that way - however, I think the main problem with flagging is the VCR trying to compensate so you'd need a model of how it does this and undo it. That's easy to find, if you can get known material with a flagging problem then find the exact flagging against the DVD version.

Anyhow I'm sure their method is just normal engineering and putting some work into analyzing the problem, the same as I'm doing.

[AusDaz]

jmac,

I actually meant missing/damaged sync could cause vertical picture jumping. Most minor vcr picture instabilities are probably caused by unwanted tape movement or deformation/stretching (not counting dropouts, loss of signal etc).

You should find that locking to the left edge of the Hsync pulse is more stable. That's the timing reference for the hardware, and pulse widths can change due to black clipping, non-linearity and bandwidth limitations - causing the right side of the Hsync pulse to shift depending on video conditions. It might be easier to capture the left Hsync edge on the RIGHT hand side of the video, then work backwards. Getting the most accurate sync will be especially important if you go for better sub-pixel accuracy - but that's when things will REALLY get tricky...

[AusDaz]

I seem to have an answer to my original question..

A reply came in on another forum showing that the Dazzle DVC107 does NOT use the ADV7180, but instead uses the older SAA7113H. The link is here. This might explain the poor performance that lordsmurf found with unstable video timing.

A Russian forum mentions the ADV7180 is used in the older DVC130 and DVC170 models, but they have no 64 bit drivers (which I would need).

Still no answer from Pinnacle or Avid, but I have made new enquiries, and am looking for another capture device featuring the ADV7180 or similar.

[Steve(MS)]

That's really the catch ( 64 bit), some of the older cap cards are superior but no way to run them on
today's more powerful computers. I remember seeing a TBC PCI cap card but I forget who made it, it probably
wouldn't work with 64 bit.

Actually it is the Datavideo card and perhaps it will work with 64 bit but I am sure it has a steep price.
I had a memory jog so decided to edit.

[jmac698]

Oh, there's a way around that!
Virtualbox lets you install 32bit drivers and run them in 64bit. It allows access to the PCI bus.
This is pretty advanced virtualization tech.
Ironically, you need a high end machine to do it, because it requires a CPU with virtualization extensions, so you need an i7 or so.
Give it a try! www.virtualbox.org

[AusDaz]

Some new information has come in..

From another forum I am told that:
The Hauppauge Colossus uses an ADI chip that has the ADLLT feature. It is advertised as having TBC.

And I just received a reply from the Australian BlackMagic Design distributor, confirming that their Intensity Pro card uses the ADV7180. 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.