Hate to post this but...

Re: Hate to post this but...

John,
I am probably going to get shot out of the sky by the tech gods on the forum but here goes. DV is a compressed format and upon the compression alot of data upwards of 80-85 percent is lost in the compression to DV. If you can come off the CCD's directly and bypass the compression, you then have a possible 4:4:4 12 bit uncompressed signal.

This then will eliminate alot of the issues with compositing, and possible up-rezing to HD...
making one hell of a impact on the quality of the final image.

My weak technical explaination ??? Sorry...

Grass Valley created a similar idea...here's some reading.
http://www.hollywoodindustry.com/2002/08_aug/features/cw_viper.htm
 
Re: Hate to post this but...

good luck.. it may be possible but the expense to do so and the storage of that uncompressed signal would be a headache. Still cool to see people are doing this kinda "experiment" though.
 
Re: Hate to post this but...

While it's technically interesting, it's one of those things that nobody will ever actually do.

First of all, whether he's getting a higher-quality signal or not is debatable, because he may be reading the raw data from the CCD, but he's giving up all the DSP options. No more CineGamma, no more gain, no more letterbox mode, no more CineMatrix, no more chroma phase adjustment, no more anything. Secondly, he'd have to slave the unit to some sort of DDR, some sort of component analog digitizing unit, and since his main complaint is about compression, I assume he's going to want to store it uncompressed, which would be (oh, ballpark here...) about 40 megabytes per second. No small task, even for today's computers.

Now, about compression: People say DV data is compressed at a rate of 5:1, but just like everything that people commonly accept as true, it isn't. The actual compression ratio is more like 8:1. The final DV compression cycle is 5:1 DCT, but that's after the color data has been decimated to 4:1:1 color sampling. Raw uncompressed D1 601 video is about 27 megabytes per second, and DV data is about 25 megabits per second, so you end up compressing the image to less than 1/8 its size. Or, in other words, yes, you're eliminating about 88% of the original source data.

Most of the data that's eliminated is actually irrelevant to human vision, however -- the DV scheme is optimized to take advantage of the way our eyes work, with more rods than cones (or is that backwards?) so luminance is preserved at a 1:1 ratio -- DV samples luminance as frequently as Digital Betacam. Eyes are less sensitive to color information, so DV samples color at 1/4 the rate of luminance.

In the end, though -- who cares? The end result looks GREAT. Look at the output of a DSR570WS -- it uses DV compression, yet it looks every bit the match of the best BetaSP camcorder ever produced (and BetaSP is uncompressed, of course, because it's not even digital).

There are two main complaints about the DV compression process: mosquito noise, and color artifacts/stairstepping. Both are valid complaints. Mosquito noise is mostly un-noticeable in motion video, but on titles and graphics it can crop up. It's the same kind of artifact you find in highly-compressed JPG files. As for the stairstepping of color artifacts, yeah, that one's regrettable -- would have been oh so much nicer if DV would have been 4:2:2 from the get-go. But it isn't.

So, best of luck to the tech wiz who's modifying his DVX, and it is that kind of innovation that is truly inspirational. But don't pay it another second of attention, because it will have no practical application for any of us.
 
Re: Hate to post this but...

I'm more or less in Barry's camp on this one, but...

I't would really depend on the technologies, how astute this Juan is, and of course MONEY! If all of these things could somehow lineup, well, it's not that this is impossible, more that it's improbable.

One thing IS for sure. If he is successful enough to put up images that prove themselves worthwhile enough, money will come to him -- Both for further development (partners) and maybe advanced purchases or deposits.

I remember when I first start doing video capture work back in the early 90's. MJPEG was the "REAL THING" but man did it take some good hardware to keep up with 4:1 compression. Now any "TOY" computer can do it without a hiccup. I think this is still going to be a HIGH hurtle for Juan to jump as he's talking about doing extended capture lengths at upwards to 40 MB per sec -- Now thats A LOT of reeeeely fast data to capture without dropping frames. And even if he succeeds in capturing it all, he still has to work all that raw data into a some meaningful format to allow it to be processed. This is about 5~6 times as fast hardware as I needed to use back then, so to me this seems almost impossible. I had to invest quite a few grand in fast SCSI drives to do 7 MB per sec MJPEG, but it seems now just about any of the manufacturers of the new 7,200 RPM drives *say* then can do 40 MB per sec or more now (based on their specs) -- I just don't know...

I guess Juan will keep trying until he does it, or finds he can't make it work. I would like to see it happen personally. If for no other reason then to open up eyes and ears to the possibility that DV isn't such a dead end after all. Especially if this *could* REALLY WORK WELL and didn't cost a fortune to implement.

-Rodger
 
Re: Hate to post this but...

Well in terms of recording the output from the camera...I work/play for a place that has the data recording issue solved.
5gigs per second....in testing now.
 
Re: Hate to post this but...

[quote author=Taylor_Moore link=board=Events;num=1075674227;start=0#9 date=02/19/04 at 14:51:30]Well in terms of recording the output from the camera...I work/play for a place that has the data recording issue solved.
5gigs per second....in testing now.[/quote]

One down, two to go ;D

-Rodger
 
You must log in or register to reply here.
Back
Top