Print

Print


Super helpful, John, thanks!!

Joel 

-----Original Message-----
From: Association for Recorded Sound Discussion List [mailto:[log in to unmask]] On Behalf Of John Schroth
Sent: Wednesday, December 14, 2016 1:20 AM
To: [log in to unmask]
Subject: Re: [ARSCLIST] Somewhat OT: Digitizing videotapes

Hi Joel:

My two cents here as I do this for a living....

I would be hard pressed to send your one-of-a-kind tapes to any chain. 
I've heard of one horror story after another.

For all the reasons you mention below and more, I would only go to DVD as a second/reference media in conjunction with the master file. Ripping files from DVD is a poor substitute for going directly to a higher quality file-based format in the first place. File based format is the direction everything is headed. You want multiple copies of the files stored in different physical locations. I don't like cloud or other services, I like to be in control of my own destiny. If you have checksums made of the files while on the original capture drive, and move the checksums with the data, you insure no data corruption. Then run the checksums on a regular basis (every 6-8 months) on each copy drive to insure the drives still hold the data without any data degradation.

In analog video, there is no "metadata" that holds the original date and time, so there is no way to capture this as part of the file.

Insure the analog video is digitized properly and with the right equipment. Multiple options for playback decks allows one to see which tapes best play back in the best deck. The analog video signal should be properly set up and adjusted (using scopes and processing amplifiers) and properly processed with the right time base correction to insure the analog signal is as stable and clean as possible before you digitize. 
Picture instability, color noise, etc is "baked" into the digital version during digitization and it is very hard to correct for after you have digitized. Choosing the right codec and compression levels for your needs is key. The less compression, the more you can do with the content and the better the content will migrate to new formats in the future.

Some camps will say uprezz now. Some like myself, say to stay as close to the original format as possible - the same image ratio high by wide as possible, holding the essence of the original analog format (interlaced, not progressive and at 29.97pfs). Color sampling rates are important. You want as much color information as practical (based on the original format). Better to have the highest quality standard-def interlaced file that you can afford to store now. There will be both hardware and software solutions down the road that will do a much better job at uprezzing and de-interlacing than can be done today.

There has been recent discussion about this on the AMIA list-serve recently. Below is the answer from Jim Linder who really knows his stuff, when your same question was asked. Hopefully he does not mind that I quoted him.

Good luck,

John Schroth
MTS

"Fortunately in analog encoded (as in NTSC) there is a good answer if a bit technical. I am going to try to give a short, coherent, and understandable answer - but there is a great deal of in depth technical information on this subject available elsewhere and if others want to discuss it off list I am happy to.

We need to go back a bit and look at the development of NTSC and how the color information was added or encoded into the black and white picture. 
The standard adopted by the FCC mandated that those who had a black and white television would not have to throw out their old tv in order to get color. Color information had to be encoded into the black and white signal - the black and white televisions would basically not see or ignore the encoded information, and color sets would see what is called "burst" which is essentially a signal to turn on the color decoding circuitry and to decode the color information and apply it to the signal. This burst signal is also called subcarrier and is of a specific frequency. The color information was encoded based on the phase relationship of the subcarrier during the period of time of the line of active black and white information. Remember this is all analog and based on time / frequency and amplitude of a signal. The phase of the subcarrier signal contains the color information.

If you think of this phase relationship you realize that it takes a certain amount of time to change from one phase relationship to another
- so color changes that are "nearby" or close to the previous can be done in plenty of time, but ones that are 180 degrees out of phase actually take a bit more time. During the time that the phase is changing, you actually do not know what the color is at all. Normally and in analog this is not an issue because your eye cant see such a small part of the line, except in extreme circumstances such as when someone is wearing a herringbone jacket or very intricate patterns with different colors. When this happens we actually see noise or moire patterns because the decoding in the set can not change fast enough and so you get visual artifacts.

So, the practical limit to the horizontal resolution in digitizing an encoded SD signal (like NTSC or PAL) relates to the burst frequency and the time available. If you sample higher then the frequency of the information that it was encoded in, you are generating noise and visually the signal can actually look worse. A great deal of experimentation was done to determine what actually is optimal and certain digitizations systems that were built actually considered this so that there was no "oversampling" and thus the reduction of quality and the adding of chroma noise in the picture.

There is another factor and that has to do with the chroma decoding of the encoded signal. There are several kinds and some are better with certain signals with others. There are comb filters that remove specific frequencies in order to present a cleaner lower noise signal and adaptive comb and other filter types that are all designed to reduce noise and artifacts.

So you do not digitize in the highest resolution possible - you digitize at the optimal resolution and with the optimal chroma decoding to get the cleanest and most accurate representation of the original analog signal. Fortunately this technology has been available for some time and a great deal of video preserved has used this technology and is now happily in files."
On 12/11/2016 9:37 PM, Joel Bresler wrote:
> Dear friends:
>
>   
>
> I would like to digitize some camcorder footage. A few questions for anyone who has tackled a similar project.
>
>   
>
> First, there are plenty of services (Costco, etc.) which will digitize tapes onto DVDrs. I am wondering in this age of increasingly cheap storage if it makes more sense to go from tapes directly to digital files instead? (Avoids any quality issues going to DVD first and then ripping to digital; avoids issues with suitability of DVDrs for archival storage; renders video in editable, digital form right at the start.) Are there any suggested services that take this approach?
>
>   
>
> Second, any suggested quality level for the video? Is there any need to preserve it at a higher resolution than the original?
>
> Third, suggested ways of storing the digitized video? (RAID array? 
> Backed up on Amazon Web Services?)
>
> And last, is there any way to capture date and time metadata on the transferred video? WITHOUT showing it continually on screen as the video plays?
>
>   
>
> Thanks in advance for your insights!
>
> Regards,
>
>   
>
> Joel
>
>   
>
>   
>
> Joel Bresler, Publisher
>
> www.sephardicmusic.org
>
> 250 E. Emerson Road
>
> Lexington, MA 02420
>
> United States
>
>   
>
> 1-781-862-4104 (Telephone & FAX)
>
> [log in to unmask]
>
> IN CASE OF VERIZON EMAIL PROBLEMS, PLEASE USE MY BACK-UP EMAIL:
>
> joelbresler-at-gmail.com
>
>   
>
>   
>
>   
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 2016.0.7924 / Virus Database: 4728/13577 - Release Date: 
> 12/11/16
>
>
>
>