[Flygon]

Kushami, on 02 August 2014 - 07:59 PM, said:

LocalH, on 12 July 2014 - 05:31 PM, said:

I still see no reason to make lossless captures of NTSC video. At least none other than an initial intermediate, certainly no need to send huge files in at least the double-digit GB range across the internet.

The need is, frankly, that these laserdiscs are going to bit rot eventually and having a lossless digital backup, untampered with as much as possible, is archival. It is to future proof this release. Mpeg2 in SD is not archival. It's good enough for casual players, but there should be an archive of these capped straight from the LD, losslessly compressed somewhere so that when the day happens that all the laserdiscs are ruined that it still exists somewhere. It's future proofing for whatever comes next.

I'm in full agreement. Completely full agreement. Particularity since, assuming hard drive capacity and internet speeds keep extending at the same rate they are already, the sizes involved will be absolutely trivial within 25 years.

Also, H.264 > MPEG2.

This post has been edited by Flygon: 03 August 2014 - 10:51 PM

[ssj]

As a contributor to a couple other software preservation groups I can't thank everyone involved in this project enough.
I totally agree on preserving them in lossless formats, the discs should be preserved as close as possible to the real thing. Once the dumps are available, I'm sure re-encoded sets will eventually show up for those that want smaller sets and aren't really interested on proper preservation.

On another note, thanks also for taking the time to scan the artwork of the games this is also part of preserving video game culture. I'd just like to ask that the spines are also scanned, I see this has been done for the Mega Drive Perfect Video '92-'93, but doesn't seem to be the case for the Power Drift & Mega Drive LaserDisc.

[Kushami]

Flygon, on 02 August 2014 - 09:47 PM, said:

I'm in full agreement. Completely full agreement. Particularity since, assuming hard drive capacity and internet speeds keep extending at the same rate they are already, the sizes involved will be absolutely trivial within 25 years.

Also, H.264 > MPEG2. />

Heck, 25 minutes of footage at SD resolution in Huffy or Lagarith is only like 15gb at this point. 1TB portable HDs cost less than 100 bucks. That's not too bad nowadays, considering most blurays are at least 25gb.

[LocalH]

I'm just saying that the need to get binary lossless copies of the content is futile, as there really is no definitive lossless binary (take 20 different lossless captures with no other changes, and you'll get 20 different output files, contrasted with digitally ripped video or audio, where if you take the same source and losslessly compress it 20 times with the same settings, you'll most likely get files that are bit-identical, or at least nearly bit-identical). There's a lot of noise and imperfection that is inherent to an NTSC signal, that will be preserved with a lossless rip, that is not at all useful from a preservation standpoint unless you're anal about leaving luma/chroma crosstalk).

The only reason I suggested high bitrate MPEG-2 is due to the maturity of the format and the widespread support, as well as the lower processing cost inherent in the use of MPEG-2. I suggested using only I-frames due to the benefits that come from having each frame stand alone, although if there are only a limited number of frames that a given game seeks to, one could set those as keyframes and gain the bitrate advantage to P-frames (and B-frames if one so chooses to use them but I always liked sticking just with I- and P-frames).

Besides, most of my concerns are with the main distribution format. I don't necessarily see the importance in lossless outside of the direct capture format, I feel that some of the analog artifacts should be processed out in a limited fashion so as to make future compression more efficient, and then *that* file should be used as the "lossless master" or whatever you want to call it, although there are resolution limitations to NTSC that make lossless less necessary than with other formats.

[ICEknight]

LocalH, on 03 August 2014 - 11:23 PM, said:

I'm just saying that the need to get binary lossless copies of the content is futile, as there really is no definitive lossless binary (take 20 different lossless captures with no other changes, and you'll get 20 different output files

Perhaps if you take all the lossless captures and combine them, a point will be reached in which the current capture just can't be improved in any way, giving an identical result?

Each capture would have to be incredibly accurate, but...

This post has been edited by ICEknight: 04 August 2014 - 12:46 PM

[Meat Miracle]

ICEknight, on 04 August 2014 - 12:45 PM, said:

LocalH, on 03 August 2014 - 11:23 PM, said:

I'm just saying that the need to get binary lossless copies of the content is futile, as there really is no definitive lossless binary (take 20 different lossless captures with no other changes, and you'll get 20 different output files

Perhaps if you take all the lossless captures and combine them, a point will be reached in which the current capture just can't be improved in any way, giving an identical result?

Each capture would have to be incredibly accurate, but...

While it is possible to average data together like that, good luck doing it with the amount of picture data in question.

[Overlord]

Meat Miracle, on 04 August 2014 - 01:06 PM, said:

ICEknight, on 04 August 2014 - 12:45 PM, said:

LocalH, on 03 August 2014 - 11:23 PM, said:

I'm just saying that the need to get binary lossless copies of the content is futile, as there really is no definitive lossless binary (take 20 different lossless captures with no other changes, and you'll get 20 different output files

Perhaps if you take all the lossless captures and combine them, a point will be reached in which the current capture just can't be improved in any way, giving an identical result?

Each capture would have to be incredibly accurate, but...

While it is possible to average data together like that, good luck doing it with the amount of picture data in question.

The process you're on about is well known in the astrophotography field, it's called stacking. I can't say I've seen it applied to anything but starfield images, but can see the logic of where you're coming from.

[Black Squirrel]

Lossless compression is pointless. This isn't super high quality footage to begin with and every LaserDisc player treats these things slightly differently. So realistically you'd have to layer multiple videos from multiple LaserDisc players to make sure things like colour aren't out of whack... and the quality still wouldn't be as good as the original masters.

But I imagine Nemesis has thought long and hard about all of this.

[doc eggfan]

ssj, on 03 August 2014 - 08:46 PM, said:

On another note, thanks also for taking the time to scan the artwork of the games this is also part of preserving video game culture. I'd just like to ask that the spines are also scanned, I see this has been done for the Mega Drive Perfect Video '92-'93, but doesn't seem to be the case for the Power Drift & Mega Drive LaserDisc.

They're there, they're just very thin.

File:Power_Drift_and_Mega_Drive_LD_JP_Spine.jpg
File:Power_Drift_and_Mega_Drive_LD_JP_Top.jpg

[ssj]

doc eggfan, on 04 August 2014 - 03:15 PM, said:

They're there, they're just very thin.

They are indeed, totally missed them. Thanks again for taking the time to scan these with such great quality.

[doc eggfan]

Triad Stone

First Mega LD complete scan

[Miles Prower]

You guys are doing some fantastic work here.

[winterhell]

How many bits are there on a LD anyway? 6GB?

[doc eggfan]

The DTS digital audio on a laserdisc is stored at 1,235 kbit/s, so on a 60min CLV disc that would be 4.24GB per side (Is that right?)

The analag video is another matter. Considering the frequency of the laser reading the analog tracks, I think it was worked out at about 20GB per side on a CLV disc and 40GB per side on a CAV.

This post has been edited by doc eggfan: 15 August 2014 - 02:31 AM

[doc eggfan]

Time Gal

Another one done